Web Applications 1.0

Contexts in which this element may be used:: In a head element containing no other title elements.
Content model:: Text (for details, see prose).
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The title element represents the document's title or name. Authors should use titles that identify their documents even when they are used out of context, for example in a user's history or bookmarks, or in search results. The document's title is often different from its first header, since the first header does not have to stand alone when taken out of context.

Here are some examples of appropriate titles, contrasted with the top-level headers that might be used on those same pages.

  <title>Introduction to The Mating Rituals of Bees</title>
    ...
  <h1>Introduction</h1>
  <p>This companion guide to the highly successful
  <cite>Introduction to Medieval Bee-Keeping</cite> book is...

The next page might be a part of the same site. Note how the title describes the subject matter unambiguously, while the first header assumes the reader knowns what the context is and therefore won't wonder if the dances are Salsa or Waltz:

  <title>Dances used during bee mating rituals</title>
    ...
  <h1>The Dances</h1>

The title element must not contain any elements.

The string to use as the document's title is given by the document.title DOM attribute. User agents should use the document's title when referring to the document in their user interface.

3.7.3. The `base` element

Contexts in which this element may be used:

In a head element, after the meta element with the charset attribute, if any, but before any other elements, and only if there are no other base elements in the document.

Content model:

Empty.

Element-specific attributes:

href (HTML only)

target

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLBaseElement : HTMLElement {
           attribute DOMString href;
           attribute DOMString target;
};

The base element allows authors to specify the document's base URI for the purposes of resolving relative URIs, and the name of the default browsing context for the purposes of following hyperlinks.

There must be no more than one base element per document.

The href content attribute, if specified, must contain a URI (or IRI).

Authors must not use the href attribute on the base element in XML documents. Authors should instead use the xml:base attribute. [XMLBASE]

User agents must use the value of the href attribute on the first base element with an href attribute in the document as the document entity's base URI for the purposes of section 5.1.1 of RFC 2396 ("Establishing a Base URI": "Base URI within Document Content"). [RFC2396] Note that this base URI from RFC 2396 is referred to by the algorithm given in XML Base, which is a normative part of this specification.

If the base URI given by this attribute is a relative URI, it must be resolved relative to the higher-level base URIs (i.e. the base URI from the encapsulating entity or the URI used to retrieve the entity) to obtain an absolute base URI.

The target attribute, if specified, must contain a valid browsing context name. User agents use this name when following hyperlinks.

The href and target DOM attributes must reflect the content attributes of the same name.

3.7.4. The `link` element

Contexts in which this element may be used:

In a head element.

Content model:

Empty.

Element-specific attributes:

href (required)

rel (required)

media

hreflang

type

Also, the title attribute has special semantics on this element.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLLinkElement : HTMLElement {
           attribute boolean disabled;
           attribute DOMString href;
           attribute DOMString rel;
  readonly attribute DOMTokenList relList;
           attribute DOMString media;
           attribute DOMString hreflang;
           attribute DOMString type;
};

The LinkStyle interface must also be implemented by this element, the styling processing model defines how. [CSSOM]

The link element allows authors to indicate explicit relationships between their document and other resources.

The destination of the link is given by the href attribute, which must be present and must contain a URI (or IRI). If the href attribute is absent, then the element does not define a link.

The type of link indicated (the relationship) is given by the value of the rel attribute, which must be present, and must have a value that is an unordered set of space-separated tokens. The allowed values and their meanings are defined in a later section. If the rel attribute is absent, or if the value used is not allowed according to the definitions in this specification, then the element does not define a link.

Two categories of links can be created using the link element. Links to external resources are links to resources that are to be used to augment the current document, and hyperlink links are links to other documents. The link types section defines whether a particular link type is an external resource or a hyperlink. One element can create multiple links (of which some might be external resource links and some might be hyperlinks). User agents should process the links on a per-link basis, not a per-element basis.

The exact behaviour for links to external resources depends on the exact relationship, as defined for the relevant link type. Some of the attributes control whether or not the external resource is to be applied (as defined below). For external resources that are represented in the DOM (for example, style sheets), the DOM representation must be made available even if the resource is not applied. (However, user agents may opt to only fetch such resources when they are needed, instead of pro-actively downloading all the external resources that are not applied.)

Interactive user agents should provide users with a means to follow the hyperlinks created using the link element, somewhere within their user interface. The exact interface is not defined by this specification, but it should include the following information (obtained from the element's attributes, again as defined below), in some form or another (possibly simplified), for each hyperlink created with each link element in the document:

The relationship between this document and the resource (given by the rel attribute)
The title of the resource (given by the title attribute).
The URI of the resource (given by the href attribute).
The language of the resource (given by the hreflang attribute).
The optimum media for the resource (given by the media attribute).

User agents may also include other information, such as the type of the resource (as given by the type attribute).

The media attribute says which media the resource applies to. The value must be a valid media query. [MQ]

If the link is a hyperlink then the media attribute is purely advisory, and describes for which media the document in question was designed.

However, if the link is an external resource link, then the media attribute is prescriptive. The user agent must only apply the external resource to views while their state match the listed media.

The default, if the media attribute is omitted, is all, meaning that by default links apply to all media.

The hreflang attribute on the link element has the same semantics as the hreflang attribute on hyperlink elements.

The type attribute gives the MIME type of the linked resource. It is purely advisory. The value must be a valid MIME type, optionally with parameters. [RFC2046]

For external resource links, user agents may use the type given in this attribute to decide whether or not to consider using the resource at all. If the UA does not support the given MIME type for the given link relationship, then the UA may opt not to download and apply the resource.

User agents must not consider the type attribute authoritative — upon fetching the resource, user agents must not use metadata included in the link to the resource to determine its type.

If the attribute is omitted, then the UA must fetch the resource to determine its type and thus determine if it supports (and can apply) that external resource.

If a document contains three style sheet links labelled as follows:

<link rel="stylesheet" href="A" type="text/css">
<link rel="stylesheet" href="B" type="text/plain">
<link rel="stylesheet" href="C">

...then a compliant UA that supported only CSS style sheets would fetch the A and C files, and skip the B file (since text/plain is not the MIME type for CSS style sheets). For these two files, it would then check the actual types returned by the UA. For those that are sent as text/css, it would apply the styles, but for those labelled as text/plain, or any other type, it would not.

The title attribute gives the title of the link. With one exception, it is purely advisory. The value is text. The exception is for style sheet links, where the title attribute defines alternative style sheet sets.

The title attribute on link elements differs from the global title attribute of most other elements in that a link without a title does not inherit the title of the parent element: it merely has no title.

Some versions of HTTP defined a Link: header, to be processed like a series of link elements. When processing links, those must be taken into consideration as well. For the purposes of ordering, links defined by HTTP headers must be assumed to come before any links in the document, in the order that they were given in the HTTP entity header. Relative URIs in these headers must be resolved according to the rules given in HTTP, not relative to base URIs set by the document (e.g. using a base element or xml:base attributes). [RFC2616] [RFC2068]

The DOM attributes href, rel, media, hreflang, and type each must reflect the respective content attributes of the same name.

The DOM attribute relList must reflect the rel content attribute.

The DOM attribute disabled only applies to style sheet links. When the link element defines a style sheet link, then the disabled attribute behaves as defined for the alternative style sheets DOM. For all other link elements it always return false and does nothing on setting.

3.7.5. The `meta` element

Contexts in which this element may be used:

In a head element.

Content model:

Empty.

Element-specific attributes:

name

http-equiv

content

charset (HTML only)

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLMetaElement : HTMLElement {
           attribute DOMString content;
           attribute DOMString name;
           attribute DOMString httpEquiv;
};

The meta element represents various kinds of metadata that cannot be expressed using the title, base, link, style, and script elements.

The meta element can represent document-level metadata with the name attribute, pragma directives with the http-equiv attribute, and the file's character encoding declaration when an HTML document is serialised to string form (e.g. for transmission over the network or for disk storage) with the charset attribute.

Exactly one of the name, http-equiv, and charset attributes must be specified.

If either name or http-equiv is specified, then the content attribute must also be specified. Otherwise, it must be omitted.

The charset attribute may only be specified in HTML documents, it must not be used in XML documents. If the charset attribute is specified, the element must be the first element in the head element of the file.

The content attribute gives the value of the document metadata or pragma directive when the element is used for those purposes. The allowed values depend on the exact context, as described in subsequent sections of this specification.

If a meta element has a name attribute, it sets document metadata. Document metadata is expressed in terms of name/value pairs, the name attribute on the meta element giving the name, and the content attribute on the same element giving the value. The name specifies what aspect of metadata is being set; valid names and the meaning of their values are described in the following sections. If a meta element has no content attribute, then the value part of the metadata name/value pair is the empty string.

The DOM attributes name and content must reflect the respective content attributes of the same name. The DOM attribute httpEquiv must reflect the content attribute http-equiv.

3.7.5.1. Standard metadata names

This specification defines a few names for the name attribute of the meta element.

generator: The value must be a free-form string that identifies the software used to generate the document. This value must not be used on hand-authored pages. WYSIWYG editors have additional constraints on the value used with this metadata name.
dns: The value must be an ordered set of unique space-separated tokens, each word of which is a host name. The list allows authors to provide a list of host names that the user is expected to subsequently need. User agents may, according to user preferences and prevailing network conditions, pre-emptively resolve the given DNS names (extracting the names from the value using the rules for splitting a string on spaces), thus precaching the DNS information for those hosts and potentially reducing the time between page loads for subsequent user interactions. Higher priority should be given to host names given earlier in the list.

3.7.5.2. Other metadata names

Extensions to the predefined set of metadata names may be registered in the WHATWG Wiki MetaExtensions page.

Anyone is free to edit the WHATWG Wiki MetaExtensions page at any time to add a type. These new names must be specified with the following information:

Keyword

The actual name being defined. The name should not be confusingly similar to any other defined name (e.g. differing only in case).

Brief description

A short description of what the metadata name's meaning is, including the format the value is required to be in.

Link to more details

A link to a more detailed description of the metadata name's semantics and requirements. It could be another page on the Wiki, or a link to an external page.

Synonyms

A list of other names that have exactly the same processing requirements. Authors should not use the names defined to be synonyms, they are only intended to allow user agents to support legacy content.

Status

One of the following:

Proposal: The name has not received wide peer review and approval. Someone has proposed it and is using it.
Accepted: The name has received wide peer review and approval. It has a specification that unambiguously defines how to handle pages that use the name, including when they use it in incorrect ways.
Unendorsed: The metadata name has received wide peer review and it has been found wanting. Existing pages are using this keyword, but new pages should avoid it. The "brief description" and "link to more details" entries will give details of what authors should use instead, if anything.

If a metadata name is added with the "proposal" status and found to be redundant with existing values, it should be removed and listed as a synonym for the existing value.

Conformance checkers must use the information given on the WHATWG Wiki MetaExtensions page to establish if a value not explicitly defined in this specification is allowed or not. When an author uses a new type not defined by either this specification or the Wiki page, conformance checkers should offer to add the value to the Wiki, with the details described above, with the "proposal" status.

This specification does not define how new values will get approved. It is expected that the Wiki will have a community that addresses this.

Metadata names whose values are to be URIs must not be proposed or accepted. Links must be represented using the link element, not the meta element.

3.7.5.3. Pragma directives

When the http-equiv attribute is specified on a meta element, the element is a pragma directive.

The http-equiv attribute is an enumerated attribute. The following table lists the keywords defined for this attribute. The states given in the first cell of the the rows with keywords give the states to which those keywords map.

State	Keywords
Refresh	`refresh`
Default style	`default-style`

When a meta element is inserted into the document, if its http-equiv attribute is present and represents one of the above states, then the user agent must run the algorithm appropriate for that state, as described in the following list:

Refresh state

If another meta element in the Refresh state has already been successfully processed (i.e. when it was inserted the user agent processed it and reached the last step of this list of steps), then abort these steps.
If the meta element has no content attribute, or if that attribute's value is the empty string, then abort these steps.
Let input be the value of the element's content attribute.
Let position point at the first character of input.
Skip whitespace.
Collect a sequence of characters in the range U+0030 DIGIT ZERO to U+0039 DIGIT NINE, and parse the resulting string using the rules for parsing non-negative integers. If the sequence of characters collected is the empty string, then no number will have been parsed; abort these steps. Otherwise, let time be the parsed number.
Collect a sequence of characters in the range U+0030 DIGIT ZERO to U+0039 DIGIT NINE and U+002E FULL STOP ("."). Ignore any collected characters.
Skip whitespace.
Let url be the address of the current page.
If the character in input pointed to by position is a U+003B SEMICOLON (";"), then advance position to the next character. Otherwise, jump to the last step.
Skip whitespace.
If the character in input pointed to by position is one of U+0055 LATIN CAPITAL LETTER U or U+0075 LATIN SMALL LETTER U, then advance position to the next character. Otherwise, jump to the last step.
If the character in input pointed to by position is one of U+0052 LATIN CAPITAL LETTER R or U+0072 LATIN SMALL LETTER R, then advance position to the next character. Otherwise, jump to the last step.
If the character in input pointed to by position is one of U+004C LATIN CAPITAL LETTER L or U+006C LATIN SMALL LETTER L, then advance position to the next character. Otherwise, jump to the last step.
Skip whitespace.
If the character in input pointed to by position is a U+003D EQUALS SIGN ("="), then advance position to the next character. Otherwise, jump to the last step.
Skip whitespace.
Let url be equal to the substring of input from the character at position to the end of the string.
Strip any trailing space characters from the end of url.
Strip any U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), and U+000D CARRIAGE RETURN (CR) characters from url.
Resolve the url value to an absolute URI using the base URI of the meta element.
Set a timer so that in time seconds, if the user has not canceled the redirect, the user agent navigates to url, with replacement enabled.

For meta elements in the Refresh state, the content attribute must have a value consisting either of:

just a valid non-negative integer, or
a valid non-negative integer, followed by a U+003B SEMICOLON (;), followed by one or more space characters, followed by either a U+0055 LATIN CAPITAL LETTER U or a U+0075 LATIN SMALL LETTER U, a U+0052 LATIN CAPITAL LETTER R or a U+0072 LATIN SMALL LETTER R, a U+004C LATIN CAPITAL LETTER L or a U+006C LATIN SMALL LETTER L, a U+003D EQUALS SIGN (=), and then a valid URI (or IRI).

In the former case, the integer represents a number of seconds before the page is to be reloaded; in the latter case the integer represents a number of seconds before the page is to be replaced by the page at the given URI.

Default style state

3.7.5.4. Specifying and establishing the document's character encoding

The meta element may also be used to provide UAs with character encoding information for HTML files, by setting the charset attribute to the name of a character encoding. This is called a character encoding declaration.

The following restrictions apply to character encoding declarations:

When serialised, the charset attribute and its value must be contained completely in the first 512 bytes of the file.
The attribute value must be serialised without the use of character entity references of any kind.
The value must be a valid character encoding name. [IANACHARSET]
The character encoding name given must be the name of the character encoding used to serialise the file.
The character encoding used must be a superset of US-ASCII (specifically, ANSI_X3.4-1968) for bytes in the range 0x09 - 0x0D, 0x20, 0x21, 0x22, 0x26, 0x27, 0x2C - 0x3F, 0x41 - 0x5A, and 0x61 - 0x7A.

If the encoding is one of UTF-8, UTF-16BE, UTF-16LE, UTF-32BE, or UTF-32LE, then authors can use a BOM at the start of the file to indicate the character encoding.

In XHTML, the XML declaration should be used for inline character encoding information, if necessary.

3.7.6. The `style` element

Contexts in which this element may be used:

In a head element.

At the start of article, aside, div, and section elements.

Content model:

Depends on the value of the type attribute.

Element-specific attributes:

media

type

scoped

Also, the title attribute has special semantics on this element.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLStyleElement : HTMLElement {
           attribute boolean disabled;
           attribute DOMString media;
           attribute DOMString type;
           attribute boolean scoped;
};

The LinkStyle interface must also be implemented by this element, the styling processing model defines how. [CSSOM]

The style element allows authors to embed style information in their documents. The style element is one of several inputs to the styling processing model.

If the type attribute is given, it must contain a valid MIME type, optionally with parameters, that designates a styling language. [RFC2046] If the attribute is absent, the type defaults to text/css. [RFC2138]

When examining types to determine if they support the language, user agents must not ignore unknown MIME parameters — types with unknown parameters must be assumed to be unsupported.

The media attribute says which media the styles apply to. The value must be a valid media query. [MQ] User agents must only apply the styles to views while their state match the listed media. [DOM3VIEWS]

The default, if the media attribute is omitted, is all, meaning that by default styles apply to all media.

The scoped attribute is a boolean attribute. If the attribute is present, then the user agent must only apply the specified style information to the style element's parent element (if any), and that element's child nodes. Otherwise, the specified styles must, if applied, be applied to the entire document.

The title attribute on style elements defines alternative style sheet sets. If the style element has no title attribute, then it has no title; the title attribute of ancestors does not apply to the style element.

The title attribute on style elements, like the title attribute on link elements, differs from the global title attribute in that a style block without a title does not inherit the title of the parent element: it merely has no title.

All descendant elements must be processed, according to their semantics, before the style element itself is evaluated. For styling languages that consist of pure text, user agents must evaluate style elements by passing the concatenation of the contents of all the text nodes that are direct children of the style element (not any other nodes such as comments or elements), in tree order, to the style system. For XML-based styling languages, user agents must pass all the children nodes of the style element to the style system.

This specification does not specify a style system, but CSS is expected to be supported by most Web browsers. [CSS21]

The media, type and scoped DOM attributes must reflect the respective content attributes of the same name.

The DOM disabled attribute behaves as defined for the alternative style sheets DOM.

3.7.7. Styling

The link and style elements can provide styling information for the user agent to use when rendering the document. The DOM Styling specification specifies what styling information is to be used by the user agent and how it is to be used. [CSSOM]

The style and link elements implement the LinkStyle interface. [CSSOM]

For style elements, if the user agent does not support the specified styling language, then the sheet attribute of the element's LinkStyle interface must return null. Similarly, link elements that do not represent external resource links that contribute to the styling processing model (i.e. that do not have a stylesheet keyword in their rel attribute), and link elements whose specified resource has not yet been downloaded, or is not in a supported styling language, must have their LinkStyle interface's sheet attribute return null.

Otherwise, the LinkStyle interface's sheet attribute must return a StyleSheet object with the attributes implemented as follows: [CSSOM]

The content type (type DOM attribute): The content type must be the same as the style's specified type. For style elements, this is the same as the type content attribute's value, or text/css if that is omitted. For link elements, this is the Content-Type metadata of the specified resource.
The location (href DOM attribute): For link elements, the location must be the URI given by the element's href content attribute. For style elements, there is no location.
The intended destination media for style information (media DOM attribute): The media must be the same as the value of the element's media content attribute.
The style sheet title (title DOM attribute): The title must be the same as the value of the element's title content attribute. If the attribute is absent, then the style sheet does not have a title. The title is used for defining alternative style sheet sets.

The disabled DOM attribute on link and style elements must return false and do nothing on setting, if the sheet attribute of their LinkStyle interface is null. Otherwise, it must return the value of the StyleSheet interface's disabled attribute on getting, and forward the new value to that same attribute on setting.

3.8. Sections

Sectioning elements are elements that divide the page into, for lack of a better word, sections. This section describes HTML's sectioning elements and elements that support them.

Some elements are scoped to their nearest ancestor sectioning element. For example, address elements apply just to their section. For such elements x, the elements that apply to a sectioning element e are all the x elements whose nearest sectioning element is e.

3.8.1. The `body` element

Sectioning element.

Contexts in which this element may be used:: As the second element in an html element.
Content model:: Zero or more block-level elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: search, warning
DOM interface:: No difference from HTMLElement.

The body element represents the main content of the document.

The body element potentially has a heading. See the section on headings and sections for further details.

In conforming documents, there is only one body element. The document.body DOM attribute provides scripts with easy access to a document's body element.

Some DOM operations (for example, parts of the drag and drop model) are defined in terms of "the body element". This refers to a particular element in the DOM, as per the definition of the term, and not any arbitrary body element.

3.8.2. The `section` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more style elements, followed by zero or more block-level elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: error, example, search, warning
DOM interface:: No difference from HTMLElement.

The section element represents a generic document or application section. A section, in this context, is a thematic grouping of content, typically with a header, possibly with a footer.

Examples of sections would be chapters, the various tabbed pages in a tabbed dialog box, or the numbered sections of a thesis. A Web site's home page could be split into sections for an introduction, news items, contact information.

Each section element potentially has a heading. See the section on headings and sections for further details.

3.8.3. The `nav` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more block-level elements, or inline-level content (but not both).
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The nav element represents a section of a page that links to other pages or to parts within the page: a section with navigation links.

When used as an inline-level content container, the element represents a paragraph.

Each nav element potentially has a heading. See the section on headings and sections for further details.

3.8.4. The `article` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more style elements, followed by zero or more block-level elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: warning
DOM interface:: No difference from HTMLElement.

The article element represents a section of a page that consists of a composition that forms an independent part of a document, page, or site. This could be a forum post, a magazine or newspaper article, a Web log entry, a user-submitted comment, or any other independent item of content.

An article element is "independent" in that its contents could stand alone, for example in syndication. However, the element is still associated with its ancestors; for instance, contact information that applies to a parent body element still covers the article as well.

When article elements are nested, the inner article elements represent articles that are in principle related to the contents of the outer article. For instance, a Web log entry on a site that accepts user-submitted comments could represent the comments as article elements nested within the article element for the Web log entry.

Author information associated with an article element (q.v. the address element) does not apply to nested article elements.

Each article element potentially has a heading. See the section on headings and sections for further details.

3.8.5. The `blockquote` element

Sectioning block-level element, and structured inline-level element.

Contexts in which this element may be used:

Where block-level elements are expected.

Where structured inline-level elements are allowed.

Content model:

Zero or more block-level elements.

Element-specific attributes:

cite

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLQuoteElement : HTMLElement {
           attribute DOMString cite;
};

The HTMLQuoteElement interface is also used by the q element.

The blockquote element represents a section that is quoted from another source.

Content inside a blockquote must be quoted from another source, whose URI, if it has one, should be cited in the cite attribute.

If the cite attribute is present, it must be a URI (or IRI). User agents should allow users to follow such citation links.

If a blockquote element is preceeded or followed by a p element that contains a single cite element and is itself not preceeded or followed by another blockquote element and does not itself have a q element descendant, then, the citation given by that cite element gives the source of the quotation contained in the blockquote element.

Each blockquote element potentially has a heading. See the section on headings and sections for further details.

The cite DOM attribute reflects the element's cite content attribte.

The best way to represent a conversation is not with the cite and blockquote elements, but with the dialog element.

3.8.6. The `aside` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more style elements, followed by either zero or more block-level elements, or inline-level content (but not both).
Element-specific attributes:: None.
Predefined classes that apply to this element:: example, issue, note, search, warning
DOM interface:: No difference from HTMLElement.

The aside element represents a section of a page that consists of content that is tangentially related to the content around the aside element, and which could be considered separate from that content. Such sections are often represented as sidebars in printed typography.

When used as an inline-level content container, the element represents a paragraph.

Each aside element potentially has a heading. See the section on headings and sections for further details.

3.8.7. The `h1`, `h2`, `h3`, `h4`, `h5`, and `h6` elements

Block-level elements.

Contexts in which these elements may be used:: Where block-level elements are expected.
Content model:: Significant strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

These elements define headers for their sections.

The semantics and meaning of these elements are defined in the section on headings and sections.

These elements have a rank given by the number in their name. The h1 element is said to have the highest rank, the h6 element has the lowest rank, and two elements with the same name have equal rank.

These elements must not be empty.

3.8.8. The `header` element

Contexts in which this element may be used:: Where block-level elements are expected and there are no header ancestors.
Content model:: Zero or more block-level elements, including at least one descendant h1, h2, h3, h4, h5, or h6 element, but no sectioning element descendants, no header element descendants, and no footer element descendants.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The header element represents the header of a section. Headers may contain more than just the section's heading — for example it would be reasonable for the header to include version history information.

header elements must not contain any header elements, footer elements, or any sectioning elements (such as section) as descendants.

header elements must have at least one h1, h2, h3, h4, h5, or h6 element as a descendant.

For the purposes of document summaries, outlines, and the like, header elements are equivalent to the highest ranked h1-h6 element descendant (the first such element if there are multiple elements with that rank).

Other heading elements indicate subheadings or subtitles.

Here are some examples of valid headers. In each case, the emphasised text represents the text that would be used as the header in an application extracting header data and ignoring subheadings.

<header>
 <h1>The reality dysfunction</h1>
 <h2>Space is not the only void</h2>
</header>

<header>
 <p>Welcome to...</p>
 <h1>Voidwars!</h1>
</header>

<header>
 <h1>Scalable Vector Graphics (SVG) 1.2</h1>
 <h2>W3C Working Draft 27 October 2004</h2>
 <dl>
  <dt>This version:</dt>
  <dd><a href="http://www.w3.org/TR/2004/WD-SVG12-20041027/">http://www.w3.org/TR/2004/WD-SVG12-20041027/</a></dd>
  <dt>Previous version:</dt>
  <dd><a href="http://www.w3.org/TR/2004/WD-SVG12-20040510/">http://www.w3.org/TR/2004/WD-SVG12-20040510/</a></dd>
  <dt>Latest version of SVG 1.2:</dt>
  <dd><a href="http://www.w3.org/TR/SVG12/">http://www.w3.org/TR/SVG12/</a></dd>
  <dt>Latest SVG Recommendation:</dt>
  <dd><a href="http://www.w3.org/TR/SVG/">http://www.w3.org/TR/SVG/</a></dd>
  <dt>Editor:</dt>
  <dd>Dean Jackson, W3C, <a href="mailto:dean@w3.org">dean@w3.org</a></dd>
  <dt>Authors:</dt>
  <dd>See <a href="#authors">Author List</a></dd>
 </dl>
 <p class="copyright"><a href="http://www.w3.org/Consortium/Legal/ipr-notic ...
</header>

The section on headings and sections defines how header elements are assigned to individual sections.

The rank of a header element is the same as for an h1 element (the highest rank).

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Either zero or more block-level elements, but with no h1, h2, h3, h4, h5, h6, header, or footer elements as descendants, and with no sectioning elements as descendants; or, inline-level content (but not both).
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The footer element represents the footer for the section it applies to. A footer typically contains information about its section such as who wrote it, links to related documents, copyright data, and the like.

footer elements must not contain any footer, header, h1, h2, h3, h4, h5, or h6 elements, or any of the sectioning elements (such as section), as descendants.

When used as an inline-level content container, the element represents a paragraph.

Contact information for the section given in a footer should be marked up using the address element.

3.8.10. The `address` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The address element represents a paragraph of contact information for the section it applies to.

For example, a page at the W3C Web site related to HTML might include the following contact information:

<ADDRESS>
 <A href="../People/Raggett/">Dave Raggett</A>, 
 <A href="../People/Arnaud/">Arnaud Le Hors</A>, 
 contact persons for the <A href="Activity">W3C HTML Activity</A>
</ADDRESS>

The address element must not be used to represent arbitrary addresses (e.g. postal addresses), unless those addresses are contact information for the section. (The p element is the appropriate element for marking up such addresses.)

The address element must not contain information other than contact information.

For example, the following is non-conforming use of the address element:

<ADDRESS>Last Modified: 1999/12/24 23:37:50</ADDRESS>

Typically, the address element would be included with other information in a footer element.

To determine the contact information for a sectioning element (such as a document's body element, which would give the contact information for the page), UAs must collect all the address elements that apply to that sectioning element and its ancestor sectioning elements. The contact information is the collection of all the information given by those elements.

Contact information for one sectioning element, e.g. a aside element, does not apply to its ancestor elements, e.g. the page's body.

3.8.11. Headings and sections

The h1-h6 elements and the header element are headings.

The first heading in a sectioning element gives the header for that section. Subsequent headers of equal or higher rank start new (implied) sections, headers of lower rank start subsections that are part of the previous one.

Sectioning elements other than blockquote are always considered subsections of their nearest ancestor sectioning element, regardless of what implied sections other headings may have created. However, blockquote elements are associated with implied sections. Effectively, blockquote elements act like sections on the inside, and act opaquely on the outside.

For the following fragment:

<body>
 <h1>Foo</h1>
 <h2>Bar</h2>
 <blockquote>
  <h3>Bla</h3>
 </blockquote>
 <p>Baz</p>
 <h2>Quux</h2>
 <section>
  <h3>Thud</h3>
 </section>
 <p>Grunt</p>
</body>

...the structure would be:

Foo (heading of explicit body section)
1. Bar (heading starting implied section)
  1. Bla (heading of explicit blockquote section)
  Baz (paragraph)
2. Quux (heading starting implied section)
3. Thud (heading of explicit section section)
Grunt (paragraph)

Notice how the blockquote nests inside an implicit section while the section does not (and in fact, ends the earlier implicit section so that a later paragraph is back at the top level).

Sections may contain headers of any rank, but authors are strongly encouraged to either use only h1 elements, or to use elements of the appropriate rank for the section's nesting level.

Authors are also encouraged to explictly wrap sections in sectioning elements, instead of relying on the implicit sections generated by having multiple heading in one sectioning element.

For example, the following is correct:

<body>
 <h4>Apples</h4>
 <p>Apples are fruit.</p>
 <section>
  <h2>Taste</h2>
  <p>They taste lovely.</p>
  <h6>Sweet</h6>
  <p>Red apples are sweeter than green ones.</p>
  <h1>Color</h1>
  <p>Apples come in various colors.</p>
 </section>
</body>

However, the same document would be more clearly expressed as:

<body>
 <h1>Apples</h1>
 <p>Apples are fruit.</p>
 <section>
  <h2>Taste</h2>
  <p>They taste lovely.</p>
  <section>
   <h3>Sweet</h3>
   <p>Red apples are sweeter than green ones.</p>
  </section>
 </section>
 <section>
  <h2>Color</h2>
  <p>Apples come in various colors.</p>
 </section>
</body>

Both of the documents above are semantically identical and would produce the same outline in compliant user agents.

3.8.11.1. Creating an outline

Documents can be viewed as a tree of sections, which defines how each element in the tree is semantically related to the others, in terms of the overall section structure. This tree is related to the document tree, but there is not a one-to-one relationship between elements in the DOM and the document's sections.

The tree of sections should be used when generating document outlines, for example when generating tables of contents.

To derive the tree of sections from the document tree, a hypothetical tree is used, consisting of a view of the document tree containing only the h1-h6 and header elements, and the sectioning elements other than blockquote. Descendants of h1-h6, header, and blockquote elements must be removed from this view.

The hypothetical tree must be rooted at the root element or at a sectioning element. In particular, while the sections inside blockquotes do not contribute to the document's tree of sections, blockquotes can have outlines of their own.

UAs must take this hypothetical tree (which will become the outline) and mutate it by walking it depth first in tree order and, for each h1-h6 or header element that is not the first element of its parent sectioning element, inserting a new sectioning element, as follows:

If the element is a header element, or if it is an h1-h6 node of rank equal to or higher than the first element in the parent sectioning element (assuming that is also an h1-h6 node), or if the first element of the parent sectioning element is a sectioning element:: Insert the new sectioning element as the immediately following sibling of the parent sectioning element, and move all the elements from the current heading element up to the end of the parent sectioning element into the new sectioning element.
Otherwise:: Move the current heading element, and all subsequent siblings up to but excluding the next sectioning element, header element, or h1-h6 of equal or higher rank, whichever comes first, into the new sectioning element, then insert the new sectioning element where the current header was.

The outline is then the resulting hypothetical tree. The ranks of the headers become irrelevant at this point: each sectioning element in the hypothetical tree contains either no or one heading element child. If there is one, then it gives the section's heading, of there isn't, the section has no heading.

Sections are nested as in the hypothetical tree. If a sectioning element is a child of another, that means it is a subsection of that other section.

When creating an interactive table of contents, entries should jump the user to the relevant section element, if it was a real element in the original document, or to the heading, if the section element was one of those created during the above process.

Selecting the first section of the document therefore always takes the user to the top of the document, regardless of where the first header in the body is to be found.

The hypothetical tree (before mutations) could be generated by creating a TreeWalker with the following NodeFilter (described here as an anonymous ECMAScript function). [DOMTR] [ECMA262]

function (n) {
  // This implementation only knows about HTML elements.
  // An implementation that supports other languages might be
  // different.

  // Reject anything that isn't an element.
  if (n.nodeType != Node.ELEMENT_NODE)
    return NodeFilter.FILTER_REJECT;

  // Skip any descendants of headings.
  if ((n.parentNode && n.parentNode.namespaceURI == 'http://www.w3.org/1999/xhtml') &&
      (n.parentNode.localName == 'h1' || n.parentNode.localName == 'h2' ||
       n.parentNode.localName == 'h3' || n.parentNode.localName == 'h4' ||
       n.parentNode.localName == 'h5' || n.parentNode.localName == 'h6' ||
       n.parentNode.localName == 'header'))
    return NodeFilter.FILTER_REJECT;

  // Skip any blockquotes.
  if ((n.namespaceURI == 'http://www.w3.org/1999/xhtml') &&
      (n.localName == 'blockquote'))
    return NodeFilter.FILTER_REJECT;

  // Accept HTML elements in the list given in the prose above.
  if ((n.namespaceURI == 'http://www.w3.org/1999/xhtml') &&
      (n.localName == 'body' || /*n.localName == 'blockquote' ||*/
       n.localName == 'section' || n.localName == 'nav' ||
       n.localName == 'article' || n.localName == 'aside' ||
       n.localName == 'h1' || n.localName == 'h2' ||
       n.localName == 'h3' || n.localName == 'h4' ||
       n.localName == 'h5' || n.localName == 'h6' ||
       n.localName == 'header'))
    return NodeFilter.FILTER_ACCEPT;

  // Skip the rest.
  return NodeFilter.FILTER_SKIP;
}

3.8.11.2. Determining which heading and section applies to a particular node

Given a particular node, user agents must use the following algorithm, in the given order, to determine which heading and section the node is most closely associated with. The processing of this algorithm must stop as soon as the associated section and heading are established (even if they are established to be nothing).

If the node has an ancestor that is a header element, then the associated heading is the most distant such ancestor. The associated section is that header's associated section (i.e. repeat this algorithm for that header).
If the node has an ancestor that is an h1-h6 element, then the associated heading is the most distant such ancestor. The associated section is that heading's section (i.e. repeat this algorithm for that heading element).
If the node is an h1-h6 element or a header element, then the associated heading is the element itself. The UA must then generate the hypothetical section tree described in the previous section, rooted at the nearest section ancestor (or the root element if there is no such ancestor). If the parent of the heading in that hypothetical tree is an element in the real document tree, then that element is the associated section. Otherwise, there is no associated section element.
If the node is a sectioning element, then the associated section is itself. The UA must then generate the hypothetical section tree described in the previous section, rooted at the section itself. If the section element, in that hypothetical tree, has a child element that is an h1-h6 element or a header element, then that element is the associated heading. Otherwise, there is no associated heading element.
If the node is a footer or address element, then the associated section is the nearest ancestor sectioning element, if there is one. The node's associated heading is the same as that sectioning element's associated heading (i.e. repeat this algorithm for that sectioning element). If there is no ancestor sectioning element, the element has no associated section nor an associated heading.
Otherwise, the node is just a normal node, and the document has to be examined more closely to determine its section and heading. Create a view rooted at the nearest ancestor sectioning element (or the root element if there is none) that has just h1-h6 elements, header elements, the node itself, and sectioning elements other than blockquote elements. (Descendants of any of the nodes in this view can be ignored, as can any node later in the tree than the node in question, as the algorithm below merely walks backwards up this view.)
Let n be an iterator for this view, initialised at the node in question.
Let c be the current best candidate heading, initially null, and initially not used. It is used when top-level heading candidates are to be searched for (see below).
Repeat these steps (which effectively goes backwards through the node's previous siblings) until an answer is found:
1. If n points to a node with no previous sibling, and c is null, then return the node's parent node as the answer. If the node has no parent node, return null as the answer.
2. Otherwise, if n points to a node with no previous sibling, return c as the answer.
3. Adjust n so that it points to the previous sibling of the current position.
4. If n is pointing at an h1 or header element, then return that element as the answer.
5. If n is pointing at an h2-h6 element, and heading candidates are not being searched for, then return that element as the answer.
6. Otherwise, if n is pointing at an h2-h6 element, and either c is still null, or c is a heading of lower rank than this one, then set c to be this element, and continue going backwards through the previous siblings.
7. If n is pointing at a sectioning element, then from this point on top-level heading candidates are being searched for. (Specifically, we are looking for the nearest top-level header for the current section.) Continue going backwards through the previous siblings.
If the answer from the previous step (the loop) is null, which can only happen if the node has no preceeding headings and is not contained in a sectioning element, then there is no associated heading and no associated section.
Otherwise, if the answer from the earlier loop step is a sectioning element, then the associated section is that element and the associated heading is that sectioning element's associated heading (i.e. repeat this algorithm for that section).
Otherwise, if the answer from that same earlier step is an h1-h6 element or a header element, then the associated heading is that element and the associated section is that heading element's associated section (i.e. repeat this algorithm for that heading).

Not all nodes have an associated header or section. For example, if a section is implied, as when multiple headers are found in one sectioning element, then a node in that section has an anonymous associated section (its section is not represented by a real element), and the algorithm above does not associate that node with any particular sectioning element.

For the following fragment:

<body>
 <h1>X</h1>
 <h2>X</h2>
 <blockquote>
  <h3>X</h3>
 </blockquote>
 <p id="a">X</p>
 <h4>Text Node A</h4>
 <section>
  <h5>X</h5>
 </section>
 <p>Text Node B</p>
</body>

The associations are as follows (not all associations are shown):

Node	Associated heading	Associated section
`<body>`	`<h1>`	`<body>`
`<h1>`	`<h1>`	`<body>`
`<h2>`	`<h2>`	None.
`<blockquote>`	`<h2>`	None.
`<h3>`	`<h3>`	`<blockquote>`
`<p id="a">`	`<h2>`	None.
`Text Node A`	`<h4>`	None.
`Text Node B`	`<h1>`	`<body>`

3.9. Prose

3.9.1. The `p` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Significant inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: copyright, error, example, issue, note, search, warning
DOM interface:: No difference from HTMLElement.

The p element represents a paragraph.

p elements can contain a mixture of strictly inline-level content, such as text, images, hyperlinks, etc, and structured inline-level elements, such as lists, tables, and block quotes. p elements must not be empty.

The following examples are conforming HTML fragments:

<p>The little kitten gently seated himself on a piece of
carpet. Later in his life, this would be referred to as the time the
cat sat on the mat.</p>

<fieldset>
 <legend>Personal information</legend>
 <p>
   <label>Name: <input name="n"></label>
   <label><input name="anon" type="checkbox"> Hide from other users</label>
 </p>
 <p><label>Address: <textarea name="a"></textarea></label></p>
</fieldset>

<p>There was once an example from Femley,<br>
Whose markup was of dubious quality.<br>
The validator complained,<br>
So the author was pained,<br>
To move the error from the markup to the rhyming.</p>

The p element should not be used when a more specific element is more appropriate.

The following example is technically correct:

<section>
 <!-- ... -->
 <p>Last modified: 2001-04-23</p>
 <p>Author: fred@example.com</p>
</section>

However, it would be better marked-up as:

<section>
 <!-- ... -->
 <footer>Last modified: 2001-04-23</footer>
 <address>Author: fred@example.com</address>
</section>

Or:

<section>
 <!-- ... -->
 <footer>
  <p>Last modified: 2001-04-23</p>
  <address>Author: fred@example.com</address>
 </footer>
</section>

3.9.2. The `hr` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Empty.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The hr element represents a paragraph-level thematic break, e.g. a scene change in a story, or a transition to another topic within a section of a reference book.

3.9.3. The `br` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Empty.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The br element represents a line break.

br elements must be empty. Any content inside br elements must not be considered part of the surrounding text.

br elements must only be used for line breaks that are actually part of the content, as in poems or addresses.

The following example is correct usage of the br element:

<p>P. Sherman<br>
42 Wallaby Way<br>
Sydney</p>

br elements must not be used for separating thematic groups in a paragraph.

The following examples are non-conforming, as they abuse the br element:

<p><a ...>34 comments.</a><br>
<a ...>Add a comment.<a></p>

<p>Name: <input name="name"><br>
Address: <input name="address"></p>

Here are alternatives to the above, which are correct:

<p><a ...>34 comments.</a></p>
<p><a ...>Add a comment.<a></p>

<p>Name: <input name="name"></p>
<p>Address: <input name="address"></p>

3.9.4. The `dialog` element

Interactive, strictly inline-level content.

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more pairs of dt and dd elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The dialog element represents a conversation.

Each part of the conversation must have an explicit talker (or speaker) given by a dt element, and a discourse (or quote) given by a dd element.

This example demonstrates this using an extract from Abbot and Costello's famous sketch, Who's on first:

<dialog>
 <dt>Costello
 <dd> Look, you gotta first baseman?
 <dt> Abbott
 <dd> Certainly.
 <dt> Costello
 <dd> Who's playing first?
 <dt> Abbott
 <dd> That's right.
 <dt> Costello
 <dd> When you pay off the first baseman every month, who gets the money?
 <dt> Abbott
 <dd> Every dollar of it. 
</dialog>

Text in a dt element in a dialog element is implicitly the source of the text given in the following dd element, and the contents of the dd element are implicitly a quote from that speaker. There is thus no need to include cite, q, or blockquote elements in this markup. Indeed, a q element inside a dd element in a conversation would actually imply the person talking were themselves quoting someone else. See the cite, q, and blockquote elements for other ways to cite or quote.

3.10. Preformatted text

3.10.1. The `pre` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:: Where block-level elements are expected.; Where structured inline-level elements are allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The pre element represents a block of preformatted text, in which structure is represented by typographic conventions rather than by elements.

Some examples of cases where the pre element could be used:

Including an e-mail, with paragraphs indicated by blank lines, lists indicated by lines prefixed with a bullet, and so on.
Including fragments of computer code, with structure indicated according to the conventions of that language.
Displaying ASCII art.

If, ignoring text nodes consisting only of whitespace, the only child of a pre is a code element, then the pre element represents a block of computer code.

If, ignoring text nodes consisting only of whitespace, the only child of a pre is a samp element, then the pre element represents a block of computer output.

3.11. Lists

3.11.1. The `ol` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:

Where block-level elements are expected.

Where structured inline-level elements are allowed.

Content model:

Zero or more li elements.

Element-specific attributes:

start

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLOListElement : HTMLElement {
           attribute long start;
};

The ol element represents an ordered list of items (which are represented by li elements).

The start attribute, if present, must be a valid integer giving the ordinal value of the first list item.

If the start attribute is present, user agents must parse it as an integer, in order to determine the attribute's value. The default value, used if the attribute is missing or if the value cannot be converted to a number according to the referenced algorithm, is 1.

The items of the list are the li element child nodes of the ol element, in tree order.

The first item in the list has the ordinal value given by the ol element's start attribute, unless that li element has a value attribute with a value that can be successfully parsed, in which case it has the ordinal value given by that value attribute.

Each subsequent item in the list has the ordinal value given by its value attribute, if it has one, or, if it doesn't, the ordinal value of the previous item, plus one.

The start DOM attribute must reflect the value of the start content attribute.

3.11.2. The `ul` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:: Where block-level elements are expected.; Where structured inline-level elements are allowed.
Content model:: Zero or more li elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The ul element represents an unordered list of items (which are represented by li elements).

The items of the list are the li element child nodes of the ul element.

3.11.3. The `li` element

Contexts in which this element may be used:

Inside ol elements.

Inside ul elements.

Inside menu elements.

Content model:

When the element is a child of an ol or ul element and the grandchild of an element that is being used as an inline-level content container, or, when the element is a child of a menu element: inline-level content.

Otherwise: zero or more block-level elements, or inline-level content (but not both).

Element-specific attributes:

If the element is a child of an ol element: value

If the element is not the child of an ol element: None.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLLIElement : HTMLElement {
           attribute long value;
};

The li element represents a list item. If its parent element is an ol, ul, or menu element, then the element is an item of the parent element's list, as defined for those elements. Otherwise, the list item has no defined list-related relationship to any other li element.

When the list item is the child of an ol or ul element, the content model of the item depends on the way that parent element was used. If it was used as structured inline content (i.e. if that element's parent was used as an inline-level content container), then the li element must only contain inline-level content. Otherwise, the element may be used either for inline content or block-level elements.

When the list item is the child of a menu element, the li element must contain only inline-level content.

When the list item is not the child of an ol, ul, or menu element, e.g. because it is an orphaned node not in the document, it may contain either for inline content or block-level elements.

When used as an inline-level content container, the list item represents a single paragraph.

The value attribute, if present, must be a valid integer giving the ordinal value of the first list item.

If the value attribute is present, user agents must parse it as an integer, in order to determine the attribute's value. If the attribute's value cannot be converted to a number, it must be treated as if the attribute was absent. The attribute has no default value.

The value attribute is processed relative to the element's parent ol element (q.v.), if there is one. If there is not, the attribute has no effect.

The value DOM attribute must reflect the value of the value content attribute.

3.11.4. The `dl` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:: Where block-level elements are expected.; Where structured inline-level elements are allowed.
Content model:: Zero or more groups each consisting of one or more dt elements followed by one or mode dd elements.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The dl element introduces an unordered association list consisting of zero or more name-value groups (a description list). Each group must consist of one or more names (dt elements) followed by one or more values (dd elements).

Name-value groups may be terms and definitions, metadata topics and values, or any other groups of name-value data.

The following are all conforming HTML fragments.

In the following example, one entry ("Authors") is linked to two values ("John" and "Luke").

<dl>
 <dt> Authors
 <dd> John
 <dd> Luke
 <dt> Editor
 <dd> Frank
</dl>

In the following example, one definition is linked to two terms.

<dl>
 <dt lang="en-US"> <dfn>color</dfn> </dt>
 <dt lang="en-GB"> <dfn>colour</dfn> </dt>
 <dd> A sensation which (in humans) derives from the ability of
 the fine structure of the eye to distinguish three differently
 filtered analyses of a view. </dd>
</dl>

The following example illustrates the use of the dl element to mark up metadata of sorts. At the end of the example, one group has two metadata labels ("Authors" and "Editors") and two values ("Robert Rothman" and "Daniel Jackson").

<dl>
 <dt> Last modified time </dt>
 <dd> 2004-12-23T23:33Z </dd>
 <dt> Recommended update interval </dt>
 <dd> 60s </dd>
 <dt> Authors </dt>
 <dt> Editors </dt>
 <dd> Robert Rothman </dd>
 <dd> Daniel Jackson </dd>
</dl>

If a dl element is empty, it contains no groups.

If a dl element contains non-whitespace text nodes, or elements other than dt and dd, then those elements or text nodes do not form part of any groups in that dl, and the document is non-conforming.

If a dl element contains only dt elements, then it consists of one group with names but no values, and the document is non-conforming.

If a dl element contains only dd elements, then it consists of one group with values but no names, and the document is non-conforming.

The dl element is inappropriate for marking up dialogue, since dialogue is ordered (each speaker/line pair comes after the next). For an example of how to mark up dialogue, see the dialog element.

3.11.5. The `dt` element

Contexts in which this element may be used:: Before dd or dt elements inside dl elements.; Before a dd element inside a dialog element.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The dt element represents the term, or name, part of a term-description group in a description list (dl element), and the talker, or speaker, part of a talker-discourse pair in a conversation (dialog element).

The dt element itself, when used in a dl element, does not indicate that its contents are a term being defined, but this can be indicated using the dfn element.

3.11.6. The `dd` element

Contexts in which this element may be used:: After dt or dd elements inside dl elements.; After a dt element inside a dialog element.
Content model:: When the element is a child of a dl element and the grandchild of an element that is being used as an inline-level content container: inline-level content.; Otherwise: zero or more block-level elements, or inline-level content (but not both).
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The dd element represents the description, definition, or value, part of a term-description group in a description list (dl element), and the discourse, or quote, part in a conversation (dialog element).

The content model of a dd element depends on the way its parent element is being used. If the parent element is a dl element that is being used as structured inline content (i.e. if the dl element's parent element is being used as an inline-level content container), then the dd element must only contain inline-level content.

Otherwise, the element may be used either for inline content or block-level elements.

3.12. Phrase elements

3.12.1. The `a` element

Contexts in which this element may be used:

Where strictly inline-level content is allowed, if there are no ancestor interactive elements.

Content model:

When used in an element whose content model is only strictly inline-level content: only significant strictly inline-level content, but there must be no interactive descendants.

Otherwise: any significant inline-level content, but there must be no interactive descendants.

Element-specific attributes:

href

target

ping

rel

media

hreflang

type

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLAnchorElement : HTMLElement {
           attribute DOMString href;
           attribute DOMString target;
           attribute DOMString ping;
           attribute DOMString rel;
  readonly attribute DOMTokenList relList;
           attribute DOMString media;
           attribute DOMString hreflang;
           attribute DOMString type;
};

The Command interface must also be implemented by this element.

If the a element has an href attribute, then it represents a hyperlink.

If the a element has no href attribute, then the element is a placeholder for where a link might otherwise have been placed, if it had been relevant.

The target, ping, rel, media, hreflang, and type attributes must be omitted if the href attribute is not present.

If a site uses a consistent navigation toolbar on every page, then the link that would normally link to the page itself could be marked up using an a element:

<nav>
 <ul>
  <li> <a href="/">Home</a> </li>
  <li> <a href="/news">News</a> </li>
  <li> <a>Examples</a> </li>
  <li> <a href="/legal">Legal</a> </li>
 </ul>
</nav>

Interactive user agents should allow users to follow hyperlinks created using the a element. The href, target and ping attributes decide how the link is followed. The rel, media, hreflang, and type attributes may be used to indicate to the user the likely nature of the target resource before the user follows the link.

The activation behavior of a elements that represent hyperlinks is to run the following steps:

If the DOMActivate event in question is not trusted (i.e. a click() method call was the reason for the event being dispatched), and the a element's target attribute is ... then raise an INVALID_ACCESS_ERR exception and abort these steps.
If the target of the DOMActivate event is an img element with an ismap attribute specified, then server-side image map processing must be performed, as follows:
1. If the DOMActivate event was dispatched as the result of a real pointing-device-triggered click event on the img element, then let x be the distance in CSS pixels from the left edge of the image to the location of the click, and let y be the distance in CSS pixels from the top edge of the image to the location of the click. Otherwise, let x and y be zero.
2. Let the hyperlink suffix be a U+003F QUESTION MARK character, the value of x expressed as a base-ten integer using ASCII digits (U+0030 DIGIT ZERO to U+0039 DIGIT NINE), a U+002C COMMA character, and the value of y expressed as a base-ten integer using ASCII digits.
Finally, the user agent must follow the hyperlink defined by the a element. If the steps above defined a hyperlink suffix, then take that into account when following the hyperlink.

One way that a user agent can enable users to follow hyperlinks is by allowing a elements to be clicked, or focussed and activated by the keyboard. This will cause the aforementioned activation behavior to be invoked.

The a element must not be empty.

The DOM attributes href, ping, target, rel, media, hreflang, and type, must each reflect the respective content attributes of the same name.

The DOM attribute relList must reflect the rel content attribute.

3.12.2. The `q` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: cite
Predefined classes that apply to this element:: None.
DOM interface:: The q element uses the HTMLQuoteElement interface.

The q element represents a part of a paragraph quoted from another source.

Content inside a q element must be quoted from another source, whose URI, if it has one, should be cited in the cite attribute.

If the cite attribute is present, it must be a URI (or IRI). User agents should allow users to follow such citation links.

If a q element is contained (directly or indirectly) in a paragraph that contains a single cite element and has no other q element descendants, then, the citation given by that cite element gives the source of the quotation contained in the q element.

3.12.3. The `cite` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The cite element represents a citation: the source, or reference, for a quote or statement made in the document.

A citation is not a quote (for which the q element is appropriate).

This is incorrect usage:

<p><cite>This is wrong!</cite>, said Ian.</p>

This is the correct way to do it:

<p><q>This is correct!</q>, said <cite>Ian</cite>.</p>

This is also wrong, because the title and the name are not references or citations:

<p>My favourite book is <cite>The Reality Dysfunction</cite>
by <cite>Peter F. Hamilton</cite>.</p>

This is correct, because even though the source is not quoted, it is cited:

<p>According to <cite>the Wikipedia article on
HTML</cite>, HTML is defined in formal specifications that were
developed and published throughout the 1990s.</p>

The cite element can apply to blockquote and q elements in certain cases described in the definitions of those elements.

3.12.4. The `em` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The em element represents stress emphasis of its contents.

The level of emphasis that a particlar piece of content has is given by its number of ancestor em elements.

The placement of emphasis changes the meaning of the sentence. The element thus forms an integral part of the content. The precise way in which emphasis is used in this way depends on the language.

These examples show how changing the emphasis changes the meaning. First, a general statement of fact, with no emphasis:

<p>Cats are cute animals.</p>

By emphasising the first word, the statement implies that the kind of animal under discussion is in question (maybe someone is asserting that dogs are cute):

<p><em>Cats</em> are cute animals.</p>

Moving the emphasis to the verb, one highlights that the truth of the entire sentence is in question (maybe someone is saying cats are not cute):

<p>Cats <em>are</em> cute animals.</p>

By moving it to the adjective, the exact nature of the cats is reasserted (maybe someone suggested cats were mean animals):

<p>Cats are <em>cute</em> animals.</p>

Similarly, if someone asserted that cats were vegetables, someone correcting this might emphasise the last word:

<p>Cats are cute <em>animals</em>.</p>

By emphasising the entire sentence, it becomes clear that the speaker is fighting hard to get the point across. This kind of emphasis also typically affects the punctuation, hence the exclamation mark here.

<p><em>Cats are cute animals!</em></p>

Anger mixed with emphasising the cuteness could lead to markup such as:

<p><em>Cats are <em>cute</em> animals!</em></p>

3.12.5. The `strong` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: error, warning
DOM interface:: No difference from HTMLElement.

The strong element represents strong importance for its contents.

The relative level of importance of a piece of content is given by its number of ancestor strong elements; each strong element increases the importance of its contents.

Changing the importance of a piece of text with the strong element does not change the meaning of the sentence.

Here is an example of a warning notice in a game, with the various parts marked up according to how important they are:

<p><strong>Warning.</strong> This dungeon is dangerous.
<strong>Avoid the ducks.</strong> Take any gold you find.
<strong><strong>Do not take any of the diamonds</strong>,
they are explosive and <strong>will destroy anything within
ten meters.</strong></strong> You have been warned.</p>

3.12.6. The `small` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The small element represents small print (part of a document often describing legal restrictions, such as copyrights or other disadvantages), or other side comments.

The small element does not "de-emphasise" or lower the importance of text emphasised by the em element or marked as important with the strong element.

In this example the footer contains contact information and a copyright.

<footer>
 <address>
  For more details, contact
  <a href="mailto:js@example.com">John Smith</a>.
 </address>
 <p><small>© copyright 2038 Example Corp.</small></p>
</footer>

In this second example, the small element is used for a side comment.

<p>Example Corp today announced record profits for the
second quarter <small>(Full Disclosure: Foo News is a subsidiary of
Example Corp)</small>, leading to speculation about a third quarter
merger with Demo Group.</p>

In this last example, the small element is marked as being important small print.

<p><strong><small>Continued use of this service will result in a kiss.</small></strong></p>

3.12.7. The `m` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The m element represents a run of text marked or highlighted.

In the following snippet, a paragraph of text refers to a specific part of a code fragment.

<p>The highlighted part below is where the error lies:</p>
<pre><code>var i: Integer;
begin
   i := <m>1.1</m>;
end.</code></pre>

Another example of the m element is highlighting parts of a document that are matching some search string. If someone looked at a document, and the server knew that the user was searching for the word "kitten", then the server might return the document with one paragraph modified as follows:

<p>I also have some <m>kitten</m>s who are visiting me
these days. They're really cute. I think they like my garden!</p>

3.12.8. The `dfn` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed, if there are no ancestor dfn elements.
Content model:: Strictly inline-level content, but there must be no descendant dfn elements.
Element-specific attributes:: None, but the title attribute has special semantics on this element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The dfn element represents the defining instance of a term. The paragraph, description list group, or section that contains the dfn element contains the definition for the term given by the contents of the dfn element.

dfn elements must not be nested.

Defining term: If the dfn element has a title attribute, then the exact value of that attribute is the term being defined. Otherwise, if it contains exactly one element child node and no child text nodes, and that child element is an abbr element with a title attribute, then the exact value of that attribute is the term being defined. Otherwise, it is the exact textContent of the dfn element that gives the term being defined.

If the title attribute of the dfn element is present, then it must only contain the term being defined.

There must only be one dfn element per document for each term defined (i.e. there must not be any duplicate terms).

The title attribute of ancestor elements does not affect dfn elements.

The dfn element enables automatic cross-references. Specifically, any span, abbr, code, var, samp, or i element that has a non-empty title attribute whose value exactly equals the term of a dfn element in the same document, or which has no title attribute but whose textContent exactly equals the term of a dfn element in the document, and that has no interactive elements or dfn elements either as ancestors or descendants, and has no other elements as ancestors that are themselves matching these conditions, should be presented in such a way that the user can jump from the element to the first dfn element giving the defining instance of that term.

In the following fragment, the term "GDO" is first defined in the first paragraph, then used in the second. A compliant UA could provide a link from the abbr element in the second paragraph to the dfn element in the first.

<p>The <dfn><abbr title="Garage Door Opener">GDO</abbr></dfn>
is a device that allows off-world teams to open the iris.</p>
<!-- ... later in the document: -->
<p>Teal'c activated his <abbr title="Garage Door Opener">GDO</abbr>
and so Hammond ordered the iris to be opened.</p>

3.12.9. The `abbr` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The abbr element represents an abbreviation or acronym. The title attribute should be used to provide an expansion of the abbreviation. If present, the attribute must only contain an expansion of the abbreviation.

The paragraph below contains an abbreviation marked up with the abbr element.

<p>The <abbr title="Web Hypertext Application Technology
Working Group">WHATWG</abbr> is a loose unofficial collaboration of
Web browser manufacturers and interested parties who wish to develop
new technologies designed to allow authors to write and deploy
Applications over the World Wide Web.</p>

The title attribute may be omitted if there is a dfn element in the document whose defining term is the abbreviation (the textContent of the abbr element).

In the example below, the word "Zat" is used as an abbreviation in the second paragraph. The abbreviation is defined in the first, so the explanatory title attribute has been omitted. Because of the way dfn elements are defined, the second abbr element in this example would be connected (in some UA-specific way) to the first.

<p>The <dfn><abbr>Zat</abbr></dfn>, short for Zat'ni'catel, is a weapon.</p>
<p>Jack used a <abbr>Zat</abbr> to make the boxes of evidence disappear.</p>

3.12.10. The `time` element

Contexts in which this element may be used:

Where strictly inline-level content is allowed.

Content model:

Element-specific attributes:

datetime

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTimeElement : HTMLElement {
           attribute DOMString datetime;
  readonly attribute DOMTimeStamp date;
  readonly attribute DOMTimeStamp time;
  readonly attribute DOMTimeStamp timezone;
};

The time element represents a date and/or a time.

The datetime attribute, if present, must contain a date or time string that identifies the date or time being specified.

If the datetime attribute is not present, then the date or time must be specified in the content of the element, such that parsing the element's textContent according to the rules for parsing date or time strings in content successfully extracts a date or time.

The datetime DOM attribute must reflect the datetime content attribute.

User agents, to obtain the date, time, and timezone represented by a time element, must follow the following steps:

If the datetime attribute is present, then parse it according to the rules for parsing date or time strings in content, and let the result be result.
Otherwise, parse the element's textContent according to the rules for parsing date or time strings in content, and let the result be result.
If result is empty (because the parsing failed), then the date is unknown, the time is unknown, and the timezone is unknown.
Otherwise: if result contains a date, then that is the date; if result contains a time, then that is the time; and if result contains a timezone, then the timezone is the element's timezone. (A timezone can only be present if both a date and a time are also present.)

The date DOM attribute must return null if the date is unknown, and otherwise must return the time corresponding to midnight UTC (i.e. the first second) of the given date.

The time DOM attribute must return null if the time is unknown, and otherwise must return the time corresponding to the given time of 1970-01-01, with the timezone UTC.

The timezone DOM attribute must return null if the timezone is unknown, and otherwise must return the time corresponding to 1970-01-01 00:00 UTC in the given timezone, with the timezone set to UTC (i.e. the time corresponding to 1970-01-01 at 00:00 UTC plus the offset corresponding to the timezone).

In the following snippet:

<p>Our first date was <time datetime="2006-09-23">a saturday</time>.</p>

...the time element's date attribute would have the value 1,158,969,600,000ms, and the time and timezone attributes would return null.

In the following snippet:

<p>We stopped talking at <time datetime="2006-09-24 05:00 -7">5am the next morning</time>.</p>

...the time element's date attribute would have the value 1,159,056,000,000ms, the time attribute would have the value 18,000,000ms, and the timezone attribute would return -25,200,000ms. To obtain the actual time, the three attributes can be added together, obtaining 1,159,048,800,000, which is the specified date and time in UTC.

Finally, in the following snippet:

<p>Many people get up at <time>08:00</time>.</p>

...the time element's date attribute would have the value null, the time attribute would have the value 28,800,000ms, and the timezone attribute would return null.

These APIs may be suboptimal. Comments on making them more useful to JS authors are welcome. The primary use cases for these elements are for marking up publication dates e.g. in blog entries, and for marking event dates in hCalendar markup. Thus the DOM APIs are likely to be used as ways to generate interactive calendar widgets or some such.

3.12.11. The `meter` element

Contexts in which this element may be used:

Where strictly inline-level content is allowed.

Content model:

Element-specific attributes:

value

min

low

high

max

optimum

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLMeterElement : HTMLElement {
           attribute long value;
           attribute long min;
           attribute long max;
           attribute long low;
           attribute long high;
           attribute long optimum;
};

The meter element represents a scalar measurement within a known range, or a fractional value; for example disk usage, the relevance of a query result, or the fraction of a voting population to have selected a particular candidate.

This is also known as a gauge.

The meter element should not be used to indicate progress (as in a progress bar). For that role, HTML provides a separate progress element.

There are six attributes that determine the semantics of the gauge represented by the element.

The min attribute specifies the lower bound of the range, and the max attribute specifies the upper bound. The value attribute specifies the value to have the gauge indicate as the "measured" value.

The other three attributes can be used to segment the gauge's range into "low", "medium", and "high" parts, and to indicate which part of the gauge is the "optimum" part. The low attribute specifies the range that is considered to be the "low" part, and the high attribute specifies the range that is considered to be the "high" part. The optimum attribute gives the position that is "optimum"; if that is higher than the "high" value then this indicates that the higher the value, the better; if it's lower than the "low" mark then it indicates that lower values are better, and naturally if it is in between then it indicates that neither high nor low values are good.

Authoring requirements: The recommended way of giving the value is to include it as contents of the element, either as two numbers (the higher number represents the maximum, the other number the current value), or as a percentage or similar (using one of the characters such as "%"), or as a fraction.

The value, min, low, high, max, and optimum attributes are all optional. When present, they must have values that are valid floating point numbers.

The following examples all represent a measurement of three quarters (of the maximum of whatever is being measured):

<meter>75%</meter>
<meter>750‰</meter>
<meter>3/4</meter>
<meter>6 blocks used (out of 8 total)</meter>
<meter>max: 100; current: 75</meter>
<meter><object data="graph75.png">0.75</object></meter>
<meter min="0" max="100" value="75"></meter>

User agent requirements: User agents must parse the min, max, value, low, high, and optimum attributes using the rules for parsing floating point number values.

If the value attribute has been omitted, the user agent must also process the textContent of the element according to the steps for finding one or two numbers of a ratio in a string. These steps will return nothing, one number, one number with a denominator punctuation character, or two numbers.

User agents must then use all these numbers to obtain values for six points on the gauge, as follows. (The order in which these are evaluated is important, as some of the values refer to earlier ones.)

The minimum value

If the min attribute is specified and a value could be parsed out of it, then the minimum value is that value. Otherwise, the minimum value is zero.

The maximum value

If the max attribute is specified and a value could be parsed out of it, the maximum value is that value.

Otherwise, if the max attribute is specified but no value could be parsed out of it, or if it was not specified, but either or both of the min or value attributes were specified, then the maximum value is 1.

Otherwise, none of the max, min, and value attributes were specified. If the result of processing the textContent of the element was either nothing or just one number with no denominator punctuation character, then the maximum value is 1; if the result was one number but it had an associated denominator punctuation character, then the maximum value is the value associated with that denominator punctuation character; and finally, if there were two numbers parsed out of the textContent, then the maximum is the higher of those two numbers.

If the above machinations result in a maximum value less than the minimum value, then the maximum value is actually the same as the minimum value.

The actual value

If the value attribute is specified and a value could be parsed out of it, then that value is the actual value.

If the value attribute is not specified but the max attribute is specified and the result of processing the textContent of the element was one number with no associated denominator punctuation character, then that number is the actual value.

If neither of the value and max attributes are specified, then, if the result of processing the textContent of the element was one number (with or without an associated denominator punctuation character), then that is the actual value, and if the result of processing the textContent of the element was two numbers, then the actual value is the lower of the two numbers found.

Otherwise, if none of the above apply, the actual value is zero.

If the above procedure results in an actual value less than the minimum value, then the actual value is actually the same as the minimum value.

If, on the other hand, the result is an actual value greater than the maximum value, then the actual value is the maximum value.

The low boundary

If the low attribute is specified and a value could be parsed out of it, then the low boundary is that value. Otherwise, the low boundary is the same as the minimum value.

If the above results in a low boundary that is less than the minimum value, the low boundary is the minimum value.

The high boundary

If the high attribute is specified and a value could be parsed out of it, then the high boundary is that value. Otherwise, the high boundary is the same as the maximum value.

If the above results in a high boundary that is higher than the maximum value, the high boundary is the maximum value.

The optimum point

If the optimum attribute is specified and a value could be parsed out of it, then the optimum point is that value. Otherwise, the optimum point is the midpoint between the minimum value and the maximum value.

If the optimum point is then less than the minimum value, then the optimum point is actually the same as the minimum value. Similarly, if the optimum point is greater than the maximum value, then it is actually the maximum value instead.

All of which should result in the following inequalities all being true:

minimum value ≤ actual value ≤ maximum value
minimum value ≤ low boundary ≤ high boundary ≤ maximum value
minimum value ≤ optimum point ≤ maximum value

UA requirements for regions of the gauge: If the optimum point is equal to the low boundary or the high boundary, or anywhere in between them, then the region between the low and high boundaries of the gauge must be treated as the optimum region, and the low and high parts, if any, must be treated as suboptimal. Otherwise, if the optimum point is less than the low boundary, then the region between the minimum value and the low boundary must be treated as the optimum region, the region between the low boundary and the high boundary must be treated as a suboptimal region, and the region between the high boundary and the maximum value must be treated as an even less good region. Finally, if the optimum point is higher than the high boundary, then the situation is reversed; the region between the high boundary and the maximum value must be treated as the optimum region, the region between the high boundary and the low boundary must be treated as a suboptimal region, and the remaining region between the low boundary and the minimum value must be treated as an even less good region.

UA requirements for showing the gauge: When representing a meter element to the user, the UA should indicate the relative position of the actual value to the minimum and maximum values, and the relationship between the actual value and the three regions of the gauge.

The following markup:

<h3>Suggested groups</h3>
<menu type="toolbar">
 <a href="?cmd=hsg" onclick="hideSuggestedGroups()">Hide suggested groups</a>
</menu>
<ul>
 <li>
  <p><a href="/group/comp.infosystems.www.authoring.stylesheets/view">comp.infosystems.www.authoring.stylesheets</a> -
     <a href="/group/comp.infosystems.www.authoring.stylesheets/subscribe">join</a></p>
  <p>Group description: <strong>Layout/presentation on the WWW.</strong></p>
  <p><meter value="0.5">Moderate activity,</meter> Usenet, 618 subscribers</p>
 </li>
 <li>
  <p><a href="/group/netscape.public.mozilla.xpinstall/view">netscape.public.mozilla.xpinstall</a> -
     <a href="/group/netscape.public.mozilla.xpinstall/subscribe">join</a></p>
  <p>Group description: <strong>Mozilla XPInstall discussion.</strong></p>
  <p><meter value="0.25">Low activity,</meter> Usenet, 22 subscribers</p>
 </li>
 <li>
  <p><a href="/group/mozilla.dev.general/view">mozilla.dev.general</a> -
     <a href="/group/mozilla.dev.general/subscribe">join</a></p>
  <p><meter value="0.25">Low activity,</meter> Usenet, 66 subscribers</p>
 </li>
</ul>

Might be rendered as follows:

With the <meter> elements rendered as inline green bars of
varying lengths.

The min, max, value, low, high, and optimum DOM attributes must reflect the elements' content attributes of the same name. When the relevant content attributes are absent, the DOM attributes must return zero. The value parsed from the textContent never affects the DOM values.

Would be cool to have the value DOM attribute update the textContent in-line...

3.12.12. The `progress` element

Contexts in which this element may be used:

Where strictly inline-level content is allowed.

Content model:

Element-specific attributes:

value

max

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLProgressElement : HTMLElement {
           attribute float value;
           attribute float max;
  readonly attribute float position;
};

The progress element represents the completion progress of a task. The progress is either indeterminate, indicating that progress is being made but that it is not clear how much more work remains to be done before the task is complete (e.g. because the task is waiting for a remote host to respond), or the progress is a number in the range zero to a maximum, giving the fraction of work that has so far been completed.

There are two attributes that determine the current task completion represented by the element.

The value attribute specifies how much of the task has been completed, and the max attribute specifies how much work the task requires in total. The units are arbitrary and not specified.

Instead of using the attributes, authors are recommended to simply include the current value and the maximum value inline as text inside the element.

Here is a snippet of a Web application that shows the progress of some automated task:

<section>
 <h2>Task Progress</h2>
 <p><label>Progress: <progress><span id="p">0</span>%</progress></p>
 <script>
  var progressBar = document.getElementById('p');
  function updateProgress(newValue) {
    progressBar.textContent = newValue;
  }
 </script>
</section>

(The updateProgress() method in this example would be called by some other code on the page to update the actual progress bar as the task progressed.)

Author requirements: The max and value attributes, when present, must have values that are valid floating point numbers. The max attribute, if present, must have a value greater than zero. The value attribute, if present, must have a value equal to or greater than zero, and less than or equal to the value of the max attribute, if present.

User agent requirements: User agents must parse the max and value attributes' values according to the rules for parsing floating point number values.

If the value attribute is omitted, then user agents must also parse the textContent of the progress element in question using the steps for finding one or two numbers of a ratio in a string. These steps will return nothing, one number, one number with a denominator punctuation character, or two numbers.

Using the results of this processing, user agents must determine whether the progress bar is an indeterminate progress bar, or whether it is a determinate progress bar, and in the latter case, what its current and maximum values are, all as follows:

If the max attribute is omitted, and the value is omitted, and the results of parsing the textContent was nothing, then the progress bar is an indeterminate progress bar. Abort these steps.
Otherwise, it is a determinate progress bar.
If the max attribute is included, then, if a value could be parsed out of it, then the maximum value is that value.
Otherwise, if the max attribute is absent but the value attribute is present, or, if the max attribute is present but no value could be parsed from it, then the maximum is 1.
Otherwise, if neither attribute is included, then, if the textContent contained one number with an associated denominator punctuation character, then the maximum value is the value associated with that denominator punctuation character; otherwise, if the textContent contained two numbers, the maximum value is the higher of the two values; otherwise, the maximum value is 1.
If the value attribute is present on the element and a value could be parsed out of it, that value is the current value of the progress bar. Otherwise, if the attribute is present but no value could be parsed from it, the current value is zero.
Otherwise if the value attribute is absent and the max attribute is present, then, if the textContent was parsed and found to contain just one number, with no associated denominator punctuation character, then the current value is that number. Otherwise, if the value attribute is absent and the max attribute is present then the current value is zero.
Otherwise, if neither attribute is present, then the current value is the lower of the one or two numbers that were found in the textContent of the element.
If the maximum value is less than or equal to zero, then it is reset to 1.
If the current value is less than zero, then it is reset to zero.
Finally, if the current value is greater than the maximum value, then the current value is reset to the maximum value.

UA requirements for showing the progress bar: When representing a progress element to the user, the UA should indicate whether it is a determinate or indeterminate progress bar, and in the former case, should indicate the relative position of the current value relative to the maximum value.

The max and value DOM attributes must reflect the elements' content attributes of the same name. When the relevant content attributes are absent, the DOM attributes must return zero. The value parsed from the textContent never affects the DOM values.

Would be cool to have the value DOM attribute update the textContent in-line...

If the progress bar is an indeterminate progress bar, then the position DOM attribute must return -1. Otherwise, it must return the result of dividing the current value by the maximum value.

3.12.13. The `code` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element when used with the dfn element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The code element represents a fragment of computer code. This could be an XML element name, a filename, a computer program, or any other string that a computer would recognise.

See the pre element for more detais.

The following example shows how a block of code could be marked up using the pre and code elements.

<pre><code>var i: Integer;
begin
   i := 1;
end.</code></pre>

3.12.14. The `var` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element when used with the dfn element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The var element represents a variable. This could be an actual variable in a mathematical expression or programming context, or it could just be a term used as a placeholder in prose.

In the paragraph below, the letter "n" is being used as a variable in prose:

<p>If there are <var>n</var> pipes leading to the ice
cream factory then I expect at <em>least</em> <var>n</var>
flavours of ice cream to be available for purchase!</p>

3.12.15. The `samp` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element when used with the dfn element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The samp element represents (sample) output from a program or computing system.

See the pre and kbd elements for more detais.

This example shows the samp element being used inline:

<p>The computer said <samp>Too much cheese in tray
two</samp> but I didn't know what that meant.</p>

This second example shows a block of sample output. Nested samp and kbd elements allow for the styling of specific elements of the sample output using a style sheet.

<pre><samp><samp class="prompt">jdoe@mowmow:~$</samp> <kbd>ssh demo.example.com</kbd>
Last login: Tue Apr 12 09:10:17 2005 from mowmow.example.com on pts/1
Linux demo 2.6.10-grsec+gg3+e+fhs6b+nfs+gr0501+++p3+c4a+gr2b-reslog-v6.189 #1 SMP Tue Feb 1 11:22:36 PST 2005 i686 unknown

<samp class="prompt">jdoe@demo:~$</samp> <samp class="cursor">_</samp></samp></pre>

3.12.16. The `kbd` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The kbd element represents user input (typically keyboard input, although it may also be used to represent other input, such as voice commands).

When the kbd element is nested inside a samp element, it represents the input as it was echoed by the system.

When the kbd element contains a samp element, it represents input based on system output, for example invoking a menu item.

When the kbd element is nested inside another kbd element, it represents an actual key or other single unit of input as appropriate for the input mechanism.

Here the kbd element is used to indicate keys to press:

<p>To make George eat an apple, press <kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd></p>

In this second example, the user is told to pick a particular menu item. The outer kbd element marks up a block of input, with the inner kbd elements representing each individual step of the input, and the samp elements inside them indicating that the steps are input based on something being displayed by the system, in this case menu labels:

<p>To make George eat an apple, select
    <kbd><kbd><samp>File</samp></kbd>|<kbd><samp>Eat Apple...</samp></kbd></kbd>
</p>

3.12.17. The `sup` and `sub` elements

Contexts in which these elements may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The sup element represents a superscript and the sub element represents a subscript.

These elements must only be used to mark up typographical conventions with specific meanings, not for typographical presentation for presentation's sake. For example, it would be inappropriate for the sup and sub elements to be used in the name of the LaTeX document preparation system. In general, authors should not use these elements if the absence of those elements would not change the meaning of the content.

When the sub element is used inside a var element, it represents the subscript that identifies the variable in a family of variables.

<p>The coordinate of the <var>i</var>th point is
(<var>x<sub><var>i</var></sub></var>, <var>y<sub><var>i</var></sub></var>).
For example, the 10th point has coordinate
(<var>x<sub>10</sub></var>, <var>y<sub>10</sub></var>).</p>

In certain languages, superscripts are part of the typographical conventions for some abbreviations.

<p>The most beautiful women are
<span lang="fr"><abbr>M<sup>lle</sup></abbr> Gwendoline</span> and 
<span lang="fr"><abbr>M<sup>me</sup></abbr> Denise</span>.</p>

Mathematical expressions often use subscripts and superscripts.

<var>E</var>=<var>m</var><var>c</var><sup>2</sup>

f(<var>x</var>, <var>n</var>) = log<sub>4</sub><var>x</var><sup><var>n</var></sup>

3.12.18. The `span` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: When used in an element whose content model is only strictly inline-level content: only strictly inline-level content.; Otherwise: any inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element when used with the dfn element.
Predefined classes that apply to this element:: copyright, error, example, issue, note, search, warning
DOM interface:: No difference from HTMLElement.

The span element doesn't mean anything on its own, but can be useful when used together with other attributes, e.g. class, lang, or dir, or when used in conjunction with the dfn element.

3.12.19. The `i` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None, but the title attribute has special semantics on this element when used with the dfn element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The i element represents a span of text in an alternate voice or mood, or otherwise offset from the normal prose, such as a taxonomic designation, a technical term, an idiomatic phrase from another language, a thought, a ship name, or some other prose whose typical typographic presentation is italicized.

Terms in languages different from the main text should be annotated with lang attributes (xml:lang in XML).

The examples below show uses of the i element:

<p>The <i>felis silvestris catus</i> is cute.</p>
<p>The <i>block-level elements</i> are defined above.</p>
<p>There is a certain <i lang="fr">je ne sais quoi</i> in the air.</p>

In the following example, a dream sequence is marked up using i elements.

<p>Raymond tried to sleep.</p>
<p><i>The ship sailed away on Thursday</i>, he
dreamt. <i>The ship had many people aboard, including a beautiful
princess called Carey. He watched her, day-in, day-out, hoping she
would notice him, but she never did.</i></p>
<p><i>Finally one night he picked up the courage to speak with
her—</i></p>
<p>Raymond woke with a start as the fire alarm rang out.</p>

The i element should be used as a last resort when no other element is more appropriate. In particular, citations should use the cite element, defining instances of terms should use the dfn element, stress emphasis should use the em element, importance should be denoted with the strong element, quotes should be marked up with the q element, and small print should use the small element.

Style sheets can be used to format i elements, just like any other element can be restyled. Thus, it is not the case that content in i elements will necessarily be italicised.

3.12.20. The `b` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The b element represents a span of text to be stylistically offset from the normal prose without conveying any extra importance, such as key words in a document abstract, product names in a review, or other spans of text whose typical typographic presentation is boldened.

The following example shows a use of the b element to highlight key words without marking them up as important:

<p>The <b>frobonitor</b> and <b>barbinator</b> components are fried.</p>

The following would be incorrect usage:

<p><b>WARNING!</b> Do not frob the barbinator!</p>

In the previous example, the correct element to use would have been strong, not b.

In the following example, objects in a text adventure are highlighted as being special by use of the b element.

<p>You enter a small room. Your <b>sword</b> glows
brighter. A <b>rat</b> scurries past the corner wall.</p>

The b element should be used as a last resort when no other element is more appropriate. In particular, headers should use the h1 to h6 elements, stress emphasis should use the em element, importance should be denoted with the strong element, and text marked or highlighted should use the m element.

Style sheets can be used to format b elements, just like any other element can be restyled. Thus, it is not the case that content in b elements will necessarily be boldened.

3.12.21. The `bdo` element

Contexts in which this element may be used:: Where strictly inline-level content is allowed.
Content model:: Strictly inline-level content.
Element-specific attributes:: None, but the dir global attribute is required on this element.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The bdo element allows authors to override the Unicode bidi algorithm by explicitly specifying a direction override. [BIDI]

Authors must specify the dir attribute on this element, with the value ltr to specify a left-to-right override and with the value rtl to specify a right-to-left override.

If the element has the dir attribute set to the exact value ltr, then for the purposes of the bidi algorithm, the user agent must act as if there was a U+202D LEFT-TO-RIGHT OVERRIDE character at the start of the element, and a U+202C POP DIRECTIONAL FORMATTING at the end of the element.

If the element has the dir attribute set to the exact value rtl, then for the purposes of the bidi algorithm, the user agent must act as if there was a U+202E RIGHT-TO-LEFT OVERRIDE character at the start of the element, and a U+202C POP DIRECTIONAL FORMATTING at the end of the element.

The requirements on handling the bdo element for the bidi algorithm may be implemented indirectly through the style layer. For example, an HTML+CSS user agent should implement these requirements by implementing the CSS unicode-bidi property. [CSS21]

3.13. Edits

The ins and del elements represent edits to the document.

3.13.1. The `ins` element

Transparent block-level element, and transparent strictly inline-level content.

Contexts in which this element may be used:: Where block-level elements is expected.; Where strictly inline-level content is allowed.
Content model:: Transparent.
Element-specific attributes:: cite; datetime
Predefined classes that apply to this element:: None.
DOM interface:: Uses the HTMLModElement interface.

The ins element represents an addition to the document.

The ins element must be used only where block-level elements or strictly inline-level content can be used.

An ins element can only contain content that would still be conformant if all elements with transparent content models were replaced by their contents.

The following would be syntactically legal:

<aside>
 <ins>
  <p>...</p>
 </ins>
</aside>

As would this:

<aside>
 <ins>
  <em>...</em>
 </ins>
</aside>

However, this last example would be illegal, as em and p cannot both be used inside an aside element at the same time:

<aside>
 <ins>
  <p>...</p>
 </ins>
 <ins>
  <em>...</em>
 </ins>
</aside>

3.13.2. The `del` element

Block-level element, and strictly inline-level content.

Contexts in which this element may be used:: Where block-level elements is expected.; Where strictly inline-level content is allowed.
Content model:: When the element has a parent: same content model as the parent element (without taking into account the other children of the parent element).; Otherwise: zero or more block-level elements, or inline-level content (but not both).
Element-specific attributes:: cite; datetime
Predefined classes that apply to this element:: None.
DOM interface:: Uses the HTMLModElement interface.

The del element represents a removal from the document.

The del element must only contain content that would be allowed inside the parent element (regardless of what the parent element actually contains).

The following would be syntactically legal:

<aside>
 <del>
  <p>...</p>
 </del>
 <ins>
  <em>...</em>
 </ins>
</aside>

...even though the p and em elements would never be allowed side by side in the aside element. This is allowed because the del element represents content that was removed, and it is quite possible that an edit could cause an element to go from being an inline-level container to a block-level container, or vice-versa.

3.13.3. Attributes common to `ins` and `del` elements

The cite attribute may be used to specify a URI that explains the change. When that document is long, for instance the minutes of a meeting, authors are encouraged to include a fragment identifier pointing to the specific part of that document that discusses the change.

If the cite attribute is present, it must be a URI (or IRI) that explains the change. User agents should allow users to follow such citation links.

The datetime attribute may be used to specify the time and date of the change.

If present, the datetime attribute must be a valid datetime value.

User agents must parse the datetime attribute according to the parse a string as a datetime value algorithm. If that doesn't return a time, then the modification has no associated timestamp (the value is non-conforming; it is not a valid datetime). Otherwise, the modification is marked as having been made at the given datetime. User agents should use the associated timezone information to determine which timezone to present the given datetime in.

The ins and del elements must implement the HTMLModElement interface:

interface HTMLModElement : HTMLElement {
           attribute DOMString cite;
           attribute DOMString datetime;
};

The cite and datetime DOM attributes must reflect the elements' content attributes of the same name.

3.14. Embedded content

3.14.1. The `figure` element

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: In any order, exactly one legend element, and exactly one embedded content element.
Element-specific attributes:: None.
Predefined classes that apply to this element:: example, warning
DOM interface:: No difference from HTMLElement.

The figure element represents a paragraph consisting of embedded content and a caption.

The first embedded content element child of the figure element, if any, is the paragraph's content.

The first legend element child of the element, if any, represents the caption of the embedded content. If there is no child legend element, then there is no caption.

If the embedded content cannot be used, then, for the purposes of establishing what the figure element represents:

If the embedded content's fallback content is a single embedded content element: The figure element must be treated as if that embedded content element was the figure element's embedded content. (If that embedded content can't be used either, then this processing must be done again, with the new embedded content's fallback content.)
If the embedded content's fallback is nothing: The entire figure element (including the caption, if any) must be ignored.
If the embedded content's fallback is inline-level content: The entire figure element (including the caption, if any) must be treated as being a single paragraph with that inline-level content as its content.
Otherwise: The entire figure element (including the caption, if any) must be treated as being replaced by that fallback content.

3.14.2. The `img` element

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

Empty.

Element-specific attributes:

alt (required)

src (required)

usemap

ismap (but only if one of the ancestor elements is an a element)

height

width

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLImageElement : HTMLElement {
           attribute DOMString alt;
           attribute DOMString src;
           attribute DOMString useMap;
           attribute boolean isMap;
           attribute long height;
           attribute long width;
  readonly attribute boolean complete;
};

An instance of HTMLImageElement can be obtained using the Image constructor.

The img element represents a piece of text with an alternate graphical representation. The text is given by the alt attribute, which must be present, and the URI to the graphical representation of that text is given in the src attribute, which must also be present.

The image given by the src attribute is the embedded content, and the value of the alt attribute is the img element's fallback content.

When the alt attribute's value is the empty string, the image supplements the surrounding content. In such cases, the image could be omitted without affecting the meaning of the document.

If the alt attribute is omitted, user agents must treat the element as if it had an alt attribute set to the empty string.

The alt attribute does not represent advisory information. User agents must not present the contents of the alt attribute in the same way as content of the title attribute.

Guidelines on writing "alt" text here.

The src attribute must contain a URI (or IRI). If the src attribute is omitted, there is no alternative image representation.

When the src attribute is set, the user agent must immediately begin to download the specified resource, unless the user agent cannot support images, or its support for images has been disabled.

The download of the image must delay the load event.

Once the download has completed, if the image is a valid image, the user agent must fire a load event on the img element. If the download fails or it completes but the image is not a valid or supported image, the user agent must fire an error event on the img element.

The remote server's response metadata (e.g. an HTTP 404 status code, or associated Content-Type headers) must be ignored when determining whether the resource obtained is a valid image or not.

This allows servers to return images with error responses.

User agents must not support non-image resources with the img element.

The usemap attribute, if present, can indicate that the image has an associated image map.

The ismap attribute, when used on an element that is a descendant of an a element with an href attribute, indicates by its presence that the element provides access to a server-side image map. This affects how events are handled on the corresponding a element.

The ismap attribute is a boolean attribute. The attribute must not be specified on an element that does not have an ancestor a element.

The height and width attributes give the preferred rendered dimensions of the image if the image is to be shown in a visual medium.

Should we require the dimensions to be correct? Should we disallow percentages?

The values of the height and width attributes must be either valid non-negative integers or valid non-negative percentages.

To parse the attributes, user agents must use the rules for parsing dimension values. This will return either an integer length, a percentage value, or nothing. When one of these attributes has no value, it must be ignored.

The user agent requirements for processing the values obtained from parsing these attributes are described in the rendering section.

The img element must be empty.

The DOM attributes alt, src, useMap, and isMap each must reflect the respective content attributes of the same name.

The DOM attributes height and width must return the rendered height and width of the image, in CSS pixels, if the image is being rendered, and is being rendered to a visual medium, or 0 otherwise. [CSS21]

The DOM attribute complete must return true if the user agent has downloaded the image specified in the src attribute, and it is a valid image.

3.14.3. The `iframe` element

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

Text (for details, see prose).

Element-specific attributes:

src

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLIFrameElement : HTMLElement {
           attribute DOMString src;
};

Objects implementing the HTMLIFrameElement interface must also implement the EmbeddingElement interface defined in the Window Object specification. [WINDOW]

The iframe element introduces a new nested browsing context.

The src attribute, if present, must be a URI (or IRI) to a page that the nested browsing context is to contain. When the browsing context is created, if the attribute is present, the user agent must navigate this browsing context to the given URI, with replacement enabled. If the user navigates away from this page, the iframe's corresponding Window object will reference new Document objects, but the src attribute will not change.

Whenever the src attribute is set, the nested browsing context must be navigated to the given URI.

If the src attribute is not set when the element is created, the browsing context will remain at the initial about:blank page.

The download of the resource must delay the load event.

When content loads in an iframe, the user agent must fire a load event at the iframe element. When content fails to load (e.g. due to a network error), then the user agent must fire an error event at the element instead.

order of events when content is also firing its own load/error?

An iframe element never has fallback content, as it will always create a nested browsing context, regardless of whether the specified initial contents are successfully used.

iframe elements may contain any text. iframe elements must not contain element nodes. Descendants of iframe elements represent nothing. (In legacy user agents that do not support iframe elements, the contents would be parsed as markup that could act as fallback content.)

restrictions for what that text must be?

The HTML parser treats markup inside iframe elements as text.

The DOM attribute src must reflect the content attribute of the same name.

3.14.4. The `embed` element

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

Empty.

Element-specific attributes:

src (required)

type

height

width

Any other attribute that has no namespace (see prose).

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLEmbedElement : HTMLElement {
           attribute DOMString src;
           attribute DOMString type;
           attribute long height;
           attribute long width;
};

Depending on the type of content instantiated by the embed element, the node may also support other interfaces.

The embed element represents an integration point for an external (typically non-HTML) application or interactive content.

The src attribute gives the address of the resource being embedded. The attribute must be present and contain a URI (or IRI).

If the src attribute is missing, then the embed element must be ignored.

When the src attribute is set, user agents are expected to find an appropriate handler for the specified resource, based on the content's type, and hand that handler the content of the resource. If the handler supports a scriptable interface, the HTMLEmbedElement object representing the element should expose that interfaces.

The download of the resource must delay the load event.

The user agent should pass the names and values of all the attributes of the embed element that have no namespace to the handler used. Any (namespace-less) attribute may be specified on the embed element.

This specification does not define a mechanism for interacting with third-party handlers, as it is expected to be user-agent-specific. Some UAs might opt to support a plugin mechanism such as the Netscape Plugin API; others may use remote content convertors or have built-in support for certain types. [NPAPI]

The embed element has no fallback content. If the user agent can't display the specified resource, e.g. because the given type is not supported, then the user agent must use a default handler for the content. (This default could be as simple as saying "Unsupported Format", of course.)

The type attribute, if present, gives the MIME type of the linked resource. The value must be a valid MIME type, optionally with parameters. [RFC2046]

The type of the content being embedded is defined as follows:

If the element has a type attribute, then the value of the type attribute is the content's type.
Otherwise, if the specified resource has explicit Content-Type metadata, then that is the content's type.
Otherwise, the content has no type and there can be no appropriate handler for it.

Should we instead say that the content-sniffing that we're going to define for top-level browsing contexts should apply here?

Should we require the type attribute to match the server information?

We should say that 404s, etc, don't affect whether the resource is used or not. Not sure how to say it here though.

Browsers should take extreme care when interacting with external content intended for third-party renderers. When third-party software is run with the same privileges as the user agent itself, vulnerabilities in the third-party software become as dangerous as those in the user agent.

height/width

The DOM attributes src and type each must reflect the respective content attributes of the same name.

3.14.5. The `object` element

Semi-transparent strictly inline-level embedded content.

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

When used as the child of a figure element, or, when used as a figure fallback object: Zero or more param elements, followed by either zero or more block-level elements or a single object element, which is then considered to be a figure fallback object.

Otherwise: Zero or more param elements, followed by inline-level content.

Element-specific attributes:

data (required if type is not given)

type (required if data is not given)

usemap

height

width

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLObjectElement : HTMLElement {
           attribute DOMString data;
           attribute DOMString type;
           attribute DOMString useMap;
           attribute long height;
           attribute long width;
};

Objects implementing the HTMLObjectElement interface must also implement the EmbeddingElement interface defined in the Window Object specification. [WINDOW]

Depending on the type of content instantiated by the object element, the node may also support other interfaces.

Shouldn't allow inline-level content to be the content model when the parent's content model is strictly inline only.

The object element can represent an external resource, which, depending on the type of the resource, will either be treated as an image, as a nested browsing context, or as an external resource to be processed by a third-party software package.

The data attribute, if present, specifies the address of the resource. If present, the attribute must be a URI (or IRI).

The type attribute, if present, specifies the type of the resource. If present, the attribute must be a valid MIME type, optionally with parameters. [RFC2046]

One or both of the data and type attributes must be present.

Whenever the data attribute changes, or, if the data attribute is not present, whenever the type attribute changes, the user agent must follow the following steps to determine what the object element represents:

If the data attribute is present, then:
1. Begin a load for the resource.
  
  The download of the resource must delay the load event.
2. If the resource is not yet available (e.g. because the resource was not available in the cache, so that loading the resource required making a request over the network), then jump to step 3 in the overall set of steps (fallback). When the resource becomes available, or if the load fails, restart this algorithm from this step. Resources can load incrementally; user agents may opt to consider a resource "available" whenever enough data has been obtained to begin processing the resource.
3. If the load failed (e.g. DNS error), fire an error event at the element, then jump to step 3 in the overall set of steps (fallback).
4. Determine the resource type, as follows:
  
  This says to trust the type. Should we instead use the same mechanism as for browsing contexts?
  
  If the resource has associated Content-Type metadata
  The type is the type specified in the resource's Content-Type metadata.
  Otherwise, if the type attribute is present
  The type is the type specified in the type attribute.
  Otherwise, there is no explicit type information
  The type is the sniffed type of the resource.
5. Handle the content as given by the first of the following cases that matches:
  
  If the resource requires a special handler (e.g. a plugin)
  
  The user agent should find an appropriate handler for the specified resource, based on the resource type found in the previous step, and pass the content of the resource to that handler. If the handler supports a scriptable interface, the HTMLObjectElement object representing the element should expose that interface. The handler is not a nested browsing context. If no appropriate handler can be found, then jump to step 3 in the overall set of steps (fallback).
  
  The user agent should pass the names and values of all the parameters given by param elements that are children of the object element to the handler used.
  
  This specification does not define a mechanism for interacting with third-party handlers, as it is expected to be user-agent-specific. Some UAs might opt to support a plugin mechanism such as the Netscape Plugin API; others may use remote content convertors or have built-in support for certain types. [NPAPI]
  
  this doesn't completely duplicate the navigation section, since it handles <param>, etc, but surely some work should be done to work with it
  
  If the type of the resource is an XML MIME type
  If the type of the resource is HTML
  If the type of the resource does not start with "image/"
  
  The object element must be associated with a nested browsing context, if it does not already have one. The element's nested browsing context must then be navigated to the given resource, with replacement enabled. (The data attribute of the object element doesn't get updated if the browsing context gets further navigated to other locations.)
  
  navigation might end up treating it as something else, because it can do sniffing. how should we handle that?
  
  If the resource is a supported image format, and support for images has not been disabled
  
  The object element represents the specified image. The image is not a nested browsing context.
  
  shouldn't we use the image-sniffing stuff here?
  
  Otherwise
  
  The object element represents the specified image, but the image cannot be shown. Jump to step 3 below in the overall set of steps (fallback).
6. The element's contents are not part of what the object element represents.
7. Once the resource is completely loaded, fire a load event at the element.
If the data attribute is absent but the type attribute is present, and if the user agent can find a handler suitable according to the value of the type attribute, then that handler should be used. If the handler supports a scriptable interface, the HTMLObjectElement object representing the element should expose that interface. The handler is not a nested browsing context. If no suitable handler can be found, jump to the next step (fallback).
(Fallback.) The object element doesn't represent anything except what the element's contents represent, ignoring any leading param element children. This is the element's fallback content.

In the absence of other factors (such as style sheets), user agents must show the user what the object element represents. Thus, the contents of object elements act as fallback content, to be used only when referenced resources can't be shown (e.g. because it returned a 404 error). This allows multiple object elements to be nested inside each other, targeting multiple user agents with different capabilities, with the user agent picking the best one it supports.

The usemap attribute, if present while the object element represents an image, can indicate that the object has an associated image map. The attribute must be ignored if the object element doesn't represent an image.

height/width

The DOM attributes data, type, useMap, height, and width each must reflect the respective content attributes of the same name.

3.14.6. The `param` element

Contexts in which this element may be used:

As a child of an object element, before any content other than param elements.

Content model:

Empty.

Element-specific attributes:

name (required)

value (required)

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLParamElement : HTMLElement {
           attribute DOMString name;
           attribute DOMString value;
};

The param element defines parameters for handlers invoked by object elements.

The name attribute gives the name of the parameter.

The value attribute gives the value of the parameter.

Both attributes must be present. They may have any value.

If both attributes are present, and if the parent element of the param is an object element, then the element defines a parameter with the given name/value pair.

The DOM attributes name and value must both reflect the respective content attributes of the same name.

3.14.7. The `video` element

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

If the element has a src attribute: transparent.

If the element does not have a src attribute: one or more source elements, then, transparent.

Element-specific attributes:

src

autoplay

start

loopstart

loopend

end

loopcount

controls

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLVideoElement : HTMLMediaElement {
  readonly attribute unsigned long videoWidth;
  readonly attribute unsigned long videoHeight;
};

A video element represents a video or movie.

Content may be provided inside the video element so that older Web browsers, which do not support video, can display text to the user informing them of how to access the video contents. User agents should not show this fallback content to the user.

The video element is a media element whose media data is ostensibly video data, possibly with associated audio data.

The src, autoplay, start, loopstart, loopend, end, loopcount, and controls attributes are the attributes common to all media elements.

The videoWidth DOM attribute must return the native width of the video in CSS pixels. The videoHeight DOM attribute must return the native height of the video in CSS pixels. In the absence of resolution information, user agents may assume that one pixel in the video corresponds to one CSS pixel. If no video data is available, then the attributes must return 0.

When no video data is available (the element's networkState attribute is either EMPTY, LOADING, or LOADED_METADATA), video elements represent nothing.

When a video element is actively playing, it represents the frame of video at the continuously increasing "current" position. When the current playback position changes such that the last frame rendered is no longer the frame corresponding to the current playback position in the video, the new frame must be rendered. Similarly, any audio associated with the video must, if played, be played synchronised with the current playback position, at the specified volume with the specified mute state.

When a video element is paused, the element represents the frame of video corresponding to the current playback position, or, if that is not available yet (e.g. because the video is seeking or buffering), the last rendered frame of video.

When a video element is neither actively playing nor paused, the element represents the last frame of the video to have been rendered.

Which frame in a video stream corresponds to a particular playback position is defined by the video stream's format.

Video content should be rendered inside the element's playback area such that the video content is shown centered in the playback area at the largest possible size that fits completely within it, with the video content's aspect ratio being preserved. Thus, if the aspect ratio of the playback area does not match the aspect ratio of the video, the video will be shown letterboxed. Areas of the element's playback area that do not contain the video represent nothing.

User agents should provide controls to enable or disable the display of closed captions associated with the video stream, though such features should, again, not interfere with the page's normal rendering.

User agents may allow users to view the video content in manners more suitable to the user (e.g. full-screen or in an independent resizable window). As for the other user interface features, controls to enable this should not interfere with the page's normal rendering unless the user agent is exposing a user interface. In such an independent context, however, user agents may make full user interfaces visible, with, e.g., play, pause, seeking, and volume controls, even if the controls attribute is absent.

User agents may allow video playback to affect system features that could interfere with the user's experience; for example, user agents could disable screensavers while video playback is in progress.

3.14.7.1. Video and audio codecs for `video` elements

User agents may support any video and audio codecs and container formats.

User agents should support Ogg Theora video and Ogg Vorbis audio, as well as the Ogg container format. [THEORA] [VORBIS] [OGG]

3.14.8. The `audio` element

Semi-transparent strictly inline-level embedded content.

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

If the element has a src attribute: transparent.

If the element does not have a src attribute: one or more source elements, then, transparent.

Element-specific attributes:

src

autoplay

start

loopstart

loopend

end

loopcount

controls

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLAudioElement : HTMLMediaElement {
  // no members
};

An audio element represents a sound or audio stream.

Content may be provided inside the audio element so that older Web browsers, which do not support audio, can display text to the user informing them of how to access the audio contents. User agents should not show this fallback content to the user.

The audio element is a media element whose media data is ostensibly audio data.

The src, autoplay, start, loopstart, loopend, end, loopcount, and controls attributes are the attributes common to all media elements.

When an audio element is actively playing, it must have its audio data played synchronised with the current playback position, at the specified volume with the specified mute state.

When an audio element is not actively playing, audio must not play for the element.

3.14.8.1. Audio codecs for `audio` elements

User agents may support any audio codecs and container formats.

User agents must support the WAVE container format with audio encoded using the PCM format.

3.14.9. Media elements

Media elements implement the following interface:

interface HTMLMediaElement : HTMLElement {

  // error state
  readonly attribute MediaError error;

  // network state
           attribute DOMString src;
  readonly attribute DOMString currentSrc;
  const unsigned short EMPTY = 0;
  const unsigned short LOADING = 1;
  const unsigned short LOADED_METADATA = 2;
  const unsigned short LOADED_FIRST_FRAME = 3;
  const unsigned short LOADED = 4;
  readonly attribute unsigned short networkState;
  readonly attribute float bufferingRate;
  readonly attribute TimeRanges buffered;
  void load();

  // ready state
  const unsigned short DATA_UNAVAILABLE = 0;
  const unsigned short CAN_SHOW_CURRENT_FRAME = 1;
  const unsigned short CAN_PLAY = 2;
  const unsigned short CAN_PLAY_THROUGH = 3;
  readonly attribute unsigned short readyState;
  readonly attribute boolean seeking;

  // playback state
           attribute float currentTime;
  readonly attribute float duration;
  readonly attribute unsigned short paused;
           attribute float defaultPlaybackRate;
           attribute float playbackRate;
  readonly attribute TimeRanges played;
  readonly attribute TimeRanges seekable;
  readonly attribute boolean ended;
           attribute boolean autoplay;
  void play();
  void pause();

  // looping
           attribute float start;
           attribute float end;
           attribute float loopStart;
           attribute float loopEnd;
           attribute unsigned long loopCount;
           attribute unsigned long currentLoop;

  // cue points
  void addCuePoint(in float time, in VoidCallback callback, in bool pause);
  void removeCuePoint(in float time, in VoidCallback callback);

  // controls
           attribute boolean controls;
           attribute float volume;
           attribute boolean muted;
};

The media element attributes, src, autoplay, start, loopstart, loopend, end, loopcount, and controls, apply to all media elements. They are defined in this section.

Media elements are used to present audio data, or video and audio data, to the user. This is referred to as media data in this section, since this section applies equally to media elements for audio or for video. The term media resource is used to refer to the complete set of media data, e.g. the complete video file, or complete audio file.

3.14.9.1. Error codes

All media elements have an associated error status, which records the last error the element encountered since the load() method was last invoked. The error attribute, on getting, must return the MediaError object created for this last error, or null if there has not been an error.

interface MediaError {
  const unsigned short MEDIA_ERR_ABORTED = 1;
  const unsigned short MEDIA_ERR_NETWORK = 2;
  const unsigned short MEDIA_ERR_DECODE = 3;
  readonly attribute unsigned short code;
};

The code attribute of a MediaError object must return the code for the error, which must be one of the following:

MEDIA_ERR_ABORTED (numeric value 1): The download of the media resource was aborted by the user agent at the user's request.
MEDIA_ERR_NETWORK (numeric value 2): A network error of some description caused the user agent to stop downloading the media resource.
MEDIA_ERR_DECODE (numeric value 3): An error of some description occurred while decoding the media resource.

3.14.9.2. Location of the media resource

The src content attribute on media elements gives the address of the video to show. The attribute, if present, must contain a URI (or IRI).

If a src attribute is specified, the resource it specifies is the media resource that will be used. Otherwise, the resource specified by the first suitable source element child of the media element is the one used.

The src DOM attribute on media elements must reflect the content attribute of the same name.

To pick a media resource for a media element, a user agent must follow the following steps:

If the media element has a src, then the address given in that attribute is the address of the media resource; jump to the last step.
Otherwise, let candidate be the first source element child in the media element, or null if there is no such child.
If either:
- candidate is null, or
- the candidate element has no src attribute, or
- the candidate element has a type attribute and that attribute's value, when parsed as a MIME type, does not represent a type that the user agent can render (including any codecs described by the codec parameter), or [RFC2046] [RFC4281]
- the candidate element has a media attribute and that attribute's value, when processed according to the rules for media queries, does not match the current environment, [MQ]
...then the candidate is not suitable; go to the next step.

Otherwise, the address given in that candidate element's src attribute is the address of the media resource; jump to the last step.
Let candidate be the next source element child in the media element, or null if there are no more such children.
If candidate is not null, return to step 3.
There is no media resource. Abort these steps.
Let the address of the chosen media resource be the one that was found before jumping to this step.

The currentSrc DOM attribute must return the empty string if the media element's networkState has the value EMPTY, and the absolute URL of the chosen media resource otherwise.

3.14.9.3. Network states

As media elements interact with the network, they go through several states. The networkState attribute, on getting, must return the current network state of the element, which must be one of the following values:

EMPTY (numeric value 0): The element has not yet been initialised. All attributes are in their initial states.
LOADING (numeric value 1): The element has picked a media resource (the chosen media resource is available from the currentSrc attribute), but none of the metadata has yet been obtained and therefore all the other attributes are still in their initial states.
LOADED_METADATA (numeric value 2): Enough of the resource has been obtained that the metadata attributes are initialized (e.g. the length is known). The API will no longer raise exceptions when used.
LOADED_FIRST_FRAME (numeric value 3): Actual media data has been obtained. In the case of video, this specifically means that a frame of video is available and can be shown.
LOADED (numeric value 4): The entire media resource has been obtained and is available to the user agent locally. Network connectivity could be lost without affecting the media playback.

The algorithm for the load() method defined below describes exactly when the networkState attribute changes value.

3.14.9.4. Loading the media resource

All media elements have a begun flag, which must begin in the false state, a loaded-first-frame flag, which must begin in the false state, and an autoplaying flag, which must begin in the true state.

When the load() method on a media element is invoked, the user agent must run the following steps. Note that this algorithm might get aborted, e.g. if the load() method itself is invoked again.

Any already-running instance of this algorithm for this element must be aborted. If those method calls have not yet returned, they must finish the step they are on, and then immediately return.
If the element's begun flag is true, then the begun flag must be set to false, the error attribute must be set to a new MediaError object whose code attribute is set to MEDIA_ERR_ABORTED, and the user agent must synchronously fire a progress event called abort at the media element.
The error attribute must be set to null, and the loaded-first-frame flag and loaded-enough-to-play-through flag must be both set to false.
The playbackRate attribute must be set to the value of the defaultPlaybackRate attribute.
If the media element's networkState is not set to EMPTY, then the following substeps must be followed:
1. The networkState attribute must be set to EMPTY.
2. If readyState is not set to DATA_UNAVAILABLE, it must be set to that state.
3. If the paused attribute is false, it must be set to true.
4. If seeking is true, it must be set to false.
5. The current playback position must be set to 0.
6. The currentLoop DOM attribute must be set to 0.
7. The user agent must synchronously fire a simple event called emptied at the media element.
The user agent must pick a media resource for the media element. If that fails, the method must raise an INVALID_STATE_ERR exception, and abort these steps.
The networkState attribute must be set to LOADING.
The currentSrc attribute starts returning the new value.
The user agent must then set the begun flag to true and fire a progress event called begin at the media element.
The method must return, but these steps must continue.
Playback of any previously playing media resource for this element stops.
If a download is in progress for the media element, the user agent should stop the download.
The user agent must then begin to download the chosen media resource. The rate of the download may be throttled, however, in response to user preferences (including throttling it to zero until the user indicates that the download can start), or to balance the download with other connections sharing the same bandwidth.
While the download is progressing, the user agent must fire a progress event called progress at the element every 350ms (±200ms) or for every byte received, whichever is least frequent.

If at any point the user agent has received no data for more than about three seconds, the user agent must fire a progress event called stalled at the element.

User agents may allow users to selectively block or slow media data downloads. When a media element's download has been blocked, the user agent must act as if it was stalled (as opposed to acting as if the connection was closed).

The user agent may use whatever means necessary to download the resource (within the constraints put forward by this and other specifications); for example, reconnecting to the server in the face of network errors, using HTTP partial range requests, or switching to a streaming protocol. The user agent must only consider a resource erroneous if it has given up trying to download it.
If the media data cannot be downloaded at all, due to network errors, causing the user agent to give up trying to download the resource
DNS errors and HTTP 4xx and 5xx errors (and equivalents in other protocols) must cause the user agent to follow the following steps. User agents may also follow these steps in response to other network errors of similar severity.
1. The user agent should cancel the download.
2. The error attribute must be set to a new MediaError object whose code attribute is set to MEDIA_ERR_NETWORK.
3. The begun flag must be set to false and the user agent must fire a progress event called error at the media element.
4. The element's networkState attribute must be switched to the EMPTY value and the user agent must fire a simple event called emptied at the element.
5. These steps must be aborted.
If the media data can be downloaded but is in an unsupported format, or can otherwise not be properly rendered at all
The server returning a file of the wrong kind (e.g. one that that turns out to not be pure audio when the media element is a audio element), or the file using unsupported codecs for all the data, must cause the user agent to follow the following steps. User agents may also follow these steps in response to other codec-related fatal errors, such as the file requiring more resources to process than the user agent can provide in real time.
1. The user agent should cancel the download.
2. The error attribute must be set to a new MediaError object whose code attribute is set to MEDIA_ERR_DECODE.
3. The begun flag must be set to false and the user agent must fire a progress event called error at the media element.
4. The element's networkState attribute must be switched to the EMPTY value and the user agent must fire a simple event called emptied at the element.
5. These steps must be aborted.
If the media data download is aborted by the user
The download is aborted by the user, e.g. because the user navigated the browsing context to another page, the user agent must follow the following steps. These steps are not followed if the load() method itself is reinvoked, as the steps above handle that particular kind of abort.
1. The user agent should cancel the download.
2. The error attribute must be set to a new MediaError object whose code attribute is set to MEDIA_ERR_ABORT.
3. The begun flag must be set to false and the user agent must fire a progress event called abort at the media element.
4. If the media element's networkState attribute has the value LOADING, the element's networkState attribute must be switched to the EMPTY value and the user agent must fire a simple event called emptied at the element.
5. These steps must be aborted.
If the media data can be downloaded but has non-fatal errors or uses, in part, codecs that are unsupported, preventing the user agent from rendering the content completely correctly but not preventing playback altogether
The server returning data that is partially usable but cannot be optimally rendered must cause the user agent to follow the following steps.
1. Should we fire a 'warning' event? Set the 'error' flag to 'MEDIA_ERR_SUBOPTIMAL' or something?
Once enough of the media data has been downloaded to determine the duration of the media resource, its dimensions, and other metadata
The user agent must follow these substeps:
1. The current playback position must be set to the effective start.
2. The networkState attribute must be set to LOADED_METADATA.
3. A number of attributes, including duration, buffered, and played, become available.
4. The user agent will fire a simple event called durationchange at the element at this point.
5. The user agent must fire a simple event called loadedmetadata at the element.
Once enough of the media data has been downloaded to enable the user agent to display the first frame of the media resource
The user agent must follow these substeps:
1. The networkState attribute must be set to LOADED_FIRST_FRAME.
2. The readyState attribute must change to CAN_SHOW_CURRENT_FRAME.
3. The loaded-first-frame flag must be set to true.
4. The user agent must fire a simple event called loadedfirstframe at the element.
5. The user agent must fire a simple event called canshowcurrentframe at the element.
When the user agent has completed the download of the entire media resource, it must move on to the next step.
If the download completes without errors, the begun flag must be set to false and the user agent must fire a progress event called load at the element.

If a media element whose networkState has the value EMPTY is inserted into a document, user agents must implicitly invoke the load() method on the media element as soon as all other scripts have finished executing. Any exceptions raised must be ignored.

The bufferingRate attribute must return the average number of bits received per second for the current download over the past few seconds. If there is no download in progress, the attribute must return 0.

The buffered attribute must return a static normalised TimeRanges object that represents the ranges of the media resource, if any, that the user agent has downloaded, at the time the attribute is evaluated.

Typically this will be a single range anchored at the zero point, but if, e.g. the user agent uses HTTP range requests in response to seeking, then there could be multiple ranges.

3.14.9.5. Offsets into the media resource

The duration attribute must return the length of the media resource, in seconds. If no media data is available, then the attributes must return 0. If media data is available but the length is not known, the attribute must return the Not-a-Number (NaN) value. If the media resource is known to be unbounded (e.g. a streaming radio), then the attribute must return the positive Infinity value.

When the length of the media resource changes (e.g. from being unknown to known, or from indeterminate to known, or from a previously established length to a new length) the user agent must, once any running scripts have finished, fire a simple event called durationchange at the media element.

Media elements have a current playback position, which must initially be zero. The current position is a time.

The currentTime attribute must, on getting, return the current playback position, expressed in seconds. On setting, the user agent must seek to the new value.

The start content attribute gives the offset into the media resource at which playback is to begin. The default value is 0.

The effective start is the smaller of start and the end of the media resource.

The loopstart content attribute gives the offset into the media resource at which playback is to begin when looping a clip. The default value is 0.

The effective loop start is the smaller of loopStart and the end of the media resource.

The loopend content attribute gives an offset into the media resource at which playback is to jump back to the loopstart, when looping the clip. The default value is infinity.

The effective loop end is the greater of start, loopStart, and loopEnd, and the end of the media resource.

The end content attribute gives an offset into the media resource at which playback is to end. The default value is infinity.

The effective end is the greater of start, loopStart, end, and the end of the media resource.

The start, loopstart, loopend, and end attributes must, if specified, contain value time offsets. To get the time values they represent, user agents must use the rules for parsing time offsets.

The start, loopStart, loopEnd, and end DOM attributes must reflect the start, loopstart, loopend, and end content attributes on the media element respectively.

The loopcount content attribute gives the number of times to play the clip. The default value is 1.

The loopCount DOM attribute must reflect the loopcount content attribute on the media element. The value must be limited to only positive non-zero numbers.

The currentLoop attribute must initially have the value 0. It gives the index of the current loop. It is changed during playback as described below.

When any of the start, loopStart, loopEnd, end, and loopCount DOM attributes change value (either through content attribute mutations reflecting into the DOM attribute, or direct mutations of the DOM attribute), the user agent must apply the following steps:

If the loopCount DOM attribute's value is less than the currentLoop DOM attribute's value, then the currentLoop DOM attribute's value must be set to the value of the loopCount DOM attribute's value (which will make the current loop the last loop).
If the media element's networkState is in the EMPTY state or the LOADING state, then the user agent must at this point abort these steps.
If the currentLoop is zero, and the current playback position is before the effective start, the user agent must seek to the effective start.
If the currentLoop is greater than zero, and the current playback position is before the effective loop start, the user agent must seek to the effective loop start.
If the currentLoop is less than loopCount, and the current playback position is after the effective loop end, the user agent must seek to the effective loop start, and increase loopCount by 1.
If the currentLoop is equal to the loopCount, and the current playback position is after the effective end, the user agent must seek to the effective end and then the looping will end.

3.14.9.6. The ready states

Media elements have a ready state, which describes to what degree they are ready to be rendered at the current playback position. The possible values are as follows; the ready state of a media element at any particular time is the greatest value describing the state of the element:

DATA_UNAVAILABLE (numeric value 0): No data for the current playback position is available. Media elements whose networkState attribute is less than LOADED_FIRST_FRAME are always in the DATA_UNAVAILABLE state.
CAN_SHOW_CURRENT_FRAME (numeric value 1): Data for the immediate current playback position is available, but not enough data is available that the user agent could successfully advance the current playback position at all without immediately reverting to the DATA_UNAVAILABLE state. In video, this corresponds to the user agent having data from the current frame, but not the next frame. In audio, this corresponds to the user agent only having audio up to the current playback position, but no further.
CAN_PLAY (numeric value 2): Data for the immediate current playback position is available, as well as enough data for the user agent to advance the current playback position at least a little without immediately reverting to the DATA_UNAVAILABLE state. In video, this corresponds to the user agent having data for the current frame and the next frame. In audio, this corresponds ot the user agent having data beyond the current playback position.
CAN_PLAY_THROUGH (numeric value 3): Data for the immediate current playback position is available, as well as enough data for the user agent to advance the current playback position at least a little without immediately reverting to the DATA_UNAVAILABLE state, and, in addition, the user agent estimates that data is being downloaded at a rate where the current playback position, if it were to advance at the rate given by the defaultPlaybackRate attribute, would not overtake the available data before playback reaches the effective end of the media resource on the last loop.

When the ready state of a media element whose networkState is not EMPTY changes, the user agent must follow the steps given below:

If the new ready state is DATA_UNAVAILABLE

The user agent must fire a simple event called dataunavailable at the element.

If the new ready state is CAN_SHOW_CURRENT_FRAME

If the element's loaded-first-frame flag is true, the user agent must fire a simple event called canshowcurrentframe event.

The first time the networkState attribute switches to this value, the loaded-first-frame flag is false, and the event is fired by the algorithm described above for the load() method, in conjunction with other steps.

If the new ready state is CAN_PLAY

The user agent must fire a simple event called canplay.

If the new ready state is CAN_PLAY_THROUGH

The user agent must fire a simple event called canplaythrough event. If the autoplaying flag is true, and the paused attribute is true, and the media element has an autoplay attribute specified, then the user agent must also set the paused attribute to false and fire a simple event called play.

The readyState DOM attribute must, on getting, return the value described above that describes the current ready state of the media element.

The autoplay attribute is a boolean attribute. When present, the algorithm described herein will cause the user agent to automatically begin playback of the media resource as soon as it can do so without stopping.

The autoplay DOM attribute must reflect the content attribute of the same name.

3.14.9.7. Playing the media resource

The paused attribute represents whether the media element is paused or not. The attribute must initially be true.

A media element is said to be actively playing when its paused attribute is false, the readyState attribute is either CAN_PLAY or CAN_PLAY_THROUGH, the element has not ended playback, playback has not stopped due to errors, and the element has not paused for user interaction.

A media element is said to have ended playback when the element's networkState attribute is LOADED_METADATA or greater, the current playback position is equal to the effective end of the media resource, and the currentLoop attribute is equal to the loopCount DOM attribute.

A media element is said to have stopped due to errors when the element's networkState attribute is LOADED_METADATA or greater, and the user agent has encounters a non-fatal error during the processing of the media data, and due to that error, is not able to play the content at the current playback position.

A media element is said to have paused for user interaction when its paused attribute is false, the readyState attribute is either CAN_PLAY or CAN_PLAY_THROUGH and the user agent has reached a point in the media resource where the user has to make a selection for the resource to continue.

It is possible for a media element to have both ended playback and paused for user interaction at the same time.

When a media element is actively playing, its current playback position must increase monotonically at playbackRate units of media time per unit time of wall clock time. If this value is not 1, the user agent may apply pitch adjustments to any audio component of the media resource.

Media resources might be internally scripted or interactive. Thus, a media element could play in a non-linear fashion. If this happens, the user agent must act as if the algorithm for seeking was used whenever the current playback position changes in a discontinuous fashion (so that the relevant events fire).

When a media element that is actively playing stops playing because its readyState attribute changes to a value lower than CAN_PLAY, without the element having ended playback, or playback having stopped due to errors, or playback having paused for user interaction, the user agent must fire a simple event called timeupdate at the element, and then must fire a simple event called waiting at the element.

When a media element that is actively playing stops playing because it has paused for user interaction, the user agent must fire a simple event called timeupdate at the element.

When currentLoop is less than loopCount-1 and the current playback position reaches the effective loop end, then the user agent must seek to the effective loop start, increase loopCount by 1, and fire a simple event called timeupdate.

When currentLoop is equal to the loopCount-1 and the current playback position reaches the effective end, then the user agent must follow these steps:

The user agent must stop playback.
The ended attribute becomes true, as described below.
The user agent must fire a simple event called timeupdate at the element.
The user agent must fire a simple event called ended at the element.

The defaultPlaybackRate attribute gives the desired speed at which the media resource is to play, as a multiple of its intrinsic speed. The attribute is mutable, but on setting, if the new value is 0.0, a NOT_SUPPORTED_ERR exception must be raised instead of the value being changed. It must initially have the value 1.0.

The playbackRate attribute gives the speed at which the media resource plays, as a multiple of its intrinsic speed. If it is not equal to the defaultPlaybackRate, then the implication is that the user is using a feature such as fast forward or slow motion playback. The attribute is mutable, but on setting, if the new value is 0.0, a NOT_SUPPORTED_ERR exception must be raised instead of the value being changed. Otherwise, the playback must change speed (if the element is actively playing). It must initially have the value 1.0.

When the defaultPlaybackRate or playbackRate attributes change value (either by being set by script or by being changed directly by the user agent, e.g. in response to user control) the user agent must, once any running scripts have finished, fire a simple event called ratechange at the media element.

When the play() method on a media element is invoked, the user agent must run the following steps.

If the media element's networkState attribute has the value EMPTY, then the user agent must sychronously invoke the load() method. If that raises an exception, that exception must be reraised by the play() method.
If the playback has ended, then the user agent must set currentLoop to zero and seek to the effective start.
The playbackRate attribute must be set to the value of the defaultPlaybackRate attribute.
If the media element's paused attribute is true, it must be set to false.
The media element's autoplaying flag must be set to false.
The method must then return.

If the second step above involved a seek, the user agent will fire a simple event called timeupdate at the media element.

If the third step above caused the playbackRate attribute to change value, the user agent will fire a simple event called ratechange at the media element.

If the fourth step above changed the value of paused, the user agent must fire a simple event called play at the media element.

When the pause() method is invoked, the user agent must run the following steps:

If the media element's networkState attribute has the value EMPTY, then the user agent must sychronously invoke the load() method. If that raises an exception, that exception must be reraised by the play() method.
If the media element's paused attribute is false, it must be set to true.
The media element's autoplaying flag must be set to false.
The method must then return.
If the second step above changed the value of paused, the user agent must first fire a simple event called timeupdate at the element, and then fire a simple event called title="event-pause">pause at the element.

The ended attribute must return true if the media element has ended playback, and false otherwise.

The played attribute must return a static normalised TimeRanges object that represents the ranges of the media resource, if any, that the user agent has so far rendered, at the time the attribute is evaluated.

3.14.9.8. Seeking

The seeking attribute must initially have the value false.

When the user agent is required to seek to a particular new playback position in the media resource, it means that the user agent must run the following steps:

If the media element's networkState is less than LOADED_METADATA, then the user agent must raise an INVALID_STATE_ERR exception (if the seek was in response to a DOM method call or setting of a DOM attribute), and abort these steps.
If currentLoop is 0, let min be the effective start. Otherwise, let it be the effective loop start.
If currentLoop is equal to the value of loopCount, let max be the effective end. Otherwise, let it be the effective loop end.
If the new playback position is more than max, let it be max.
If the new playback position is less than min, let it be min.
If the (possibly now changed) new playback position is not in one of the ranges given in the seekable attribute, then the user agent must raise an INDEX_SIZE_ERR exception (if the seek was in response to a DOM method call or setting of a DOM attribute), and abort these steps.
The current playback position must be set to the given new playback position.
The seeking DOM attribute must be set to true.
The user agent must fire a simple event called timeupdate at the element.
As soon as the user agent has established whether or not the media data for the new playback position is available, and, if it is, decoded enough data to play back that position, the seeking DOM attribute must be set to false.

The seekable attribute must return a static normalised TimeRanges object that represents the ranges of the media resource, if any, that the user agent is able to seek to, at the time the attribute is evaluated, notwithstanding the looping attributes (i.e. the effective start and effective end, etc, don't affect the seeking attribute).

If the user agent can seek to anywhere in the media resource, e.g. because the user agent and the server support HTTP Range requests, then the attribute would return an object with one range, whose start is 0, and whose end is the same as the duration attribute's value.

3.14.9.9. Cue points

Media elements have an ordered list of (time, callback) tuples called the cue point list. Each entry in the list also has a boolean pause associated with it.

The addCuePoint(time, callback, pause) method must, when called, add a tuple made of the given time and callback to the element's cue point list, and associate with that entry the value of the pause argument.

The removeCuePoint(time, callback) method must, when called, remove the first tuple matching the given time and callback from the element's cue point list.

When the current playback position of a media element reaches one of the times given in the element's cue point list, the user agent must follow these steps:

First, if any of the entries in the cue point list with that time have their associated pause boolean set to true, then the user agent must immediately act as if the element's pause() method had been invoked.
The user agent must then fire a simple event called timeupdate at the element.
The user agent must then invoke all the non-null callbacks for all the entries in the list that match the current playback position time, in the order they were added to the list.

Invoking a callback (an object implementing the VoidCallback interface) means calling its handleEvent() method.

interface VoidCallback {
  void handleEvent();
};

The handleEvent method of objects implementing the VoidCallback interface is the entrypoint for the callback represented by the object.

In the ECMAScript DOM binding, the ECMAScript native Function type must implement the VoidCallback interface such that invoking the handleEvent() method of that interface on the object from another language binding invokes the function itself. In the ECMAScript binding itself, however, the handleEvent() method of the interface is not directly accessible on Function objects. Such functions, when invoked, must be called at the scope of the browsing context.

3.14.9.10. User interface

The controls attribute is a boolean attribute. If the attribute is present, or if scripting is disabled, then the user agent should expose a user interface to the user. This user interface should include features to begin playback, pause playback, seek to an arbitrary position in the content (if the content supports arbitrary seeking), change the volume, and showe the media content in manners more suitable to the user (e.g. full-screen video or in an independent resizable window). Other controls may also be made available.

If the attribute is absent, then the user agent should avoid making a user interface available that could conflict with an author-provided user interface. User agents may make the following features available, however, even when the attribute is absent:

User agents may provide controls to affect playback of the media resource (e.g. play, pause, seeking, and volume controls), but such features should not interfere with the page's normal rendering. For example, such features could be exposed in the media element's context menu.

Where possible (specifically, for starting, stopping, pausing, and unpausing playback, for muting or changing the volume of the audio, and for seeking), user interface features exposed by the user agent must be implemented in terms of the DOM API described above, so that, e.g., all the same events fire.

The controls DOM attribute must reflect the content attribute of the same name.

The volume attribute must return the playback volume of any audio portions of the media element, in the range 0.0 (silent) to 1.0 (loudest). Initially, the volume must be 0.5, but user agents may remember the last set value across sessions, on a per-site basis or otherwise, so the volume may start at other values. On setting, if the new value is in the range 0.0 to 1.0 inclusive, the attribute must be set to the new value and the playback volume must be correspondingly adjusted as soon as possible after setting the attribute, with 0.0 being silent, and 1.0 being the loudest setting, values in between increasing in loudness. The range need not be linear. The loudest setting may be lower than the system's loudest possible setting; for example the user could have set a maximum volume. If the new value is outside the range 0.0 to 1.0 inclusive, then, on setting, an INDEX_SIZE_ERR exception must be raised instead.

The muted attribute must return true if the audio channels are muted and false otherwise. On setting, the attribute must be set to the new value; if the new value is true, audio playback for this media resource must then be muted, and if false, audio playback must then be enabled.

Whenever either the muted or volume attributes are changed, after any running scripts have finished executing, the user agent must fire a simple event called volumechange at the media element.

3.14.9.11. Time range

Objects implementing the TimeRanges interface represent a list of ranges (periods) of time.

interface TimeRanges {
  readonly attribute unsigned long length;
  float start(in unsigned long index);
  float end(in unsigned long index);
};

The length DOM attribute must return the number of ranges represented by the object.

The start(index) method must return the position of the start of the indexth range represented by the object, in seconds measured from the start of the timeline that the object covers.

The end(index) method must return the position of the end of the indexth range represented by the object, in seconds measured from the start of the timeline that the object covers.

When a TimeRanges object is said to be a normalised TimeRanges object, the ranges it represents must obey the following criteria:

The start of a range must be greater than the end of all earlier ranges.
The start of a range must be less than the end of that same range.

In other words, the ranges in such an object are ordered, don't overlap, and don't touch (adjacent ranges are folded into one bigger range).

The timelines used by the objects returned by the buffered, seekable and played DOM attributes of media elements must be the same as that element's media resource's timeline.

3.14.9.12. Event summary

The following events fire on media elements as part of the processing model described above:

Event name	Interface	Dispatched when...	Preconditions
`begin`	`ProgressEvent` [PROGRESS]	The user agent begins fetching the media data, synchronously during the `load()` method call.	`networkState` equals `LOADING`
`progress`	`ProgressEvent` [PROGRESS]	The user agent is fetching media data.	`networkState` is more than `EMPTY` and less than `LOADED`
`loadedmetadata`	`Event`	The user agent is fetching media data, and the media resource's metadata has just been received.	`networkState` equals `LOADED_METADATA`
`loadedfirstframe`	`Event`	The user agent is fetching media data, and the media resource's metadata has just been received.	`networkState` equals `LOADED_FIRST_FRAME`
`load`	`ProgressEvent` [PROGRESS]	The user agent finishes downloading the entire media resource.	`networkState` equals `LOADED`
`abort`	`ProgressEvent` [PROGRESS]	The user agent stops fetching the media data before it is completely downloaded. This can be fired synchronously during the `load()` method call.	`error` is an object with the code `MEDIA_ERR_ABORTED`. `networkState` equals either `EMPTY` or `LOADED`, depending on when the download was aborted.
`error`	`ProgressEvent` [PROGRESS]	An error occurs while fetching the media data.	`error` is an object with the code `MEDIA_ERR_NETWORK_ERROR` or higher. `networkState` equals either `EMPTY` or `LOADED`, depending on when the download was aborted.
`emptied`	`Event`	A media element whose `networkState` was previously not in the `EMPTY` state has just switched to that state (either because of a fatal error during load that's about to be reported, or because the `load()` method was reinvoked, in which case it is fired synchronously during the `load()` method call).	`networkState` is `EMPTY`; all the DOM attributes are in their initial states.
`stalled`	`ProgressEvent`	The user agent is trying to fetch media data, but data is unexpectedly not forthcoming.
`play`	`Event`	Playback has begun. Fired after the `play` method has returned.	`paused` is newly false.
`pause`	`Event`	Playback has been paused. Fired after the `pause` method has returned.	`paused` is newly true.
`waiting`	`Event`	Playback has stopped because the next frame is not available, but the user agent expects that frame to become available in due course.	`readyState` is either `DATA_UNAVAILABLE` or `CAN_SHOW_CURRENT_FRAME`, and `paused` is false. Either `seeking` is true, or the current playback position is not contained in any of the ranges in `buffered`. It is possible for playback to stop for two other reasons without `paused` being false, but those two reasons do not fire this event: maybe playback ended, or playback stopped due to errors.
`timeupdate`	`Event`	The current playback position changed in an interesting way, for example discontinuously.
`ended`	`Event`	Playback has stopped because the end of the media resource was reached.	`currentTime` equals the `effective end`; `ended` is true.
`dataunavailable`	`Event`	The user agent cannot render the data at the current playback position because data for the current frame is not immediately available.	The `readyState` attribute is newly equal to `DATA_UNAVAILABLE`.
`canshowcurrentframe`	`Event`	The user agent cannot render the data after the current playback position because data for the next frame is not immediately available.	The `readyState` attribute is newly equal to `CAN_SHOW_CURRENT_FRAME`.
`canplay`	`Event`	The user agent can resume playback of the media data, but estimates that if playback were to be started now, the media resource could not be rendered at the current playback rate up to its end without having to stop for further buffering of content.	The `readyState` attribute is newly equal to `CAN_PLAY`.
`canplaythrough`	`Event`	The user agent estimates that if playback were to be started now, the media resource could be rendered at the current playback rate all the way to its end without having to stop for further buffering.	The `readyState` attribute is newly equal to `CAN_PLAY_THROUGH`.
`ratechange`	`Event`	Either the `defaultPlaybackRate` or the `playbackRate` attribute has just been updated.
`durationchange`	`Event`	The `duration` attribute has just been updated.
`volumechange`	`Event`	Either the `volume` attribute or the `muted` attribute has changed. Fired after the relevant attribute's setter has returned.

3.14.9.13. Security and privacy considerations

Talk about making sure interactive media files (e.g. SVG) don't have access to the container DOM (XSS potential); talk about not exposing any sensitive data like metadata from tracks in the media files (intranet snooping risk)

3.14.10. The `source` element

Contexts in which this element may be used:

As a child of a media element, before any content other than source elements.

Content model:

Empty.

Element-specific attributes:

src (required)

type

media

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLSourceElement : HTMLElement {
           attribute DOMString src;
           attribute DOMString type;
           attribute DOMString media;
};

The source element allows authors to specify multiple media resources for media elements.

The src attribute gives the address of the media resource. The value must be a URI (or IRI). This attribute must be present.

The type attribute gives the type of the media resource, to help the user agent determine if it can play this media resource before downloading it. Its value must be a MIME type. The codec parameter may be specified and might be necessary to specify exactly how the resource is encoded. [RFC2046] [RFC4281]

The media attribute gives the intended media type of the media resource, to help the user agent determine if this media resource is useful to the user before downloading it. Its value must be a valid media query. [MQ]

Either the type attribute, the media attribute or both, must be specified, unless this is the last source element child of the parent element.

If a source element is inserted into a media element that is already in a document and whose networkState is in the EMPTY state, the user agent must implicitly invoke the load() method on the media element as soon as all other scripts have finished executing. Any exceptions raised must be ignored.

The DOM attributes src, type, and media must reflect the respective content attributes of the same name.

3.14.11. The `canvas` element

Contexts in which this element may be used:

As the only embedded content child of a figure element.

Where strictly inline-level content is allowed.

Content model:

Inline-level content.

Element-specific attributes:

height

width

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLCanvasElement : HTMLElement {
         attribute long width;
         attribute long height;

  DOMString toDataURL();
  DOMString toDataURL(in DOMString type);

  DOMObject getContext(in DOMString contextID);
};

Shouldn't allow inline-level content to be the content model when the parent's content model is strictly inline only.

The canvas element represents a resolution-dependent bitmap canvas, which can be used for rendering graphs, game graphics, or other visual images on the fly.

Authors should not use the canvas element in a document when a more suitable element is available. For example, it is inappropriate to use a canvas element to render a page heading: if the desired presentation of the heading is graphically intense, it should be marked up using appropriate elements (typically h1) and then styled using CSS and supporting technologies such as XBL.

When authors use the canvas element, they should also provide content that, when presented to the user, conveys essentially the same function or purpose as the bitmap canvas. This content may be placed as content of the canvas element. The contents of the canvas element, if any, are the element's fallback content.

In interactive visual media with scripting enabled, the canvas element is an embedded element with a dynamically created image.

In non-interactive, static, visual media, if the canvas element has been previously painted on (e.g. if the page was viewed in an interactive visual medium and is now being printed, or if some script that ran during the page layout process painted on the element), then the canvas element must be treated as embedded content with the current image and size. Otherwise, the element's fallback content must be used instead.

In non-visual media, and in visual media with scripting disabled, the canvas element's fallback content must be used instead.

The canvas element has two attributes to control the size of the coordinate space: height and width. These attributes, when specified, must have values that are valid non-negative integers. The rules for parsing non-negative integers must be used to obtain their numeric values. If an attribute is missing, or if parsing its value returns an error, then the default value must be used instead. The width attribute defaults to 300, and the height attribute defaults to 150.

The intrinsic dimensions of the canvas element equal the size of the coordinate space, with the numbers interpreted in CSS pixels. However, the element can be sized arbitrarily by a style sheet. During rendering, the image is scaled to fit this layout size.

The size of the coordinate space does not necessarily represent the size of the actual bitmap that the user agent will use internally or during rendering. On high-definition displays, for instance, the user agent may internally use a bitmap with two device pixels per unit in the coordinate space, so that the rendering remains at high quality throughout.

The canvas must initially be fully transparent black.

If the width and height attributes are dynamically modified, the bitmap and any associated contexts must be cleared back to their initial state and reinitialised with the newly specified coordinate space dimensions.

The width and height DOM attributes must reflect the content attributes of the same name.

To draw on the canvas, authors must first obtain a reference to a context using the getContext method of the canvas element.

This specification only defines one context, with the name "2d". If getContext() is called with that exact string, then the UA must return a reference to an object implementing CanvasRenderingContext2D. Other specifications may define their own contexts, which would return different objects.

Vendors may also define experimental contexts using the syntax vendorname-context, for example, moz-3d.

When the UA is passed an empty string or a string specifying a context that it does not support, then it must return null. String comparisons should be literal and case-sensitive.

A future version of this specification will probably define a 3d context (probably based on the OpenGL ES API).

The toDataURL() method must, when called with no arguments, return a data: URI containing a representation of the image as a PNG file. [PNG].

The toDataURL(type) method (when called with one or more arguments) must return a data: URI containing a representation of the image in the format given by type. The possible values are MIME types with no parameters, for example image/png, image/jpeg, or even maybe image/svg+xml if the implementation actually keeps enough information to reliably render an SVG image from the canvas.

Only support for image/png is required. User agents may support other types. If the user agent does not support the requested type, it must return the image using the PNG format.

User agents must convert the provided type to lower case before establishing if they support that type and before creating the data: URL.

When trying to use types other than image/png, authors can check if the image was really returned in the requested format by checking to see if the returned string starts with one the exact strings "data:image/png," or "data:image/png;". If it does, the image is PNG, and thus the requested type was not supported.

Arguments other than the type must be ignored, and must not cause the user agent to raise an exception (as would normally occur if a method was called with the wrong number of arguments). A future version of this specification will probably allow extra parameters to be passed to toDataURL() to allow authors to more carefully control compression settings, image metadata, etc.

Security: To prevent information leakage, the toDataURL() and getImageData() methods should raise a security exception if the canvas has ever had an image painted on it whose origin is different from that of the script calling the method.

3.14.11.1. The 2D context

When the getContext() method of a canvas element is invoked with 2d as the argument, a CanvasRenderingContext2D object is returned.

There is only one CanvasRenderingContext2D object per canvas, so calling the getContext() method with the 2d argument a second time must return the same object.

The 2D context represents a flat cartesian surface whose origin (0,0) is at the top left corner, with the coordinate space having x values increasing when going right, and y values increasing when going down.

interface CanvasRenderingContext2D {

  // back-reference to the canvas readonly attribute
  HTMLCanvasElement canvas;

  // state
  void save(); // push state on state stack
  void restore(); // pop state stack and restore state

  // transformations (default transform is the identity matrix)
  void scale(in float x, in float y);
  void rotate(in float angle);
  void translate(in float x, in float y);
  void transform(in float m11, in float m12, in float m21, in float m22, in float dx, in float dy);
  void setTransform(in float m11, in float m12, in float m21, in float m22, in float dx, in float dy);

  // compositing
           attribute float globalAlpha; // (default 1.0)
           attribute DOMString globalCompositeOperation; // (default over)

  // colors and styles
           attribute DOMObject strokeStyle; // (default black)
           attribute DOMObject fillStyle; // (default black)
  CanvasGradient createLinearGradient(in float x0, in float y0, in float x1, in float y1);
  CanvasGradient createRadialGradient(in float x0, in float y0, in float r0, in float x1, in float y1, in float r1);
  CanvasPattern createPattern(in HTMLImageElement image, DOMString repetition);
  CanvasPattern createPattern(in HTMLCanvasElement image, DOMString repetition);

  // line caps/joins
           attribute float lineWidth; // (default 1)
           attribute DOMString lineCap; // "butt", "round", "square" (default "butt")
           attribute DOMString lineJoin; // "round", "bevel", "miter" (default "miter")
           attribute float miterLimit; // (default 10)

  // shadows
           attribute float shadowOffsetX; // (default 0)
           attribute float shadowOffsetY; // (default 0)
           attribute float shadowBlur; // (default 0)
           attribute DOMString shadowColor; // (default black)

  // rects
  void clearRect(in float x, in float y, in float w, in float h);
  void fillRect(in float x, in float y, in float w, in float h);
  void strokeRect(in float x, in float y, in float w, in float h);

  // path API
  void beginPath();
  void closePath();
  void moveTo(in float x, in float y);
  void lineTo(in float x, in float y);
  void quadraticCurveTo(in float cpx, in float cpy, in float x, in float y);
  void bezierCurveTo(in float cp1x, in float cp1y, in float cp2x, in float cp2y, in float x, in float y);
  void arcTo(in float x1, in float y1, in float x2, in float y2, in float radius);
  void rect(in float x, in float y, in float w, in float h);
  void arc(in float x, in float y, in float radius, in float startAngle, in float endAngle, in boolean anticlockwise);
  void fill();
  void stroke();
  void clip();
  boolean isPointInPath(in float x, in float y);

  // drawing images
  void drawImage(in HTMLImageElement image, in float dx, in float dy);
  void drawImage(in HTMLImageElement image, in float dx, in float dy, in float dw, in float dh);
  void drawImage(in HTMLImageElement image, in float sx, in float sy, in float sw, in float sh, in float dx, in float dy, in float dw, in float dh);
  void drawImage(in HTMLCanvasElement image, in float dx, in float dy);
  void drawImage(in HTMLCanvasElement image, in float dx, in float dy, in float dw, in float dh);
  void drawImage(in HTMLCanvasElement image, in float sx, in float sy, in float sw, in float sh, in float dx, in float dy, in float dw, in float dh);

  // pixel manipulation
  ImageData getImageData(in float sx, in float sy, in float sw, in float sh);
  void putImageData(in ImageData image, in float dx, in float dy);

  // drawing text is not supported in this version of the API
  // (there is no way to predict what metrics the fonts will have,
  // which makes fonts very hard to use for painting)

};

interface CanvasGradient {
  // opaque object
  void addColorStop(in float offset, in DOMString color);
};

interface CanvasPattern {
  // opaque object
};

interface ImageData {
  readonly attribute long int width;
  readonly attribute long int height;
  readonly attribute int[] data;
};

The canvas attribute must return the canvas element that the context paints on.

3.14.11.1.1. The canvas state

Each context maintains a stack of drawing states. Drawing states consist of:

The current transformation matrix.
The current clip region.
The current values of the following attributes: strokeStyle, fillStyle, globalAlpha, lineWidth, lineCap, lineJoin, miterLimit, shadowOffsetX, shadowOffsetY, shadowBlur, shadowColor, globalCompositeOperation.

The current path and the current bitmap are not part of the drawing state. The current path is persistent, and can only be reset using the beginPath() method. The current bitmap is a property of the canvas, not the context.

The save() method pushes a copy of the current drawing state onto the drawing state stack.

The restore() method pops the top entry in the drawing state stack, and resets the drawing state it describes. If there is no saved state, the method does nothing.

3.14.11.1.2. Transformations

The transformation matrix is applied to all drawing operations prior to their being rendered. It is also applied when creating the clip region.

When the context is created, the transformation matrix must initially be the identity transform. It may then be adjusted using the three transformation methods.

The transformations must be performed in reverse order. For instance, if a scale transformation that doubles the width is applied, followed by a rotation transformation that rotates drawing operations by a quarter turn, and a rectangle twice as wide as it is tall is then drawn on the canvas, the actual result will be a square.

The scale(x, y) method must add the scaling transformation described by the arguments to the transformation matrix. The x argument represents the scale factor in the horizontal direction and the y argument represents the scale factor in the vertical direction. The factors are multiples.

The rotate(angle) method must add the rotation transformation described by the argument to the transformation matrix. The angle argument represents a clockwise rotation angle expressed in radians.

The translate(x, y) method must add the translation transformation described by the arguments to the transformation matrix. The x argument represents the translation distance in the horizontal direction and the y argument represents the translation distance in the vertical direction. The arguments are in coordinate space units.

The transform(m11, m12, m21, m22, dx, dy) method must multiply the current transformation matrix with the matrix described by:

`m11`	`m21`	`dx`
`m12`	`m22`	`dy`
0	0	1

The setTransform(m11, m12, m21, m22, dx, dy) method must reset the current transform to the identity matrix, and then invoke the transform(m11, m12, m21, m22, dx, dy) method with the same arguments.

3.14.11.1.3. Compositing

All drawing operations are affected by the global compositing attributes, globalAlpha and globalCompositeOperation.

The globalAlpha attribute gives an alpha value that is applied to shapes and images before they are composited onto the canvas. The value must be in the range from 0.0 (fully transparent) to 1.0 (no additional transparency). If an attempt is made to set the attribute to a value outside this range, the attribute must retain its previous value. When the context is created, the globalAlpha attribute must initially have the value 1.0.

The globalCompositeOperation attribute sets how shapes and images are drawn onto the existing bitmap, once they have had globalAlpha and the current transformation matrix applied. It must be set to a value from the following list. In the descriptions below, the source image is the shape or image being rendered, and the destination image is the current state of the bitmap.

The source-* descriptions below don't define what should happen with semi-transparent regions.

source-atop: Display the source image wherever both images are opaque. Display the destination image wherever the destination image is opaque but the source image is transparent. Display transparency elsewhere.
source-in: Display the source image wherever both the source image and destination image are opaque. Display transparency elsewhere.
source-out: Display the source image wherever the source image is opaque and the destination image is transparent. Display transparency elsewhere.
source-over (default): Display the source image wherever the source image is opaque. Display the destination image elsewhere.
destination-atop: Same as source-atop but using the destination image instead of the source image and vice versa.
destination-in: Same as source-in but using the destination image instead of the source image and vice versa.
destination-out: Same as source-out but using the destination image instead of the source image and vice versa.
destination-over: Same as source-over but using the destination image instead of the source image and vice versa.
darker: Display the sum of the source image and destination images, with color values approaching 0 as a limit.
lighter: Display the sum of the source image and destination image, with color values approaching 1 as a limit.
copy: Display the source image instead of the destination image.
xor: Exclusive OR of the source and destination images.
vendorName-operationName: Vendor-specific extensions to the list of composition operators should use this syntax.

These values are all case-sensitive — they must be used exactly as shown. User agents must only recognise values that exactly match the values given above.

On setting, if the user agent does not recognise the specified value, it must be ignored, leaving the value of globalCompositeOperation unaffected.

When the context is created, the globalCompositeOperation attribute must initially have the value source-over.

3.14.11.1.4. Colors and styles

The strokeStyle attribute represents the color or style to use for the lines around shapes, and the fillStyle attribute represents the color or style to use inside the shapes.

Both attributes can be either strings, CanvasGradients, or CanvasPatterns. On setting, strings should be parsed as CSS <color> values and the color assigned, and CanvasGradient and CanvasPattern objects must be assigned themselves. [CSS3COLOR] If the value is a string but is not a valid color, or is neither a string, a CanvasGradient, nor a CanvasPattern, then it must be ignored, and the attribute must retain its previous value.

On getting, if the value is a color, then: if it has alpha equal to 1.0, then the color must be returned as an uppercase six-digit hex value, prefixed with a "#" character (U+0023 NUMBER SIGN), with the first two digits representing the red component, the next two digits representing the green component, and the last two digits representing the blue component, the digits being in the range 0-9 A-F (U+0030 to U+0039 and U+0041 to U+0046). If the value has alpha less than 1.0, then the value must instead be returned in the CSS rgba() functional-notation format: the literal string rgba (U+0072 U+0067 U+0062 U+0061) followed by a U+0028 LEFT PARENTHESIS, a base-ten integer in the range 0-255 representing the red component (using digits 0-9, U+0030 to U+0039, in the shortest form possible), a literal U+002C COMMA and U+0020 SPACE, an integer for the green component, a comma and a space, an integer for the blue component, another comma and space, a U+0030 DIGIT ZERO, a U+002E FULL STOP (representing the decimal point), one or more digits in the range 0-9 (U+0030 to U+0039) representing the fractional part of the alpha value, and finally a U+0029 RIGHT PARENTHESIS.

Otherwise, if it is not a color but a CanvasGradient or CanvasPattern, then an object supporting those interfaces must be returned. Such objects are opaque and therefore only useful for assigning to other attributes or for comparison to other gradients or patterns.

When the context is created, the strokeStyle and fillStyle attributes must initially have the string value #000000.

There are two types of gradients, linear gradients and radial gradients, both represented by objects implementing the opaque CanvasGradient interface.

Once a gradient has been created (see below), stops must be placed along it to define how the colors are distributed along the gradient. Between each such stop, the colors and the alpha component must be interpolated over the RGBA space to find the color to use at that offset. Immediately before the 0 offset and immediately after the 1 offset, transparent black stops are be assumed.

The addColorStop(offset, color) method on the CanvasGradient interface adds a new stop to a gradient. If the offset is less than 0 or greater than 1 then an INDEX_SIZE_ERR exception must be raised. If the color cannot be parsed as a CSS color, then a SYNTAX_ERR exception must be raised. Otherwise, the gradient must be updated with the new stop information.

The createLinearGradient(x0, y0, x1, y1) method takes four arguments, representing the start point (x0, y0) and end point (x1, y1) of the gradient, in coordinate space units, and must return a linear CanvasGradient initialised with that line.

Linear gradients must be rendered such that at the starting point on the canvas the color at offset 0 is used, that at the ending point the color at offset 1 is used, that all points on a line perpendicular to the line between the start and end points have the color at the point where those two lines cross (interpolation happening as described above), and that any points beyond the start or end points are a transparent black.

The createRadialGradient(x0, y0, r0, x1, y1, r1) method takes six arguments, the first three representing the start circle with origin (x0, y0) and radius r0, and the last three representing the end circle with origin (x1, y1) and radius r1. The values are in coordinate space units. The method must return a radial CanvasGradient initialised with those two circles.

Radial gradients must be rendered such that a cone is created from the two circles, so that at the circumference of the starting circle the color at offset 0 is used, that at the circumference around the ending circle the color at offset 1 is used, that the circumference of a circle drawn a certain fraction of the way along the line between the two origins with a radius the same fraction of the way between the two radii has the color at that offset (interpolation happening as described above), that the end circle appear to be above the start circle when the end circle is not completely enclosed by the start circle, that the end circle be filled by the color at offset 1, and that any points not described by the gradient are a transparent black.

If a gradient has no stops defined, then the gradient must be treated as a solid transparent black. Gradients are, naturally, only painted where the stroking or filling effect requires that they be drawn.

Support for actually painting gradients is optional. Instead of painting the gradients, user agents may instead just paint the first stop's color. However, createLinearGradient() and createRadialGradient() must always return objects when passed valid arguments.

Patterns are represented by objects implementing the opaque CanvasPattern interface.

To create objects of this type, the createPattern(image, repetition) method is used. The first argument gives the image to use as the pattern (either an HTMLImageElement or an HTMLCanvasElement). Modifying this image after calling the createPattern() method must not affect the pattern. The second argument must be a string with one of the following values: repeat, repeat-x, repeat-y, no-repeat. If the empty string or null is specified, repeat must be assumed. If an unrecognised value is given, then the user agent must raise a SYNTAX_ERR exception. User agents must recognise the four values described above exactly (e.g. they must not do case folding). The method must return a CanvasPattern object suitably initialised.

The image argument must be an instance of an HTMLImageElement or HTMLCanvasElement. If the image is of the wrong type, the implementation must raise a TYPE_MISMATCH_ERR exception.

Patterns must be painted so that the top left of the first image is anchored at the origin of the coordinate space, and images are then repeated horizontally to the left and right (if the repeat-x string was specified) or vertically up and down (if the repeat-y string was specified) or in all four directions all over the canvas (if the repeat string was specified). The images are not be scaled by this process; one CSS pixel of the image must be painted on one coordinate space unit. Of course, patterns must only actually painted where the stroking or filling effect requires that they be drawn, and are affected by the current transformation matrix.

Support for patterns is optional. If the user agent doesn't support patterns, then createPattern() must return null.

3.14.11.1.5. Line styles

The lineWidth attribute gives the default width of lines, in coordinate space units. On setting, zero and negative values must be ignored, leaving the value unchanged.

When the context is created, the lineWidth attribute must initially have the value 1.0.

The lineCap attribute defines the type of endings that UAs shall place on the end of lines. The three valid values are butt, round, and square. The butt value means that the end of each line is a flat edge perpendicular to the direction of the line. The round value means that a semi-circle with the diameter equal to the width of the line is then added on to the end of the line. The square value means that at the end of each line is a rectangle with the length of the line width and the width of half the line width, placed flat against the edge perpendicular to the direction of the line. On setting, any other value than the literal strings butt, round, and square must be ignored, leaving the value unchanged.

When the context is created, the lineCap attribute must initially have the value butt.

The lineJoin attribute defines the type of corners that that UAs will place where two lines meet. The three valid values are round, bevel, and miter.

On setting, any other value than the literal strings round, bevel and miter must be ignored, leaving the value unchanged.

When the context is created, the lineJoin attribute must initially have the value miter.

The round value means that a filled arc connecting the corners on the outside of the join, with the diameter equal to the line width, and the origin at the point where the inside edges of the lines touch, must be rendered at joins. The bevel value means that a filled triangle connecting those two corners with a straight line, the third point of the triangle being the point where the lines touch on the inside of the join, must be rendered at joins. The miter value means that a filled four- or five-sided polygon must be placed at the join, with two of the lines being the perpendicular edges of the joining lines, and the other two being continuations of the outside edges of the two joining lines, as long as required to intersect without going over the miter limit.

The miter length is the distance from the point where the lines touch on the inside of the join to the intersection of the line edges on the outside of the join. The miter limit ratio is the maximum allowed ratio of the miter length to the line width. If the miter limit would be exceeded, then a fifth line must be added to the polygon, connecting the two outside lines, such that the distance from the inside point of the join to the point in the middle of this fifth line is the maximum allowed value for the miter length.

The miter limit ratio can be explicitly set using the miterLimit attribute. On setting, zero and negative values must be ignored, leaving the value unchanged.

When the context is created, the miterLimit attribute must initially have the value 10.0.

3.14.11.1.6. Shadows

All drawing operations are affected by the four global shadow attributes. Shadows form part of the source image during composition.

The shadowColor attribute sets the color of the shadow.

When the context is created, the shadowColor attribute initially must be fully-transparent black.

The shadowOffsetX and shadowOffsetY attributes specify the distance that the shadow will be offset in the positive horizontal and positive vertical distance respectively. Their values are in coordinate space units.

When the context is created, the shadow offset attributes initially have the value 0.

The shadowBlur attribute specifies the number of coordinate space units that the blurring is to cover. On setting, negative numbers must be ignored, leaving the attribute unmodified.

When the context is created, the shadowBlur attribute must initially have the value 0.

Support for shadows is optional. When they are supported, then, when shadows are drawn, they must be rendered using the specified color, offset, and blur radius.

3.14.11.1.7. Simple shapes (rectangles)

There are three methods that immediately draw rectangles to the bitmap. They each take four arguments; the first two give the x and y coordinates of the top left of the rectangle, and the second two give the width and height of the rectangle, respectively.

Shapes are painted without affecting the current path, and are subject to transformations, shadow effects, global alpha, clipping paths, and global composition operators.

Negative values for width and height must cause the implementation to raise an INDEX_SIZE_ERR exception.

The clearRect() method must clear the pixels in the specified rectangle to a fully transparent black, erasing any previous image. If either height or width are zero, this method has no effect.

The fillRect() method must paint the specified rectangular area using the fillStyle. If either height or width are zero, this method has no effect.

The strokeRect() method must draw a rectangular outline of the specified size using the strokeStyle, lineWidth, lineJoin, and (if appropriate) miterLimit attributes. What should happen with zero heights or widths?

3.14.11.1.8. Complex shapes (paths)

The context always has a current path. There is only one current path, it is not part of the drawing state.

A path has a list of zero or more subpaths. Each subpath consists of a list of one or more points, connected by straight or curved lines, and a flag indicating whether the subpath is closed or not. A closed subpath is one where the last point of the subpath is connected to the first point of the subpath by a straight line. Subpaths with fewer than two points are ignored when painting the path.

Initially, the context's path must have zero subpaths.

The beginPath() method must empty the list of subpaths so that the context once again has zero subpaths.

The moveTo(x, y) method must create a new subpath with the specified point as its first (and only) point.

The closePath() method must do nothing if the context has no subpaths. Otherwise, it must mark the last subpath as closed, create a new subpath whose first point is the same as the previous subpath's first point, and finally add this new subpath to the path. (If the last subpath had more than one point in its list of points, then this is equivalent to adding a straight line connecting the last point back to the first point, thus "closing" the shape, and then repeating the last moveTo() call.)

New points and the lines connecting them are added to subpaths using the methods described below. In all cases, the methods only modify the last subpath in the context's paths.

The lineTo(x, y) method must do nothing if the context has no subpaths. Otherwise, it must connect the last point in the subpath to the given point (x, y) using a straight line, and must then add the given point (x, y) to the subpath.

The quadraticCurveTo(cpx, cpy, x, y) method must do nothing if the context has no subpaths. Otherwise it must connect the last point in the subpath to the given point (x, y) by a quadratic curve with control point (cpx, cpy), and must then add the given point (x, y) to the subpath.

The bezierCurveTo(cp1x, cp1y, cp2x, cp2y, x, y) method must do nothing if the context has no subpaths. Otherwise, it must connect the last point in the subpath to the given point (x, y) using a bezier curve with control points (cp1x, cp1y) and (cp2x, cp2y). Then, it must add the point (x, y) to the subpath.

The arcTo(x1, y1, x2, y2, radius) method must do nothing if the context has no subpaths. If the context does have a subpath, then the behaviour depends on the arguments and the last point in the subpath.

Let the point (x0, y0) be the last point in the subpath. Let The Arc be the shortest arc given by circumference of the circle that has one point tangent to the line defined by the points (x0, y0) and (x1, y1), another point tangent to the line defined by the points (x1, y1) and (x2, y2), and that has radius radius. The points at which this circle touches these two lines are called the start and end tangent points respectively.

If the point (x2, y2) is on the line defined by the points (x0, y0) and (x1, y1) then the method must do nothing, as no arc would satisfy the above constraints.

Otherwise, the method must connect the point (x0, y0) to the start tangent point by a straight line, then connect the start tangent point to the end tangent point by The Arc, and finally add the start and end tangent points to the subpath.

Negative or zero values for radius must cause the implementation to raise an INDEX_SIZE_ERR exception.

The arc(x, y, radius, startAngle, endAngle, anticlockwise) method draws an arc. If the context has any subpaths, then the method must add a straight line from the last point in the subpath to the start point of the arc. In any case, it must draw the arc between the start point of the arc and the end point of the arc, and add the start and end points of the arc to the subpath. The arc and its start and end points are defined as follows:

Consider a circle that has its origin at (x, y) and that has radius radius. The points at startAngle and endAngle along the circle's circumference, measured in radians clockwise from the positive x-axis, are the start and end points respectively. The arc is the path along the circumference of this circle from the start point to the end point, going anti-clockwise if the anticlockwise argument is true, and clockwise otherwise.

Negative or zero values for radius must cause the implementation to raise an INDEX_SIZE_ERR exception.

The rect(x, y, w, h) method must create a new subpath containing just the four points (x, y), (x+w, y), (x+w, y+h), (x, y+h), with those four points connected by straight lines, and must then mark the subpath as closed. It must then create a new subpath with the point (x, y) as the only point in the subpath.

Negative values for w and h must cause the implementation to raise an INDEX_SIZE_ERR exception.

The fill() method must fill each subpath of the current path in turn, using fillStyle, and using the non-zero winding number rule. Open subpaths must be implicitly closed when being filled (without affecting the actual subpaths).

The stroke() method must stroke each subpath of the current path in turn, using the strokeStyle, lineWidth, lineJoin, and (if appropriate) miterLimit attributes.

Paths, when filled or stroked, must be painted without affecting the current path, and must be subject to transformations, shadow effects, global alpha, clipping paths, and global composition operators.

The transformation is applied to the path when it is drawn, not when the path is constructed. Thus, a single path can be constructed and then drawn according to different transformations without recreating the path.

The clip() method must create a new clipping path by calculating the intersection of the current clipping path and the area described by the current path (after applying the current transformation), using the non-zero winding number rule. Open subpaths must be implicitly closed when computing the clipping path, without affecting the actual subpaths.

When the context is created, the initial clipping path is the rectangle with the top left corner at (0,0) and the width and height of the coordinate space.

The isPointInPath(x, y) method must return true if the point given by the x and y coordinates passed to the method, when treated as coordinates in the canvas' coordinate space unaffected by the current transformation, is within the area of the canvas that is inside the current path; and must return false otherwise.

3.14.11.1.9. Images

To draw images onto the canvas, the drawImage method can be used.

This method is overloaded with three variants: drawImage(image, dx, dy), drawImage(image, dx, dy, dw, dh), and drawImage(image, sx, sy, sw, sh, dx, dy, dw, dh). (Actually it is overloaded with six; each of those three can take either an HTMLImageElement or an HTMLCanvasElement for the image argument.) If not specified, the dw and dh arguments default to the values of sw and sh, interpreted such that one CSS pixel in the image is treated as one unit in the canvas coordinate space. If the sx, sy, sw, and sh arguments are omitted, they default to 0, 0, the image's intrinsic width in image pixels, and the image's intrinsic height in image pixels, respectively.

The image argument must be an instance of an HTMLImageElement or HTMLCanvasElement. If the image is of the wrong type, the implementation must raise a TYPE_MISMATCH_ERR exception. If one of the sy, sw, sw, and sh arguments is outside the size of the image, or if one of the dw and dh arguments is negative, the implementation must raise an INDEX_SIZE_ERR exception.

When drawImage() is invoked, the specified region of the image specified by the source rectangle (sx, sy, sw, sh) must be painted on the region of the canvas specified by the destination rectangle (dx, dy, dw, dh).

Images are painted without affecting the current path, and are subject to transformations, shadow effects, global alpha, clipping paths, and global composition operators.

3.14.11.1.10. Pixel manipulation

The getImageData(sx, sy, sw, sh) method must return an ImageData object representing the underlying pixel data for the area of the canvas denoted by the rectangle which has one corner at the (sx, sy) coordinate, and that has width sw and height sh. Pixels outside the canvas must be returned as transparent black.

ImageData objects must be initialised so that their height attribute is set to h, the number of rows in the image data, their width attribute is set to w, the number of physical device pixels per row in the image data, and the data attribute is initialised to an array of h×w×4 integers. The pixels must be represented in this array in left-to-right order, row by row, starting at the top left, with each pixel's red, green, blue, and alpha components being given in that order. Each component of each device pixel represented in this array must be in the range 0..255, representing the 8 bit value for that component.

The putImageData(image, dx, dy) method must take the given ImageData structure, and draw it at the specified location dx,dy in the canvas coordinate space, mapping each pixel represented by the ImageData structure into one device pixel.

The handling of pixel rounding when the specified coordinates do not exactly map to the device coordinate space is not defined by this specification, except that the following must result in no visible changes to the rendering:

context.putImageData(context.getImageData(x, y, w, h), x, y);

...for any value of x and y. In other words, while user agents may round the arguments of the two methods so that they map to device pixel boundaries, any rounding performed must be performed consistently for both the getImageData() and putImageData() operations.

The current transformation matrix must not affect the getImageData() and putImageData() methods.

3.14.11.1.11. Drawing model

When a shape or image is painted, user agents must follow these steps, in the order given (or act as if they do):

The coordinates are transformed by the current transformation matrix.
The shape or image is rendered, creating image A, as described in the previous sections. For shapes, the current fill, stroke, and line styles must be honoured.
The shadow is rendered from image A, using the current shadow styles, creating image B.
Image A is composited over image B creating the source image.
The source image has its alpha adjusted by globalAlpha.
Within the clip region (as affected by the current transformation matrix), the source image is composited over the current canvas bitmap using the current composition operator.

3.14.12. The `map` element

Contexts in which this element may be used:

Where block-level elements are expected.

Content model:

Zero or more block-level elements.

Element-specific attributes:

None.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLMapElement : HTMLElement {
  readonly attribute HTMLCollection areas;
};

The map element, in conjuction with any area element descendants, defines an image map.

The areas attribute must return an HTMLCollection rooted at the map element, whose filter matches only area elements.

3.14.13. The `area` element

Interactive, block-level element.

Contexts in which this element may be used:

Where strictly inline-level content is allowed, but only as a descendant of a map element.

Content model:

Empty.

Element-specific attributes:

alt

coords

shape

href

target

ping

rel

media

hreflang

type

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLAreaElement : HTMLElement {
           attribute DOMString alt;
           attribute DOMString coords;
           attribute DOMString shape;
           attribute DOMString href;
           attribute DOMString target;
           attribute DOMString ping;
           attribute DOMString rel;
  readonly attribute DOMTokenList relList;
           attribute DOMString media;
           attribute DOMString hreflang;
           attribute DOMString type;
};

The area element represents either a hyperlink with some text and a corresponding area on an image map, or a dead area on an image map.

If the area element has an href attribute, then the area element represents a hyperlink; the alt attribute, which must then be present, specifies the text.

However, if the area element has no href attribute, then the area represented by the element cannot be selected, and the alt attribute must be omitted.

In both cases, the shape and coords attributes specify the area.

The shape attribute is an enumerated attribute. The following table lists the keywords defined for this attribute. The states given in the first cell of the the rows with keywords give the states to which those keywords map. Some of the keywords are non-conforming, as noted in the last column.

State	Keywords	Notes
Circle state	`circ`	Non-conforming
Circle state	`circle`
Default state	`default`
Polygon state	`poly`
Polygon state	`polygon`	Non-conforming
Rectangle state	`rect`
Rectangle state	`rectangle`	Non-conforming

The attribute may be ommited. The missing value default is the rectangle state.

The coords attribute must, if specified, contain a valid list of integers. This attribute gives the coordinates for the shape described by the shape attribute. The processing for this attribute is described as part of the image map processing model.

In the circle state, area elements must have a coords attribute present, with three integers, the last of which must be non-negative. The first integer must be the distance in CSS pixels from the left edge of the image to the center of the circle, the second integer must be the distance in CSS pixels from the top edge of the image to the center of the circle, and the third integer must be the radius of the circle, again in CSS pixels.

In the default state state, area elements must not have a coords attribute.

In the polygon state, area elements must have a coords attribute with at least six integers, and the number of integers must be even. Each pair of integers must represent a coordinate given as the distances from the left and the top of the image in CSS pixels respectively, and all the coordinates together must represent the points of the polygon, in order.

In the rectangle state, area elements must have a coords attribute with exactly four integers, the first of which must be less than the third, and the second of which must be less than the fourth. The four points must represent, respectively, the distance from the left edge of the image to the top left side of the rectangle, the distance from the top edge to the top side, the distance from the left edge to the right side, and the distance from the top edge to the bottom side, all in CSS pixels.

When user agents allow users to follow hyperlinks created using the area element, as described in the next section, the href, target and ping attributes decide how the link is followed. The rel, media, hreflang, and type attributes may be used to indicate to the user the likely nature of the target resource before the user follows the link.

The target, ping, rel, media, hreflang, and type attributes must be omitted if the href attribute is not present.

The activation behavior of area elements is to run the following steps:

If the DOMActivate event in question is not trusted (i.e. a click() method call was the reason for the event being dispatched), and the area element's target attribute is ... then raise an INVALID_ACCESS_ERR exception.
Otherwise, the user agent must follow the hyperlink defined by the area element, if any.

One way that a user agent can enable users to follow hyperlinks is by allowing area elements to be clicked, or focussed and activated by the keyboard. This will cause the aforementioned activation behavior to be invoked.

The DOM attributes alt, coords, shape, href, target, ping, rel, media, hreflang, and type, each must reflect the respective content attributes of the same name.

The DOM attribute relList must reflect the rel content attribute.

3.14.14. Image maps

An image map allows geometric areas on an image to be associated with hyperlinks.

An image, in the form of an img element or an object element representing an image, may be associated with an image map (in the form of a map element) by specifying a usemap attribute on the img or object element. The usemap attribute, if specified, must be a valid hashed ID reference to a map element.

If an img element or an object element representing an image has a usemap attribute specified, user agents must process it as follows:

First, rules for parsing a hashed ID reference to a map element must be followed. This will return either an element (the map) or null.
If that returned null, then abort these steps. The image is not associated with an image map after all.
Otherwise, the user agent must collect all the area elements that are descendants of the map. Let those be the areas.

Having obtained the list of area elements that form the image map (the areas), interactive user agents must process the list in one of two ways.

If the user agent intends to show the text that the img element represents, then it must use the following steps.

In user agents that do not support images, or that have images disabled, object elements cannot represent images, and thus this section never applies (the fallback content is shown instead). The following steps therefore only apply to img elements.

Remove all the area elements in areas that have no href attribute.
Remove all the area elements in areas that have no alt attribute, or whose alt attribute's value is the empty string, if there is another area element in areas with the same value in the href attribute and with a non-empty alt attribute.
Each remaining area element in areas represents a hyperlink. Those hyperlinks should all be made available to the user in a manner associated with the text of the img element.

In this context, user agents may represent area and img elements with no specified alt attributes, or whose alt attributes are the empty string or some other non-visible text, in a user-agent-defined fashion intended to indicate the lack of suitable author-provided text.

If the user agent intends to show the image and allow interaction with the image to select hyperlinks, then the image must be associated with a set of layered shapes, taken from the area elements in areas, in reverse tree order (so the last specified area element in the map is the bottom-most shape, and the first element in the map, in tree order, is the top-most shape).

Each area element in areas must be processed as follows to obtain a shape to layer onto the image:

Find the state that the element's shape attribute represents.
Use the rules for parsing a list of integers to parse the element's coords attribute, if it is present, and let the result be the coords list. If the attribute is absent, let the coords list be the empty list.

If the number of items in the coords list is less than the minimum number given for the area element's current state, as per the following table, then the shape is empty; abort these steps.

State	Minimum number of items
Circle state	3
Default state	0
Polygon state	6
Rectangle state	4

Check for excess items in the coords list as per the entry in the following list corresponding to the shape attribute's state:

Circle state
Drop any items in the list beyond the third.
Default state
Drop all items in the list.
Polygon state
Drop the last item if there's an odd number of items.
Rectangle state
Drop any items in the list beyond the fourth.
If the shape attribute represents the rectangle state, and the first number in the list is numerically less than the third number in the list, then swap those two numbers around.
If the shape attribute represents the rectangle state, and the second number in the list is numerically less than the fourth number in the list, then swap those two numbers around.
If the shape attribute represents the circle state, and the third number in the list is less than or equal to zero, then the shape is empty; abort these steps.
Now, the shape represented by the element is the one described for the entry in the list below corresponding to the state of the shape attribute:

Circle state

Let x be the first number in coords, y be the second number, and r be the third number.

The shape is a circle whose center is x CSS pixels from the left edge of the image and x CSS pixels from the top edge of the image, and whose radius is r pixels.

Default state

The shape is a rectangle that exactly covers the entire image.

Polygon state

Let x_i be the (2i)th entry in coords, and y_i be the (2i+1)th entry in coords (the first entry in coords being the one with index 0).

Let the coordinates be (x_i, y_i), interpreted in CSS pixels measured from the top left of the image, for all integer values of i from 0 to (N/2)-1, where N is the number of items in coords.

The shape is a polygon whose vertices are given by the coordinates, and whose interior is established using the even-odd rule. [GRAPHICS]

Rectangle state

Let x1 be the first number in coords, y1 be the second number, x2 be the third number, and y2 be the fourth number.

The shape is a rectangle whose top-left corner is given by the coordinate (x1, y1) and whose bottom right corner is given by the coordinate (x2, y2), those coordinates being interpreted as CSS pixels from the top left corner of the image.

Mouse clicks on an image associated with a set of layered shapes per the above algorithm must be dispatched to the top-most shape covering the point that the pointing device indicated (if any), and then, must be dispatched again (with a new Event object) to the image element itself. User agents may also allow individual area elements representing hyperlinks to be selected and activated (e.g. using a keyboard); events from this are not also propagated to the image.

Because a map element (and its area elements) can be associated with multiple img elements, it is possible for an area element to correspond to multiple focusable areas of the document.

Image maps are live; if the DOM is mutated, then the user agent must act as if it had rerun the algorithms for image maps.

3.15. Tabular data

3.15.1. The `table` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:

Where block-level elements are expected.

Where structured inline-level elements are allowed.

Content model:

In this order: optionally a caption element, followed by either zero or more colgroup elements, followed optionally by a thead element, followed optionally by a tfoot element, followed by either zero or more tbody elements or one or more tr elements, followed optionally by a tfoot element (but there can only be one tfoot element child in total).

Element-specific attributes:

None.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableElement : HTMLElement {
           attribute HTMLTableCaptionElement caption;
  HTMLElement createCaption();
  void deleteCaption();
           attribute HTMLTableSectionElement tHead;
  HTMLElement createTHead();
  void deleteTHead();
           attribute HTMLTableSectionElement tFoot;
  HTMLElement createTFoot();
  void deleteTFoot();
  readonly attribute HTMLCollection tBodies;
  readonly attribute HTMLCollection rows;
  HTMLElement insertRow(in long index);
  void deleteRow(in long index);
};

The table element represents data with more than one dimension (a table).

The children of a table element must be, in order:

Zero or one caption elements.
Zero or more colgroup elements.
Zero or one thead elements.
Zero or one tfoot elements, if the last element in the table is not a tfoot element.
Either:
- Zero or more tbody elements, or
- One or more tr elements.
Zero or one tfoot element, if there are no other tfoot elements in the table.

The table element takes part in the table model.

The caption DOM attribute must return, on getting, the first caption element child of the table element. On setting, if the new value is a caption element, the first caption element child of the table element, if any, must be removed, and the new value must be inserted as the first node of the table element. If the new value is not a caption element, then a HIERARCHY_REQUEST_ERR DOM exception must be raised instead.

The createCaption() method must return the first caption element child of the table element, if any; otherwise a new caption element must be created, inserted as the first node of the table element, and then returned.

The deleteCaption() method must remove the first caption element child of the table element, if any.

The tHead DOM attribute must return, on getting, the first thead element child of the table element. On setting, if the new value is a thead element, the first thead element child of the table element, if any, must be removed, and the new value must be inserted immediately before the first element in the table element that is neither a caption element nor a colgroup element, if any, or at the end of the table otherwise. If the new value is not a thead element, then a HIERARCHY_REQUEST_ERR DOM exception must be raised instead.

The createTHead() method must return the first thead element child of the table element, if any; otherwise a new thead element must be created and inserted immediately before the first element in the table element that is neither a caption element nor a colgroup element, if any, or at the end of the table otherwise, and then that new element must be returned.

The deleteTHead() method must remove the first thead element child of the table element, if any.

The tFoot DOM attribute must return, on getting, the first tfoot element child of the table element. On setting, if the new value is a tfoot element, the first tfoot element child of the table element, if any, must be removed, and the new value must be inserted immediately before the first element in the table element that is neither a caption element, a colgroup element, nor a thead element, if any, or at the end of the table if there are no such elements. If the new value is not a tfoot element, then a HIERARCHY_REQUEST_ERR DOM exception must be raised instead.

The createTFoot() method must return the first tfoot element child of the table element, if any; otherwise a new tfoot element must be created and inserted immediately before the first element in the table element that is neither a caption element, a colgroup element, nor a thead element, if any, or at the end of the table if there are no such elements, and then that new element must be returned.

The deleteTFoot() method must remove the first tfoot element child of the table element, if any.

The tBodies attribute must return an HTMLCollection rooted at the table node, whose filter matches only tbody elements that are children of the table element.

The rows attribute must return an HTMLCollection rooted at the table node, whose filter matches only tr elements that are either children of the table element, or children of thead, tbody, or tfoot elements that are themselves children of the table element. The elements in the collection must be ordered such that those elements whose parent is a thead are included first, in tree order, followed by those elements whose parent is either a table or tbody element, again in tree order, followed finally by those elements whose parent is a tfoot element, still in tree order.

The behaviour of the insertRow(index) method depends on the state of the table. When it is called, the method must act as required by the first item in the following list of conditions that describes the state of the table and the index argument:

If index is less than -1 or greater than the number of elements in rows collection:: The method must raise an INDEX_SIZE_ERR exception.
If the rows collection has zero elements in it, and the table has no tbody elements in it:: The method must create a tbody element, then create a tr element, then append the tr element to the tbody element, then append the tbody element to the table element, and finally return the tr element.
If the rows collection has zero elements in it:: The method must create a tr element, append it to the last tbody element in the table, and return the tr element.
If index is equal to -1 or equal to the number of items in rows collection:: The method must create a tr element, and append it to the parent of the last tr element in the rows collection. Then, the newly created tr element must be returned.
Otherwise:: The method must create a tr element, insert it immediately before the indexth tr element in the rows collection, in the same parent, and finally must return the newly created tr element.

The deleteRow(index) method must remove the indexth element in the rows collection from its parent. If index is less than zero or greater than or equal to the number of elements in the rows collection, the method must instead raise an INDEX_SIZE_ERR exception.

3.15.2. The `caption` element

Contexts in which this element may be used:: As the first element child of a table element.
Content model:: Significant strictly inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The caption element represents the title of the table that is its parent, if it has a parent and that is a table element.

The caption element takes part in the table model.

3.15.3. The `colgroup` element

Contexts in which this element may be used:

As a child of a table element, after any caption elements and before any thead, tbody, tfoot, and tr elements.

Content model:

Zero or more col elements.

Element-specific attributes:

span, but only if the element contains no col elements

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableColElement : HTMLElement {
           attribute unsigned long span;
};

The colgroup element represents a group of one or more columns in the table that is its parent, if it has a parent and that is a table element.

If the colgroup element contains no col elements, then the element may have a span content attribute specified, whose value must be a valid non-negative integer greater than zero. Its default value, which must be used if parsing the attribute as a non-negative integer returns either an error or zero, is 1.

The colgroup element and its span attribute take part in the table model.

The span DOM attribute must reflect the content attribute of the same name, with the exception that on setting, if the new value is 0, then an INDEX_SIZE_ERR exception must be raised.

3.15.4. The `col` element

Contexts in which this element may be used:: As a child of a colgroup element that doesn't have a span attribute.
Content model:: Empty.
Element-specific attributes:: span
Predefined classes that apply to this element:: None.
DOM interface:: HTMLTableColElement, same as for colgroup elements. This interface defines one member, span.

If a col element has a parent and that is a colgroup element that itself has a parent that is a table element, then the col element represents one or more columns in the column group represented by that colgroup.

The element may have a span content attribute specified, whose value must be a valid non-negative integer greater than zero. Its default value, which must be used if parsing the attribute as a non-negative integer returns either an error or zero, is 1.

The col element and its span attribute take part in the table model.

The span DOM attribute must reflect the content attribute of the same name, with the exception that on setting, if the new value is 0, then an INDEX_SIZE_ERR exception must be raised.

3.15.5. The `tbody` element

Contexts in which this element may be used:

As a child of a table element, after any caption, colgroup, and thead elements, but only if there are no tr elements that are children of the table element.

Content model:

One or more tr elements

Element-specific attributes:

None.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableSectionElement : HTMLElement {
  readonly attribute HTMLCollection rows;
  HTMLElement insertRow(in long index);
  void deleteRow(in long index);
};

The HTMLTableSectionElement interface is also used for thead and tfoot elements.

The tbody element represents a block of rows that consist of a body of data for the parent table element, if the tbody element has a parent and it is a table.

The tbody element takes part in the table model.

The rows attribute must return an HTMLCollection rooted at the element, whose filter matches only tr elements that are children of the element.

The insertRow(index) method must, when invoked on an element table section, act as follows:

If index is less than -1 or greater than the number of elements in the rows collection, the method must raise an INDEX_SIZE_ERR exception.

If index is equal to -1 or equal to the number of items in the rows collection, the method must create a tr element, append it to the element table section, and return the newly created tr element.

Otherwise, the method must create a tr element, insert it as a child of the table section element, immediately before the indexth tr element in the rows collection, and finally must return the newly created tr element.

3.15.6. The `thead` element

Contexts in which this element may be used:: As a child of a table element, after any caption, and colgroup elements and before any tbody, tfoot, and tr elements, but only if there are no other thead elements that are children of the table element.
Content model:: One or more tr elements
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: HTMLTableSectionElement, as defined for tbody elements.

The thead element represents the block of rows that consist of the column labels (headers) for the parent table element, if the thead element has a parent and it is a table.

The thead element takes part in the table model.

3.15.7. The `tfoot` element

Contexts in which this element may be used:: As a child of a table element, after any caption, colgroup, and thead elements and before any tbody and tr elements, but only if there are no other tfoot elements that are children of the table element.; As a child of a table element, after any caption, colgroup, thead, tbody, and tr elements, but only if there are no other tfoot elements that are children of the table element.
Content model:: One or more tr elements
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: HTMLTableSectionElement, as defined for tbody elements.

The tfoot element represents the block of rows that consist of the column summaries (footers) for the parent table element, if the tfoot element has a parent and it is a table.

The tfoot element takes part in the table model.

3.15.8. The `tr` element

Contexts in which this element may be used:

As a child of a thead element.

As a child of a tbody element.

As a child of a tfoot element.

As a child of a table element, after any caption, colgroup, and thead elements, but only if there are no tbody elements that are children of the table element.

Content model:

One or more td or th elements

Element-specific attributes:

None.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableRowElement : HTMLElement {
  readonly attribute long rowIndex;
  readonly attribute long sectionRowIndex;
  readonly attribute HTMLCollection cells;
  HTMLElement insertCell(in long index);
  void deleteCell(in long index);
};

The tr element represents a row of cells in a table.

The tr element takes part in the table model.

The rowIndex element must, if the element has a parent table element, or a parent tbody, thead, or tfoot element and a grandparent table element, return the index of the tr element in that table element's rows collection. If there is no such table element, then the attribute must return 0.

The rowIndex DOM attribute must, if the element has a parent table, tbody, thead, or tfoot element, return the index of the tr element in the parent element's rows collection (for tables, that's the rows collection; for table sections, that's the rows collection). If there is no such parent element, then the attribute must return 0.

The cells attribute must return an HTMLCollection rooted at the tr element, whose filter matches only td and th elements that are children of the tr element.

The insertCell(index) method must act as follows:

If index is less than -1 or greater than the number of elements in the cells collection, the method must raise an INDEX_SIZE_ERR exception.

If index is equal to -1 or equal to the number of items in cells collection, the method must create a td element, append it to the tr element, and return the newly created td element.

Otherwise, the method must create a td element, insert it as a child of the tr element, immediately before the indexth td or th element in the cells collection, and finally must return the newly created td element.

The deleteCell(index) method must remove the indexth element in the cells collection from its parent. If index is less than zero or greater than or equal to the number of elements in the cells collection, the method must instead raise an INDEX_SIZE_ERR exception.

3.15.9. The `td` element

Contexts in which this element may be used:

As a child of a tr element.

Content model:

Zero or more block-level elements, or inline-level content (but not both).

Element-specific attributes:

colspan

rowspan

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableCellElement : HTMLElement {
           attribute long colSpan;
           attribute long rowSpan;
  readonly attribute long cellIndex;
};

The td element represents a data cell in a table.

The td element may have a colspan content attribute specified, whose value must be a valid non-negative integer greater than zero. Its default value, which must be used if parsing the attribute as a non-negative integer returns either an error or zero, is 1.

The td element may also have a rowspan content attribute specified, whose value must be a valid non-negative integer. Its default value, which must be used if parsing the attribute as a non-negative integer returns an error, is also 1.

The td element and its colspan and rowspan attributes take part in the table model.

The colspan DOM attribute must reflect the content attribute of the same name, with the exception that on setting, if the new value is 0, then an INDEX_SIZE_ERR exception must be raised.

The rowspan DOM attribute must reflect the content attribute of the same name.

The cellIndex DOM attribute must, if the element has a parent tr element, return the index of the cell's element in the parent element's cells collection. If there is no such parent element, then the attribute must return 0.

3.15.10. The `th` element

Contexts in which this element may be used:

As a child of a tr element.

Content model:

Zero or more block-level elements, or inline-level content (but not both).

Element-specific attributes:

colspan

rowspan

scope

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLTableHeaderCellElement : HTMLTableCellElement {
           attribute DOMString scope;
};

The th element represents a header cell in a table.

The th element may have a colspan content attribute specified, whose value must be a valid non-negative integer greater than zero. Its default value, which must be used if parsing the attribute as a non-negative integer returns either an error or zero, is 1.

The th element may also have a rowspan content attribute specified, whose value must be a valid non-negative integer. Its default value, which must be used if parsing the attribute as a non-negative integer returns an error, is also 1.

The th element may have a scope content attribute specified. The scope attribute is an enumerated attribute with five states, four of which have explicit keywords:

The row keyword, which maps to the row state: The row state means the header cell applies to all the remaining cells in the row.
The col keyword, which maps to the column state: The column state means the header cell applies to all the remaining cells in the column.
The rowgroup keyword, which maps to the row group state: The row group state means the header cell applies to all the remaining cells in the row group.
The colgroup keyword, which maps to the column group state: The column group state means the header cell applies to all the remaining cells in the column group.
The auto state: The auto state makes the header cell apply to a set of cells selected based on context.

The scope attribute's missing value default is the auto state.

The exact effect of these values is described in detail in the algorithm for assigning header cells to data cells, which user agents must apply to determine the relationships between data cells and header cells.

The th element and its colspan, rowspan, and scope attributes take part in the table model.

The scope DOM attribute must reflect the content attribute of the same name.

The HTMLTableHeaderCellElement interface inherits from the HTMLTableCellElement interface and therefore also has the DOM attributes defined above in the td section.

3.15.11. Processing model

The various table elements and their content attributes together define the table model.

A table consists of cells aligned on a two-dimensional grid of slots with coordinates (x, y). The grid is finite, and is either empty or has one or more slots. If the grid has one or more slots, then the x coordinates are always in the range 1 ≤ x ≤ x_max, and the y coordinates are always in the range 1 ≤ y ≤ y_max. If one or both of x_max and y_max are zero, then the table is empty (has no slots). Tables correspond to table elements.

A cell is a set of slots anchored at a slot (cell_x, cell_y), and with a particular width and height such that the cell covers all the slots with coordinates (x, y) where cell_x ≤ x < cell_x+width and cell_y ≤ y < cell_y+height. Cell can either be data cells or header cells. Data cells correspond to td elements, and have zero or more associated header cells. Header cells correspond to th elements.

A row is a complete set of slots from x=1 to x=x_max, for a particular value of y. Rows correspond to tr elements.

A column is a complete set of slots from y=1 to y=y_max, for a particular value of x. Columns can correspond to col elements, but in the absense of col elements are implied.

A row group is a set of rows anchored at a slot (1, group_y) with a particular height such that the row group covers all the slots with coordinates (x, y) where 1 ≤ x < x_max and group_y ≤ y < group_y+height. Row groups correspond to tbody, thead, and tfoot elements. Not every row is necessarily in a row group.

A column group is a set of columns anchored at a slot (group_x, 1) with a particular width such that the column group covers all the slots with coordinates (x, y) where group_x ≤ x < group_x+width and 1 ≤ y < y_max. Column groups correspond to colgroup elements. Not every column is necessarily in a column group.

Row groups cannot overlap each other. Similarly, column groups cannot overlap each other.

A cell cannot cover slots that are from two or more row groups. It is, however, possible for a cell to be in multiple column groups. All the slots that form part of one cell are part of zero or one row groups and zero or more column groups.

In addition to cells, columns, rows, row groups, and column groups, tables can have a caption element associated with them. This gives the table a heading, or legend.

A table model error is an error with the data represented by table elements and their descendants. Documents must not have table model errors.

3.15.11.1. Forming a table

To determine which elements correspond to which slots in a table associated with a table element, to determine the dimensions of the table (x_max and y_max), and to determine if there are any table model errors, user agents must use the following algorithm:

Let x_max be zero.
Let y_max be zero.
Let the table be the table represented by the table element. The x_max and y_max variables give the table's extent. The table is initially empty.
If the table element has no table children, then return the table (which will be empty), and abort these steps.
Let the current element be the first element child of the table element.

If a step in this algorithm ever requires the current element to be advanced to the next child of the table when there is no such next child, then the algorithm must be aborted at that point and the algorithm must return the table.
While the current element is not one of the following elements, advance the current element to the next child of the table:
- caption
- colgroup
- thead
- tbody
- tfoot
- tr
If the current element is a caption, then that is the caption element associated with the table. Otherwise, it has no associated caption element.
If the current element is a caption, then while the current element is not one of the following elements, advance the current element to the next child of the table:
- colgroup
- thead
- tbody
- tfoot
- tr
(Otherwise, the current element will already be one of those elements.)
If the current element is a colgroup, follow these substeps:
1. Let next column be 1.
2. Column groups. Process the current element according to the appropriate one of the following two cases:
  If the current element has any col element children
  Follow these steps:
  1. Let x_start have the value x_max+1.
  2. Let the current column be the first col element child of the colgroup element.
  3. Columns. If the current column col element has a span attribute, then parse its value using the rules for parsing non-negative integers.
    
    If the result of parsing the value is not an error or zero, then let span be that value.
    
    Otherwise, if the col element has no span attribute, or if trying to parse the attribute's value resulted in an error, then let span be 1.
  4. Increase x_max by span.
  5. Let the last span columns in the table correspond to the current column col element.
  6. If current column is not the last col element child of the colgroup element, then let the current column be the next col element child of the colgroup element, and return to the third step of this innermost group of steps (columns).
  7. Let all the last columns in the table from x=x_start to x=x_max form a new column group, anchored at the slot (x_start, 1), with width x_max-x_start-1, corresponding to the colgroup element.
  If the current element has no col element children
  1. If the colgroup element has a span attribute, then parse its value using the rules for parsing non-negative integers.
    
    If the result of parsing the value is not an error or zero, then let span be that value.
    
    Otherwise, if the colgroup element has no span attribute, or if trying to parse the attribute's value resulted in an error, then let span be 1.
  2. Increase x_max by span.
  3. Let the last span columns in the table form a new column group, anchored at the slot (x_max-span+1, 1), with width span, corresponding to the colgroup element.
3. Advance the current element to the next child of the table.
4. While the current element is not one of the following elements, advance the current element to the next child of the table:
  - colgroup
  - thead
  - tbody
  - tfoot
  - tr
5. If the current element is a colgroup element, jump to step 2 in these substeps (column groups).
Let y_current be zero. When the algorithm is aborted, if y_current does not equal y_max, then that is a table model error.
Let the list of downward-growing cells be an empty list.
Rows. While the current element is not one of the following elements, advance the current element to the next child of the table:
- thead
- tbody
- tfoot
- tr
If the current element is a tr, then run the algorithm for processing rows (defined below), then return to the previous step (rows).
Otherwise, run the algorithm for ending a row group.
Let y_start have the value y_max+1.
For each tr element that is a child of the current element, in tree order, run the algorithm for processing rows (defined below).
If y_max ≥ y_start, then let all the last rows in the table from y=y_start to y=y_max form a new row group, anchored at the slot with coordinate (1, y_start), with height y_max-y_start+1, corresponding to the current element.
Run the algorithm for ending a row group again.
Return to step 12 (rows).

The algorithm for ending a row group, which is invoked by the set of steps above when starting and eding a block of rows, is:

If y_current is less than y_max, then this is a table model error.
While y_current is less than y_max, follow these steps:
1. Increase y_current by 1.
2. Run the algorithm for growing downward-growing cells.
Empty the list of downward-growing cells.

The algorithm for processing rows, which is invoked by the set of steps above for processing tr elements, is:

Increase y_current by 1.
Run the algorithm for growing downward-growing cells.
Let x_current be 1.
If the tr element being processed contains no td or th elements, then abort this set of steps and return to the algorithm above.
Let current cell be the first td or th element in the tr element being processed.
Cells. While x_current is less than or equal to x_max and the slot with coordinate (x_current, y_current) already has a cell assigned to it, increase x_current by 1.
If x_current is greater than x_max, increase x_max by 1 (which will make them equal).
If the current cell has a colspan attribute, then parse that attribute's value, and let colspan be the result.

If parsing that value failed, or returned zero, or if the attribute is absent, then let colspan be 1, instead.
If the current cell has a rowspan attribute, then parse that attribute's value, and let rowspan be the result.

If parsing that value failed or if the attribute is absent, then let rowspan be 1, instead.
If rowspan is zero, then let cell grows downward be true, and set rowspan to 1. Otherwise, let cell grows downward be false.
If x_max < x_current+colspan-1, then let x_max be x_current+colspan-1.
If y_max < y_current+rowspan-1, then let y_max be y_current+rowspan-1.
Let the slots with coordinates (x, y) such that x_current ≤ x < x_current+colspan and y_current ≤ y < y_current+rowspan be covered by a new cell c, anchored at (x_current, y_current), which has width colspan and height rowspan, corresponding to the current cell element.

If the current cell element is a th element, let this new cell c be a header cell; otherwise, let it be a data cell. To establish what header cells apply to a data cell, use the algorithm for assigning header cells to data cells described in the next section.

If any of the slots involved already had a cell covering them, then this is a table model error. Those slots now have two cells overlapping.
If cell grows downward is true, then add the tuple {c, x_current, colspan} to the list of downward-growing cells.
Increase x_current by colspan.
If current cell is the last td or th element in the tr element being processed, then abort this set of steps and return to the algorithm above.
Let current cell be the next td or th element in the tr element being processed.
Return to step 5 (cells).

The algorithm for growing downward-growing cells, used when adding a new row, is as follows:

If the list of downward-growing cells is empty, do nothing. Abort these steps; return to the step that invoked this algorithm.
Otherwise, if y_max is less than y_current, then increase y_max by 1 (this will make it equal to y_current).
For each {cell, cell_x, width} tuple in the list of downward-growing cells, extend the cell cell so that it also covers the slots with coordinates (x, y_current), where cell_x ≤ x < cell_x+width-1.

If, after establishing which elements correspond to which slots, there exists a column in the table containing only slots that do not have a cell anchored to them, then this is a table model error.

3.15.11.2. Forming relationships between data cells and header cells

Each data cell can be assigned zero or more header cells. The algorithm for assigning header cells to data cells is as follows.

For each header cell in the table, in tree order:

Let (header_x, header_y) be the coordinate of the slot to which the header cell is anchored.
Examine the scope attribute of the th element corresponding to the header cell, and, based on its state, apply the appropriate substep:

If it is in the row state

Assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where header_x < data_x ≤ x_max and data_y = header_y.

If it is in the column state

Assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where data_x = header_x and header_y < data_y ≤ y_max.

If it is in the row group state

If the header cell is not in a row group, then don't assign the header cell to any data cells.

Otherwise, let (1, group_y) be the slot at which the row group is anchored, let height be the number of rows in the row group, and assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where header_x ≤ data_x ≤ x_max and header_y ≤ data_y < group_y+height.

If it is in the column group state

If the header cell is not in a column group, then don't assign the header cell to any data cells.

Otherwise, let (group_x, 1) be the slot at which the column group is anchored, let width be the number of columns in the column group, and assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where header_x ≤ data_x < group_x+width and header_y ≤ data_y ≤ y_max.

Otherwise, it is in the auto state

If the header cell is not in the first row of the table, or not in the first cell of a row, then don't assign the header cell to any data cells.

Otherwise, if the header cell is in the first row of the table, assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where data_x = header_x and header_y < data_y ≤ y_max.

Otherwise, the header cell is in the first column of the table; assign the header cell to any data cells anchored at slots with coordinates (data_x, data_y) where header_x < data_x ≤ x_max and data_y = header_y.

3.16. Forms

This section will contain definitions of the form element and so forth.

This section will be a rewrite of the HTML4 Forms and Web Forms 2.0 specifications, with hopefully no normative changes.

3.16.1. The `form` element

3.16.2. The `fieldset` element

3.16.3. The `input` element

3.16.4. The `button` element

3.16.5. The `label` element

3.16.6. The `select` element

3.16.7. The `datalist` element

3.16.8. The `optgroup` element

3.16.9. The `option` element

3.16.10. The `textarea` element

3.16.11. The `output` element

3.16.12. Processing model

See WF2 for now

3.16.12.1. Form submission

See WF2 for now

3.17. Scripting

3.17.1. The `script` element

Block-level element, strictly inline-level content, and metadata element.

Contexts in which this element may be used:

In a head element.

Where block-level elements are expected.

Where inline-level content is expected.

Content model:

If there is no src attribute, depends on the value of the type attribute.

If there is a src attribute, the element must be empty.

Element-specific attributes:

src

defer (if the src attribute is present)

async (if the src attribute is present)

type

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLScriptElement : HTMLElement {
           attribute DOMString src;
           attribute boolean defer;
           attribute boolean async;
           attribute DOMString type;
           attribute DOMString text;
};

The script element allows authors to include dynamic script in their documents.

When the src attribute is set, the script element refers to an external file. The value of the attribute must be a URI (or IRI).

If the src attribute is not set, then the script is given by the contents of the element.

The language of the script is given by the type attribute. The value must be a valid MIME type, optionally with parameters. [RFC2046]

The defer and async attributes are boolean attributes that indicate how the script should be executed.

There are three possible modes that can be selected using these attributes. If the defer attribute is present, then the script is executed when the page has finished parsing. If the defer attribute is not present but the async attribute is present, then the script will be executed asynchronously, as soon as it is available. If neither attribute is present, then the script is downloaded and executed immediately, before the user agent continues parsing the page. The exact processing details for these attributes is described below.

Changing the src, type, defer and async attributes dynamically has no direct effect; these attribute are only used at specific times described below (namely, when the element is inserted into the document).

script elements have two associated pieces of metadata. The first is a flag indicating whether or not the script block has been "already executed". Initially, script elements must have this flag unset (script blocks, when created, are not "already executed"). When a script element is cloned, the "already executed" flag, if set, must be propagated to the clone when it is created. The second is a flag indicating whether the element was "parser-inserted". This flag is set by the HTML parser and is used to handle document.write() calls.

Running a script: when a script block is inserted into a document, the user agent must act as follows:

How to handle the type and language attributes should be defined here, probably with reference to the next section.
If scripting is disabled, or if the user agent does not support the scripting language in question, or if the Document has designMode enabled, or if the script element has its "already executed" flag set, then the user agent must abort these steps at this point. The script is not executed.
The user agent must set the element's "already executed" flag.
If the element has a src attribute, then a load for the specified content must be started.

Later, once the load has completed, the user agent will have to complete the steps described below.

For performance reasons, user agents may start loading the script as soon as the attribute is set, instead, in the hope that the element will be inserted into the document. Either way, once the element is inserted into the document, the load must have started. If the UA performs such prefetching, but the element is never inserted in the document, or the src attribute is dynamically changed, then the user agent will not execute the script, and the load will have been effectively wasted.
Then, the first of the following options that describes the situation must be followed:

If the document is still being parsed, and the element has a defer attribute
The element must be added to the end of the list of scripts that will execute when the document has finished parsing. The user agent must begin the next set of steps when the script is ready. This isn't compatible with IE for inline deferred scripts, but then what IE does is pretty hard to pin down exactly. Do we want to keep this like it is? Be more compatible?
If the element has an async attribute and a src attribute
The element must be added to the end of the list of scripts that will execute asynchronously. The user agent must jump to the next set of steps once the script is ready.
If the element has an async attribute but no src attribute, and the list of scripts that will execute asynchronously is not empty
The element must be added to the end of the list of scripts that will execute asynchronously.
If the element has a src attribute and has been flagged as "parser-inserted"
The element is the script that will execute as soon as the parser resumes. (There can only be one such script at a time.)
If the element has a src attribute
The element must be added to the end of the list of scripts that will execute as soon as possible. The user agent must jump to the next set of steps when the script is ready.
Otherwise
The user agent must immediately execute the script, even if other scripts are already executing.

When a script completes loading: If a script whose element was added to one of the lists mentioned above completes loading while the document is still being parsed, then the parser handles it. Otherwise, when a script completes loading, the UA must follow the following steps as soon as as any other scripts that may be executing have finished executing:

If the script's element was added to the list of scripts that will execute when the document has finished parsing:

If the script's element is not the first element in the list, then do nothing yet. Stop going through these steps.
Otherwise, execute the script (that is, the script associated with the first element in the list).
Remove the script's element from the list (i.e. shift out the first entry in the list).
If there are any more entries in the list, and if the script associated with the element that is now the first in the list is already loaded, then jump back to step two to execute it.

If the script's element was added to the list of scripts that will execute asynchronously:

If the script is not the first element in the list, then do nothing yet. Stop going through these steps.
Execute the script (the script associated with the first element in the list).
Remove the script's element from the list (i.e. shift out the first entry in the list).
If there are any more scripts in the list, and the element now at the head of the list had no src attribute when it was added to the list, or had one, but its associated script has finished loading, then jump back to step two to execute the script associated with this element.

If the script's element was added to the list of scripts that will execute as soon as possible:

Execute the script.
Remove the script's element from the list.

If the script is the script that will execute as soon as the parser resumes:

The script will be handled when the parser resumes (amazingly enough).

The download of an external script must delay the load event.

Executing a script block: If the load resulted in an error (for example a DNS error, or an HTTP 404 error), then executing the script must just consist of firing an error event at the element.

If the load was successful, then first the user agent must fire a load event at the element, and then, if scripting is enabled and the Document does not have designMode enabled, the user agent must run the script according to the semantics of the relevant scripting language defines.

If the script is from an external file, then that file must be used as the file to execute.

If the script is inline, then, for scripting languages that consist of pure text, user agents must use the value of the DOM text attribute (defined below) as the script to execute, and for XML-based scripting languages, user agents must use all the child nodes of the script element as the script to execute.

In any case, the user agent must execute the script according to the semantics of the relevant language specification, as determined by the type and language attributes on the script element when the element was inserted into the document, as described above.

Scripts must execute in the scope of the browsing context of the element's Document.

The element's attributes' values might have changed between when the element was inserted into the document and when the script has finished loading, as may its other attributes; similarly, the element itself might have been taken back out of the DOM, or had other changes made. These changes do not in any way affect the above steps; only the values of the attributes at the time the script element is first inserted into the document matter.

The DOM attributes src, type, defer, async, each must reflect the respective content attributes of the same name.

The DOM attribute text must return a concatenation of the contents of all the text nodes that are direct children of the script element (ignoring any other nodes such as comments or elements), in tree order. On setting, it must act the same way as the textContent DOM attribute.

3.17.1.1. Script languages

The following lists some MIME types and the languages to which they refer:

text/javascript: ECMAScript. [ECMA262]
text/javascript;e4x=1: ECMAScript with ECMAScript for XML. [ECMA357]

User agents may support other MIME types and other languages.

When examining types to determine if they support the language, user agents must not ignore unknown MIME parameters — types with unknown parameters must be assumed to be unsupported.

3.17.2. The `noscript` element

When scripting is disabled: transparent block-level element, and transparent strictly inline-level content.

When scripting is enabled: block-level element, and strictly inline-level content.

Contexts in which this element may be used:: In a head element of an HTML document, if there are no ancestor noscript elements.; Where block-level elements are expected in HTML documents, if there are no ancestor noscript elements.; Where inline-level content is expected in HTML documents, if there are no ancestor noscript elements.
Content model:: When scripting is disabled: transparent, but there must be no noscript element descendants.; When scripting is enabled: Text that conforms to the requirements given in the prose.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The noscript element does not represent anything. It is used to present different markup to user agents that support scripting and those that don't support scripting, by affecting how the document is parsed.

The noscript element must not be used in XML documents.

When used in HTML documents, the allowed content model depends on whether scripting is enabled or not.

If scripting is disabled, then the content model of a noscript element is transparent, with the additional restriction that a noscript element must not have a noscript element as an ancestor (that is, noscript can't be nested).

If scripting is enabled, then the content model of a noscript element is text, except that the text must be such that running the following algorithm results in a conforming document with no noscript elements and no script elements, and such that no step in the algorithm causes an HTML parser to flag a parse error:

Remove every script element from the document.
Make a list of every noscript element in the document. For every noscript element in that list, perform the following steps:
1. Let the parent element be the parent element of the noscript element.
2. Take all the children of the parent element that come before the noscript element, and call these elements the before children.
3. Take all the children of the parent element that come after the noscript element, and call these elements the after children.
4. Let s be the concatenation of all the text node children of the noscript element.
5. Set the innerHTML attribute of the parent element to the value of s. (This, as a side-effect, causes the noscript element to be removed from the document.)
6. Insert the before children at the start of the parent element, preserving their original relative order.
7. Insert the after children at the end of the parent element, preserving their original relative order.

The noscript element has no other requirements. In particular, children of the noscript element are not exempt from form submission, scripting, and so forth, even when scripting is enabled.

All these contortions are required because, for historical reasons, the noscript element causes the HTML parser to act differently based on whether scripting is enabled or not. The element is not allowed in XML, because in XML the parser is not affected by such state, and thus the element would not have the desired effect.

3.17.3. The `event-source` element

Block-level element, strictly inline-level content, and metadata element.

Contexts in which this element may be used:

In a head element.

Where block-level elements are expected.

Where inline-level content is expected.

Content model:

Empty.

Element-specific attributes:

src

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLEventSourceElement : HTMLElement {
           attribute DOMString src;
};

The event-source element represents a target for events generated by a remote server.

The src attribute, if specified, must give a URI (or IRI) pointing to a resource that uses the application/x-dom-event-stream format.

When the element is inserted into the document, if it has the src attribute specified, the user agent must act as if the addEventSource() method on the event-source element had been invoked with the URI resulting from resolving the src attribute's value to an absolute URI.

While the element is in a document, if its src attribute is mutated, the user agent must act as if first the removeEventSource() method on the event-source element had been invoked with the URI resulting from resolving the old value of the attribute to an absolute URI, and then as if the addEventSource() method on the element had been invoked with the URI resulting from resolving the new value of the src attribute to an absolute URI.

When the element is removed from the document, if it has the src attribute specified, or, when the src attribute is about to be removed, the user agent must act as if the removeEventSource() method on the event-source element had been invoked with the URI resulting from resolving the src attribute's value to an absolute URI.

There can be more than one event-source element per document, but authors should take care to avoid opening multiple connections to the same server as HTTP recommends a limit to the number of simultaneous connections that a user agent can open per server.

The src DOM attribute must reflect the content attribute of the same name.

3.18. Interactive elements

3.18.1. The `details` element

Contexts in which this element may be used:

Where block-level elements are expected.

Content model:

One legend element followed by either one or more block-level elements or inline-level content (but not both).

Element-specific attributes:

open

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLDetailsElement : HTMLElement {
  attribute boolean open;
};

The details element represents additional information or controls which the user can obtain on demand.

The first element child of a details element, if it is a legend element, represents the summary of the details.

If the first element is not a legend element, the UA should provide its own legend (e.g. "Details").

The open content attribute is a boolean attribute. If present, it indicates that the details should be shown to the user. If the attribute is absent, the details should not be shown.

If the attribute is removed, then the details should be hidden. If the attribute is added, the details should be shown.

The user should be able to request that the details be shown or hidden.

The open attribute must reflect the open content attribute.

Rendering will be described in the Rendering section in due course. Basically CSS :open and :closed match the element, it's a block-level element by default, and when it matches :closed it renders as if it had an XBL binding attached to it whose template was just <template>▶<content includes="legend:first-child">Details</content></template>, and when it's :open it acts as if it had an XBL binding attached to it whose template was just <template>▼<content includes="legend:first-child">Details</content><content/></template> or some such.

Clicking the legend would make it open/close (and would change the content attribute). Question: Do we want the content attribute to reflect the actual state like this? I think we do, the DOM not reflecting state has been a pain in the neck before. But is it semantically ok?

3.18.2. The `datagrid` element

Interactive, block-level element.

Contexts in which this element may be used:

Where block-level elements are expected, if there are no ancestor interactive elements.

Content model:

Zero or more block-level elements.

Element-specific attributes:

multiple

disabled

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLDataGridElement : HTMLElement {
           attribute DataGridDataProvider data;
  readonly attribute DataGridSelection selection;
           attribute boolean multiple;
           attribute boolean disabled;
  void updateEverything();
  void updateRowsChanged(in RowSpecification row, in unsigned long count);
  void updateRowsInserted(in RowSpecification row, in unsigned long count);
  void updateRowsRemoved(in RowSpecification row, in unsigned long count);
  void updateRowChanged(in RowSpecification row);
  void updateColumnChanged(in unsigned long column);
  void updateCellChanged(in RowSpecification row, in unsigned long column);
};

One possible thing to be added is a way to detect when a row/selection has been deleted, activated, etc, by the user (delete key, enter key, etc).

This element is defined as interactive, which means it can't contain other interactive elements, despite the fact that we expect it to work with other interactive elements e.g. checkboxes and input fields. It should be called something like a Leaf Interactive Element or something, which counts for ancestors looking in and not descendants looking out.

The datagrid element represents an interactive representation of tree, list, or tabular data.

The data being presented can come either from the content, as elements given as children of the datagrid element, or from a scripted data provider given by the data DOM attribute.

The multiple and disabled attributes are boolean attributes. Their effects are described in the processing model sections below.

The multiple and disabled DOM attributes must reflect the multiple and disabled content attributes respectively.

3.18.2.1. The `datagrid` data model

This section is non-normative.

In the datagrid data model, data is structured as a set of rows representing a tree, each row being split into a number of columns. The columns are always present in the data model, although individual columns may be hidden in the presentation.

Each row can have child rows. Child rows may be hidden or shown, by closing or opening (respectively) the parent row.

Rows are referred to by the path along the tree that one would take to reach the row, using zero-based indices. Thus, the first row of a list is row "0", the second row is row "1"; the first child row of the first row is row "0,0", the second child row of the first row is row "0,1"; the fourth child of the seventh child of the third child of the tenth row is "9,2,6,3", etc.

The columns can have captions. Those captions are not considered a row in their own right, they are obtained separately.

Selection of data in a datagrid operates at the row level. If the multiple attribute is present, multiple rows can be selected at once, otherwise the user can only select one row at a time.

The datagrid element can be disabled entirely by setting the disabled attribute.

Columns, rows, and cells can each have specific flags, known as classes, applied to them by the data provider. These classes affect the functionality of the datagrid element, and are also passed to the style system. They are similar in concept to the class attribute, except that they are not specified on elements but are given by scripted data providers.

3.18.2.2. How rows are identified

The chains of numbers that give a row's path, or identifier, are represented by objects that implement the RowSpecification interface.

interface RowSpecification {
  // binding-specific interface
};

In ECMAScript, two classes of objects are said to implement this interface: Numbers representing non-negative integers, and homogeneous arrays of Numbers representing non-negative integers. Thus, [1,0,9] is a RowSpecification, as is 1 on its own. However, [1,0.2,9] is not a RowSpecification object, since its second value is not an integer.

User agents must always represent RowSpecifications in ECMAScript by using arrays, even if the path only has one number.

The root of the tree is represented by the empty path; in ECMAScript, this is the empty array ([]). Only the getRowCount() and GetChildAtPosition() methods ever get called with the empty path.

3.18.2.3. The data provider interface

The conformance criteria in this section apply to any implementation of the DataGridDataProvider, including (and most commonly) the content author's implementation(s).

// To be implemented by Web authors as a JS object
interface DataGridDataProvider {
  void initialize(in HTMLDataGridElement datagrid);
  unsigned long getRowCount(in RowSpecification row);
  unsigned long getChildAtPosition(in RowSpecification parentRow, in unsigned long position);
  unsigned long getColumnCount();
  DOMString getCaptionText(in unsigned long column);
  void getCaptionClasses(in unsigned long column, in DOMTokenList classes);
  DOMString getRowImage(in RowSpecification row);
  HTMLMenuElement getRowMenu(in RowSpecification row);
  void getRowClasses(in RowSpecification row, in DOMTokenList classes);
  DOMString getCellData(in RowSpecification row, in unsigned long column);
  void getCellClasses(in RowSpecification row, in unsigned long column, in DOMTokenList classes);
  void toggleColumnSortState(in unsigned long column);
  void setCellCheckedState(in RowSpecification row, in unsigned long column, in long state);
  void cycleCell(in RowSpecification row, in unsigned long column);
  void editCell(in RowSpecification row, in unsigned long column, in DOMString data);
};

The DataGridDataProvider interface represents the interface that objects must implement to be used as custom data views for datagrid elements.

Not all the methods are required. The minimum number of methods that must be implemented in a useful view is two: the getRowCount() and getCellData() methods.

Once the object is written, it must be hooked up to the datagrid using the data DOM attribute.

The following methods may be usefully implemented:

initialize(datagrid): Called by the datagrid element (the one given by the datagrid argument) after it has first populated itself. This would typically be used to set the initial selection of the datagrid element when it is first loaded. The data provider could also use this method call to register a select event handler on the datagrid in order to monitor selection changes.
getRowCount(row): Must return the number of rows that are children of the specified row, including rows that are off-screen. If row is empty, then the number of rows at the top level must be returned. If the value that this method would return for a given row changes, the relevant update methods on the datagrid must be called first. Otherwise, this method must always return the same number. For a list (as opposed to a tree), this method must return 0 whenever it is called with a row identifier that is not empty.
getChildAtPosition(parentRow, position): Must return the index of the row that is a child of parentRow and that is to be positioned as the positionth row under parentRow when rendering the children of parentRow. If parentRow is empty, then position refers to the positionth row at the top level of the data grid. May be omitted if the rows are always to be sorted in the natural order. (The natural order is the one where the method always returns position.) For a given parentRow, this method must never return the same value for different values of position. The returned value x must be in the range 0 ≤ x < n, where n is the value returned by getRowCount(parentRow).
getColumnCount(): Must return the number of columns currently in the data model (including columns that might be hidden). May be omitted if there is only one column. If the value that this method would return changes, the datagrid's updateEverything() method must be called.
getCaptionText(column): Must return the caption, or label, for column column. May be omitted if the columns have no captions. If the value that this method would return changes, the datagrid's updateColumnChanged() method must be called with the appropriate column index.
getCaptionClasses(column, classes): Must add the classes that apply to column column to the classes object. May be omitted if the columns have no special classes. If the classes that this method would add changes, the datagrid's updateColumnChanged() method must be called with the appropriate column index. Some classes have predefined meanings.
getRowImage(row): Must return a URI to an image that represents row row, or the empty string if there is no applicable image. May be omitted if no rows have associated images. If the value that this method would return changes, the datagrid's update methods must be called to update the row in question.
getRowMenu(row): Must return an HTMLMenuElement object that is to be used as a context menu for row row, or null if there is no particular context menu. May be omitted if none of the rows have a special context menu. As this method is called immediately before showing the menu in question, no precautions need to be taken if the return value of this method changes.
getRowClasses(row, classes): Must add the classes that apply to row row to the classes object. May be omitted if the rows have no special classes. If the classes that this method would add changes, the datagrid's update methods must be called to update the row in question. Some classes have predefined meanings.
getCellData(row, column): Must return the value of the cell on row row in column column. For text cells, this must be the text to show for that cell. For progress bar cells, this must be either a floating point number in the range 0.0 to 1.0 (converted to a string representation), indicating the fraction of the progress bar to show as full (1.0 meaning complete), or the empty string, indicating an indeterminate progress bar. If the value that this method would return changes, the datagrid's update methods must be called to update the rows that changed. If only one cell changed, the updateCellChanged() method may be used.
getCellClasses(row, column, classes): Must add the classes that apply to the cell on row row in column column to the classes object. May be omitted if the cells have no special classes. If the classes that this method would add changes, the datagrid's update methods must be called to update the rows or cells in question. Some classes have predefined meanings.
toggleColumnSortState(column): Called by the datagrid when the user tries to sort the data using a particular column column. The data provider must update its state so that the GetChildAtPosition() method returns the new order, and the classes of the columns returned by getCaptionClasses() represent the new sort status. There is no need to tell the datagrid that it the data has changed, as the datagrid automatically assumes that the entire data model will need updating.
setCellCheckedState(row, column, state): Called by the datagrid when the user changes the state of a checkbox cell on row row, column column. The checkbox should be toggled to the state given by state, which is a positive integer (1) if the checkbox is to be checked, zero (0) if it is to be unchecked, and a negative number (-1) if it is to be set to the indeterminate state. There is no need to tell the datagrid that the cell has changed, as the datagrid automatically assumes that the given cell will need updating.
cycleCell(row, column): Called by the datagrid when the user changes the state of a cyclable cell on row row, column column. The data provider should change the state of the cell to the new state, as appropriate. There is no need to tell the datagrid that the cell has changed, as the datagrid automatically assumes that the given cell will need updating.
editCell(row, column, data): Called by the datagrid when the user edits the cell on row row, column column. The new value of the cell is given by data. The data provider should update the cell accordingly. There is no need to tell the datagrid that the cell has changed, as the datagrid automatically assumes that the given cell will need updating.

The following classes (for rows, columns, and cells) may be usefully used in conjunction with this interface:

Class name	Applies to	Description
`checked`	Cells	The cell has a checkbox and it is checked. (The `cyclable` and `progress` classes override this, though.)
`cyclable`	Cells	The cell can be cycled through multiple values. (The `progress` class overrides this, though.)
`editable`	Cells	The cell can be edited. (The `cyclable`, `progress`, `checked`, `unchecked` and `indeterminate` classes override this, though.)
`header`	Rows	The row is a heading, not a data row.
`indeterminate`	Cells	The cell has a checkbox, and it can be set to an indeterminate state. If neither the `checked` nor `unchecked` classes are present, then the checkbox is in that state, too. (The `cyclable` and `progress` classes override this, though.)
`initially-hidden`	Columns	The column will not be shown when the `datagrid` is initially rendered. If this class is not present on the column when the `datagrid` is initially rendered, the column will be visible if space allows.
`initially-closed`	Rows	The row will be closed when the `datagrid` is initially rendered. If neither this class nor the `initially-open` class is present on the row when the `datagrid` is initially rendered, the initial state will depend on platform conventions.
`initially-open`	Rows	The row will be opened when the `datagrid` is initially rendered. If neither this class nor the `initially-closed` class is present on the row when the `datagrid` is initially rendered, the initial state will depend on platform conventions.
`progress`	Cells	The cell is a progress bar.
`reversed`	Columns	If the cell is sorted, the sort direction is descending, instead of ascending.
`selectable-separator`	Rows	The row is a normal, selectable, data row, except that instead of having data, it only has a separator. (The `header` and `separator` classes override this, though.)
`separator`	Rows	The row is a separator row, not a data row. (The `header` class overrides this, though.)
`sortable`	Columns	The data can be sorted by this column.
`sorted`	Columns	The data is sorted by this column. Unless the `reversed` class is also present, the sort direction is ascending.
`unchecked`	Cells	The cell has a checkbox and, unless the `checked` class is present as well, it is unchecked. (The `cyclable` and `progress` classes override this, though.)

3.18.2.4. The default data provider

The user agent must supply a default data provider for the case where the datagrid's data attribute is null. It must act as described in this section.

The behaviour of the default data provider depends on the nature of the first element child of the datagrid.

While the first element child is a table

getRowCount(row): The number of rows returned by the default data provider for the root of the tree (when row is empty) must be the total number of tr elements that are children of tbody elements that are children of the table, if there are any such child tbody elements. If there are no such tbody elements then the number of rows returned for the root must be the number of tr elements that are children of the table.

When row is not empty, the number of rows returned must be zero.

The table-based default data provider cannot represent a tree.

Rows in thead elements do not contribute to the number of rows returned, although they do affect the columns and column captions. Rows in tfoot elements are ignored completely by this algorithm.

getChildAtPosition(row, i): The default data provider must return the mapping appropriate to the current sort order.

getColumnCount(): The number of columns returned must be the number of td element children in the first tr element child of the first tbody element child of the table, if there are any such tbody elements. If there are no such tbody elements, then it must be the number of td element children in the first tr element child of the table, if any, or otherwise 1. If the number that would be returned by these rules is 0, then 1 must be returned instead.

getCaptionText(i): If the table has no thead element child, or if its first thead element child has no tr element child, the default data provider must return the empty string for all captions. Otherwise, the value of the textContent attribute of the ith th element child of the first tr element child of the first thead element child of the table element must be returned. If there is no such th element, the empty string must be returned.

getCaptionClasses(i, classes): If the table has no thead element child, or if its first thead element child has no tr element child, the default data provider must not add any classes for any of the captions. Otherwise, each class in the class attribute of the ith th element child of the first tr element child of the first thead element child of the table element must be added to the classes. If there is no such th element, no classes must be added. The user agent must then:

Remove the sorted and reversed classes.
If the table element has a class attribute that includes the sortable class, add the sortable class.
If the column is the one currently being used to sort the data, add the sorted class.
If the column is the one currently being used to sort the data, and it is sorted in descending order, add the reversed class as well.

The various row- and cell- related methods operate relative to a particular element, the element of the row or cell specified by their arguments.

For rows: Since the default data provider for a table always returns 0 as the number of children for any row other than the root, the path to the row passed to these methods will always consist of a single number. In the prose below, this number is referred to as i.

If the table has tbody element children, the element for the ith row is the ith tr element that is a child of a tbody element that is a child of the table element. If the table does not have tbody element children, then the element for the ith real row is the ith tr element that is a child of the table element.

For cells: Given a row and its element, the row's ith cell's element is the ith td element child of the row element.

The colspan and rowspan attributes are ignored by this algorithm.

getRowImage(i): If the row's first cell's element has an img element child, then the URI of the row's image is the URI of the first img element child of the row's first cell's element. Otherwise, the URI of the row's image is the empty string.

getRowMenu(i): If the row's first cell's element has a menu element child, then the row's menu is the first menu element child of the row's first cell's element. Otherwise, the row has no menu.

getRowClasses(i, classes): The default data provider must never add a class to the row's classes.

toggleColumnSortState(i): If the data is already being sorted on the given column, then the user agent must change the current sort mapping to be the inverse of the current sort mapping; if the sort order was ascending before, it is now descending, otherwise it is now ascending. Otherwise, if the current sort column is another column, or the data model is currently not sorted, the user agent must create a new mapping, which maps rows in the data model to rows in the DOM so that the rows in the data model are sorted by the specified column, in ascending order. (Which sort comparison operator to use is left up to the UA to decide.)

When the sort mapping is changed, the values returned by the getChildAtPosition() method for the default data provider will change appropriately.

getCellData(i, j), getCellClasses(i, j, classes), getCellCheckedState(i, j, state), cycleCell(i, j), and editCell(i, j, data): See the common definitions below.

The data provider must call the datagrid's update methods appropriately whenever the descendants of the datagrid mutate. For example, if a tr is removed, then the updateRowsRemoved() methods would probably need to be invoked, and any change to a cell or its descendants must cause the cell to be updated. If the table element stops being the first child of the datagrid, then the data provider must call the updateEverything() method on the datagrid. Any change to a cell that is in the column that the data provider is currently using as its sort column must also cause the sort to be reperformed, with a call to updateEverything() if the change did affect the sort order.

While the first element child is a select

The default data provider must return 1 for the column count, the empty string for the column's caption, and must not add any classes to the column's classes.

For the rows, assume the existence of a node filter view of the descendants of the first select element child of the datagrid element, that skips all nodes other than optgroup and option elements, as well as any descendents of any option elements.

Given a path row, the corresponding element is the one obtained by drilling into the view, taking the child given by the path each time.

Given the following XML markup:

<datagrid>
 <select>
  <!-- the options and optgroups have had their labels and values removed
       to make the underlying structure clearer -->
  <optgroup>
   <option/>
   <option/>
  </optgroup>
  <optgroup>
   <option/>
   <optgroup id="a">
    <option/>
    <option/>
    <bogus/>
    <option id="b"/>
   </optgroup>
   <option/>
  </optgroup>
 </select>
</datagrid>

The path "1,1,2" would select the element with ID "b". In the filtered view, the text nodes, comment nodes, and bogus elements are ignored; so for instance, the element with ID "a" (path "1,1") has only 3 child nodes in the view.

getRowCount(row) must drill through the view to find the element corresponding to the method's argument, and return the number of child nodes in the filtered view that the corresponding element has. (If the row is empty, the corresponding element is the select element at the root of the filtered view.)

getChildAtPosition(row, position) must return position. (The select default data provider does not support sorting the data grid.)

getRowImage(i) must return the empty string, getRowMenu(i) must return null.

getRowClasses(row, classes) must add the classes from the following list to classes when their condition is met:

If the row's corresponding element is an optgroup element: header
If the row's corresponding element contains other elements that are also in the view, and the element's class attribute contains the closed class: initially-closed
If the row's corresponding element contains other elements that are also in the view, and the element's class attribute contains the open class: initially-open

The getCellData(row, cell) method must return the value of the label attribute if the row's corresponding element is an optgroup element, otherwise, if the row's corresponding element is an optionelement, its label attribute if it has one, otherwise the value of its textContent DOM attribute.

The getCellClasses(row, cell, classes) method must add no classes.

autoselect some rows when initialised, reflect the selection in the select, reflect the multiple attribute somehow.

The data provider must call the datagrid's update methods appropriately whenever the descendants of the datagrid mutate.

While the first element child is another element

The default data provider must return 1 for the column count, the empty string for the column's caption, and must not add any classes to the column's classes.

For the rows, assume the existence of a node filter view of the descendants of the datagrid that skips all nodes other than li, h1-h6, and hr elements, and skips any descendants of menu elements.

Given this view, each element in the view represents a row in the data model. The element corresponding to a path row is the one obtained by drilling into the view, taking the child given by the path each time. The element of the row of a particular method call is the element given by drilling into the view along the path given by the method's arguments.

getRowCount(row) must return the number of child elements in this view for the given row, or the number of elements at the root of the view if the row is empty.

In the following example, the elements are identified by the paths given by their child text nodes:

<datagrid>
 <ol>
  <li> row 0 </li>
  <li> row 1
   <ol>
    <li> row 1,0 </li>
   </ol>
  </li>
  <li> row 2 </li>
 </ol>
</datagrid>

In this example, only the li elements actually appear in the data grid; the ol element does not affect the data grid's processing model.

getChildAtPosition(row, position) must return position. (The generic default data provider does not support sorting the data grid.)

getRowImage(i) must return the URI of the image given by the first img element descendant (in the real DOM) of the row's element, that is not also a descendant of another element in the filtered view that is a descendant of the row's element.

In the following example, the row with path "1,0" returns "http://example.com/a" as its image URI, and the other rows (including the row with path "1") return the empty string:

<datagrid>
 <ol>
  <li> row 0 </li>
  <li> row 1
   <ol>
    <li> row 1,0 <img src="http://example.com/a" alt=""> </li>
   </ol>
  </li>
  <li> row 2 </li>
 </ol>
</datagrid>

getRowMenu(i) must return the first menu element descendant (in the real DOM) of the row's element, that is not also a descendant of another element in the filtered view that is a decsendant of the row's element. (This is analogous to the image case above.)

getRowClasses(i, classes) must add the classes from the following list to classes when their condition is met:

If the row's element contains other elements that are also in the view, and the element's class attribute contains the closed class: initially-closed
If the row's element contains other elements that are also in the view, and the element's class attribute contains the open class: initially-open
If the row's element is an h1-h6 element: header
If the row's element is an hr element: separator

The getCellData(i, j), getCellClasses(i, j, classes), getCellCheckedState(i, j, state), cycleCell(i, j), and editCell(i, j, data) methods must act as described in the common definitions below, treating the row's element as being the cell's element.

selection handling?

The data provider must call the datagrid's update methods appropriately whenever the descendants of the datagrid mutate.

Otherwise, while there is no element child

The data provider must return 0 for the number of rows, 1 for the number of columns, the empty string for the first column's caption, and must add no classes when asked for that column's classes. If the datagrid's child list changes such that there is a first element child, then the data provider must call the updateEverything() method on the datagrid.

3.18.2.4.1. Common default data provider method definitions for cells

These definitions are used for the cell-specific methods of the default data providers (other than in the select case). How they behave is based on the contents of an element that represents the cell given by their first two arguments. Which element that is is defined in the previous section.

Cyclable cells

If the first element child of a cell's element is a select element that has a no multiple attribute and has at least one option element descendent, then the cell acts as a cyclable cell.

The "current" option element is the selected option element, or the first option element if none is selected.

The getCellData() method must return the textContent of the current option element (the label attribute is ignored in this context as the optgroups are not displayed).

The getCellClasses() method must add the cyclable class and then all the classes of the current option element.

The cycleCell() method must change the selection of the select element such that the next option element after the current option element is the only one that is selected (in tree order). If the current option element is the last option element descendent of the select, then the first option element descendent must be selected instead.

The setCellCheckedState() and editCell() methods must do nothing.

Progress bar cells

If the first element child of a cell's element is a progress element, then the cell acts as a progress bar cell.

The getCellData() method must return the value returned by the progress element's position DOM attribute.

The getCellClasses() method must add the progress class.

The setCellCheckedState(), cycleCell(), and editCell() methods must do nothing.

Checkbox cells

If the first element child of a cell's element is an input element that has a type attribute with the value checkbox, then the cell acts as a check box cell.

The getCellData() method must return the textContent of the cell element.

The getCellClasses() method must add the checked class if the input element is checked, and the unchecked class otherwise.

The setCellCheckedState() method must set the input element's checkbox state to checked if the method's third argument is 1, and to unchecked otherwise.

The cycleCell() and editCell() methods must do nothing.

Editable cells

If the first element child of a cell's element is an input element that has a type attribute with the value text or that has no type attribute at all, then the cell acts as an editable cell.

The getCellData() method must return the value of the input element.

The getCellClasses() method must add the editable class.

The editCell() method must set the input element's value DOM attribute to the value of the third argument to the method.

The setCellCheckedState() and cycleCell() methods must do nothing.

3.18.2.5. Populating the `datagrid` element

A datagrid must be disabled until its end tag has been parsed (in the case of a datagrid element in the original document markup) or until it has been inserted into the document (in the case of a dynamically created element). After that point, the element must fire a single load event at itself, which doesn't bubble and cannot be canceled.

The end-tag parsing thing should be moved to the parsing section.

The datagrid must then populate itself using the data provided by the data provider assigned to the data DOM attribute. After the view is populated (using the methods described below), the datagrid must invoke the initialize() method on the data provider specified by the data attribute, passing itself (the HTMLDataGridElement object) as the only argument.

When the data attribute is null, the datagrid must use the default data provider described in the previous section.

To obtain data from the data provider, the element must invoke methods on the data provider object in the following ways:

To determine the total number of columns

Invoke the getColumnCount() method with no arguments. The return value is the number of columns. If the return value is zero or negative, not an integer, or simply not a numeric type, or if the method is not defined, then 1 must be used instead.

To get the captions to use for the columns

Invoke the getCaptionText() method with the index of the column in question. The index i must be in the range 0 ≤ i < N, where N is the total number of columns. The return value is the string to use when referring to that column. If the method returns null or the empty string, the column has no caption. If the method is not defined, then none of the columns have any captions.

To establish what classes apply to a column

Invoke the getCaptionClasses() method with the index of the column in question, and an object implementing the DOMTokenList interface, associated with an anonymous empty string. The index i must be in the range 0 ≤ i < N, where N is the total number of columns. The tokens contained in the string underlying DOMTokenList object when the method returns represent the classes that apply to the given column. If the method is not defined, no classes apply to the column.

To establish whether a column should be initially included in the visible columns

Check whether the initially-hidden class applies to the column. If it does, then the column should not be initially included; if it does not, then the column should be initially included.

To establish whether the data can be sorted relative to a particular column

Check whether the sortable class applies to the column. If it does, then the user should be able to ask the UA to display the data sorted by that column; if it does not, then the user agent must not allow the user to ask for the data to be sorted by that column.

To establish if a column is a sorted column

If the user agent can handle multiple columns being marked as sorted simultaneously: Check whether the sorted class applies to the column. If it does, then that column is the sorted column, otherwise it is not.

If the user agent can only handle one column being marked as sorted at a time: Check each column in turn, starting with the first one, to see whether the sorted class applies to that column. The first column that has that class, if any, is the sorted column. If none of the columns have that class, there is no sorted column.

To establish the sort direction of a sorted column

Check whether the reversed class applies to the column. If it does, then the sort direction is descending (down; first rows have the highest values), otherwise it is ascending (up; first rows have the lowest values).

To determine the total number of rows

Determine the number of rows for the root of the data grid, and determine the number of child rows for each open row. The total number of rows is the sum of all these numbers.

To determine the number of rows for the root of the data grid

Invoke the getRowCount() method with a RowSpecification object representing the empty path as its only argument. The return value is the number of rows at the top level of the data grid. If the return value of the method is negative, not an integer, or simply not a numeric type, or if the method is not defined, then zero must be used instead.

To determine the number of child rows for a row

Invoke the getRowCount() method with a RowSpecification object representing the path to the row in question. The return value is the number of child rows for the given row. If the return value of the method is negative, not an integer, or simply not a numeric type, or if the method is not defined, then zero must be used instead.

To determine what order to render rows in

Invoke the getChildAtPosition() method with a RowSpecification object representing the path to the parent of the rows that are being rendered as the first argument, and the position that is being rendered as the second argument. The return value is the index of the row to render in that position.

If the rows are:

Row "0"
1. Row "0,0"
2. Row "0,1"
Row "1"
1. Row "1,0"
2. Row "1,1"

...and the getChildAtPosition() method is implemented as follows:

function getChildAtPosition(parent, child) {
  // always return the reverse order
  return getRowCount(parent)-child-1;
}

...then the rendering would actually be:

Row "1"
1. Row "1,1"
2. Row "1,0"
Row "0"
1. Row "0,1"
2. Row "0,0"

If the return value of the method is negative, larger than the number of rows that the getRowCount() method reported for that parent, not an integer, or simply not a numeric type, then the entire data grid should be disabled. Similarly, if the method returns the same value for two or more different values for the second argument (with the same first argument, and assuming that the data grid hasn't had relevant update methods invoked in the meantime), then the data grid should be disabled. Instead of disabling the data grid, the user agent may act as if the getChildAtPosition() method was not defined on the data provider (thus disabling sorting for that data grid, but still letting the user interact with the data). If the method is not defined, then the return value must be assumed to be the same as the second argument (an indentity transform; the data is rendered in its natural order).

To establish what classes apply to a row

Invoke the getRowClasses() method with a RowSpecification object representing the row in question, and a DOMTokenList associated with an empty string. The tokens contained in the DOMTokenList object's underlying string when the method returns represent the classes that apply to the row in question. If the method is not defined, no classes apply to the row.

To establish whether a row is a data row or a special row

Examine the classes that apply to the row. If the header class applies to the row, then it is not a data row, it is a subheading. The data from the first cell of the row is the text of the subheading, the rest of the cells must be ignored. Otherwise, if the separator class applies to the row, then in the place of the row, a separator should be shown. Otherwise, if the selectable-separator class applies to the row, then the row should be a data row, but represented as a separator. (The difference between a separator and a selectable-separator is that the former is not an item that can be actually selected, whereas the second can be selected and thus has a context menu that applies to it, and so forth.) For both kinds of separator rows, the data of the rows' cells must all be ignored. If none of those three classes apply then the row is a simple data row.

To establish whether a row is openable

Determine the number of child rows for that row. If there are one or more child rows, then the row is openable.

To establish whether a row should be initially open or closed

If the row is openable, examine the classes that apply to the row. If the initially-open class applies to the row, then it should be initially open. Otherwise, if the initially-closed class applies to the row, then it must be initially closed. Otherwise, if neither class applies to the row, or if the row is not openable, then the initial state of the row is entirely up to the UA.

To obtain a URI to an image representing a row

Invoke the getRowImage() method with a RowSpecification object representing the row in question. The return value is a string representing a URI (or IRI) to an image. Relative URIs must be interpreted relative to the datagrid's base URI. If the method returns the empty string, null, or if the method is not defined, then the row has no associated image.

To obtain a context menu appropriate for a particular row

Invoke the getRowMenu() method with a RowSpecification object representing the row in question. The return value is a reference to an object implementing the HTMLMenuElement interface, i.e. a menu element DOM node. (This element must then be interpreted as described in the section on context menus to obtain the actual context menu to use.) If the method returns something that is not an HTMLMenuElement, or if the method is not defined, then the row has no associated context menu. User agents may provide their own default context menu, and may add items to the author-provided context menu. For example, such a menu could allow the user to change the presentation of the datagrid element.

To establish the value of a particular cell

Invoke the getCellData() method with the first argument being a RowSpecification object representing the row of the cell in question and the second argument being the index of the cell's column. The second argument must be a non-negative integer less than the total number of columns. The return value is the value of the cell. If the return value is null or the empty string, or if the method is not defined, then the cell has no data. (For progress bar cells, the cell's value must be further interpreted, as described below.)

To establish what classes apply to a cell

Invoke the getCellClasses() method with the first argument being a RowSpecification object representing the row of the cell in question, the second argument being the index of the cell's column, and the third being an object implementing the DOMTokenList interface, associated with an empty string. The second argument must be a non-negative integer less than the total number of columns. The tokens contained in the DOMTokenList object's underlying string when the method returns represent the classes that apply to that cell. If the method is not defined, no classes apply to the cell.

To establish how the type of a cell

Examine the classes that apply to the cell. If the progress class applies to the cell, it is a progress bar. Otherwise, if the cyclable class applies to the cell, it is a cycling cell whose value can be cycled between multiple states. Otherwise, none of these classes apply, and the cell is a simple text cell.

To establish the value of a progress bar cell

If the value x of the cell is a string that can be converted to a floating-point number in the range 0.0 ≤ x ≤ 1.0, then the progress bar has that value (0.0 means no progress, 1.0 means complete). Otherwise, the progress bar is an indeterminate progress bar.

To establish how a simple text cell should be presented

Check whether one of the checked, unchecked, or indeterminate classes applies to the cell. If any of these are present, then the cell has a checkbox, otherwise none are present and the cell does not have a checkbox. If the cell has no checkbox, check whether the editable class applies to the cell. If it does, then the cell value is editable, otherwise the cell value is static.

To establish the state of a cell's checkbox, if it has one

Check whether the checked class applies to the cell. If it does, the cell is checked. Otherwise, check whether the unchecked class applies to the cell. If it does, the cell is unchecked. Otherwise, the indeterminate class appplies to the cell and the cell's checkbox is in an indeterminate state. When the indeterminate class appplies to the cell, the checkbox is a tristate checkbox, and the user can set it to the indeterminate state. Otherwise, only the checked and/or unchecked classes apply to the cell, and the cell can only be toggled betwen those two states.

If the data provider ever raises an exception while the datagrid is invoking one of its methods, the datagrid must act, for the purposes of that particular method call, as if the relevant method had not been defined.

A RowSpecification object p with n path components passed to a method of the data provider must fulfill the constraint 0 ≤ p_i < m-1 for all integer values of i in the range 0 ≤ i < n-1, where m is the value that was last returned by the getRowCount() method when it was passed the RowSpecification object q with i-1 items, where p_i = q_i for all integer values of i in the range 0 ≤ i < n-1, with any changes implied by the update methods taken into account.

The data model is considered stable: user agents may assume that subsequent calls to the data provider methods will return the same data, until one of the update methods is called on the datagrid element. If a user agent is returned inconsistent data, for example if the number of rows returned by getRowCount() varies in ways that do not match the calls made to the update methods, the user agent may disable the datagrid. User agents that do not disable the datagrid in inconsistent cases must honour the most recently returned values.

User agents may cache returned values so that the data provider is never asked for data that could contradict earlier data. User agents must not cache the return value of the getRowMenu method.

The exact algorithm used to populate the data grid is not defined here, since it will differ based on the presentation used. However, the behaviour of user agents must be consistent with the descriptions above. For example, it would be non-conformant for a user agent to make cells have both a checkbox and be editable, as the descriptions above state that cells that have a checkbox cannot be edited.

3.18.2.6. Updating the `datagrid`

Whenever the data attribute is set to a new value, the datagrid must clear the current selection, remove all the displayed rows, and plan to repopulate itself using the information from the new data provider at the earliest opportunity.

There are a number of update methods that can be invoked on the datagrid element to cause it to refresh itself in slightly less drastic ways:

When the updateEverything() method is called, the user agent must repopulate the entire datagrid. If the number of rows decreased, the selection must be updated appropriately. If the number of rows increased, the new rows should be left unselected.

When the updateRowsChanged(row, count) method is called, the user agent must refresh the rendering of the rows starting from the row specified by row, and including the count next siblings of the row (or as many next siblings as it has, if that is less than count), including all descendant rows.

When the updateRowsInserted(row, count) method is called, the user agent must assume that count new rows have been inserted, such that the first new row is indentified by row. The user agent must update its rendering and the selection accordingly. The new rows should not be selected.

When the updateRowsRemoved(row, count) method is called, the user agent must assume that count rows have been removed starting from the row that used to be identifier by row. The user agent must update its rendering and the selection accordingly.

The updateRowChanged(row) method must be exactly equivalent to calling updateRowsChanged(row, 1).

When the updateColumnChanged(column) method is called, the user agent must refresh the rendering of the specified column column, for all rows.

When the updateCellChanged(row, column) method is called, the user agent must refresh the rendering of the cell on row row, in column column.

Any effects the update methods have on the datagrid's selection is not considered a change to the selection, and must therefore not fire the select event.

These update methods should only be called by the data provider, or code acting on behalf of the data provider. In particular, calling the updateRowsInserted() and updateRowsRemoved() methods without actually inserting or removing rows from the data provider is likely to result in inconsistent renderings, and the user agent is likely to disable the data grid.

3.18.2.7. Requirements for interactive user agents

This section only applies to interactive user agents.

If the datagrid element has a disabled attribute, then the user agent must disable the datagrid, preventing the user from interacting with it. The datagrid element should still continue to update itself when the data provider signals changes to the data, though. Obviously, conformance requirements stating that datagrid elements must react to users in particular ways do not apply when one is disabled.

If a row is openable, then the user should be able to toggle its open/closed state. When a row's open/closed state changes, the user agent must update the rendering to match the new state.

If a cell is a cell whose value can be cycled between multiple states, then the user must be able to activate the cell to cycle its value. When the user activates this "cycling" behaviour of a cell, then the datagrid must invoke the data provider's cycleCell() method, with a RowSpecification object representing the cell's row as the first argument and the cell's column index as the second. The datagrid must act as if the datagrid's updateCellChanged() method had been invoked with those same arguments immediately before the provider's method was invoked.

When a cell has a checkbox, the user must be able to set the checkbox's state. When the user changes the state of a checkbox in such a cell, the datagrid must invoke the data provider's setCellCheckedState() method, with a RowSpecification object representing the cell's row as the first argument, the cell's column index as the second, and the checkbox's new state as the third. The state should be represented by the number 1 if the new state is checked, 0 if the new state is unchecked, and -1 if the new state is indeterminate (which must only be possible if the cell has the indeterminate class set). The datagrid must act as if the datagrid's updateCellChanged() method had been invoked, specifying the same cell, immediately before the provider's method was invoked.

If a cell is editable, the user must be able to edit the data for that cell, and doing so must cause the user agent to invoke the editCell() method of the data provider with three arguments: a RowSpecification object representing the cell's row, the cell's column's index, and the new text entered by the user. The user agent must act as if the updateCellChanged() method had been invoked, with the same row and column specified, immediately before the provider's method was invoked.

3.18.2.8. The selection

This section only applies to interactive user agents. For other user agents, the selection attribute must return null.

interface DataGridSelection {
  readonly attribute unsigned long length;
  RowSpecification item(in unsigned long index);
  boolean isSelected(in RowSpecification row);
  void setSelected(in RowSpecification row, in boolean selected);

  void selectAll();
  void invert();
  void clear();
};

Each datagrid element must keep track of which rows are currently selected. Initially no rows are selected, but this can be changed via the methods described in this section.

The selection of a datagrid is represented by its selection DOM attribute, which must be a DataGridSelection object.

DataGridSelection objects represent the rows in the selection. In the selection the rows must be ordered in the natural order of the data provider (and not, e.g., the rendered order). Rows that are not rendered because one of their ancestors is closed must share the same selection state as their nearest rendered ancestor. Such rows are not considered part of the selection for the purposes of iterating over the selection.

This selection API doesn't allow for hidden rows to be selected because it is trivial to create a data provider that has infinite depth, which would then require the selection to be infinite if every row, including every hidden row, was selected.

The length attribute must return the number of rows currently present in the selection. The item(index) method must return the indexth row in the selection. If the argument is out of range (less than zero or greater than the number of selected rows minus one), then it must raise an INDEX_SIZE_ERR exception. [DOM3CORE]

The isSelected() method must return the selected state of the row specified by its argument. If the specified row exists and is selected, it must return true, otherwise it must return false.

The setSelected() method takes two arguments, row and selected. When invoked, it must set the selection state of row row to selected if selected is true, and unselected if it is false. If row is not a row in the data grid, the method must raise an INDEX_SIZE_ERR exception. If the specified row is not rendered because one of its ancestors is closed, the method must do nothing.

The selectAll() method must mark all the rows in the data grid as selected. After a call to selectAll(), the length attribute will return the number of rows in the data grid, not counting children of closed rows.

The invert() method must cause all the rows in the selection that were marked as selected to now be marked as not selected, and vice versa.

The clear() method must mark all the rows in the data grid to be marked as not selected. After a call to clear(), the length attribute will return zero.

If the datagrid element has a multiple attribute, then the user must be able to select any number of rows (zero or more). If the attribute is not present, then the user must only be able to select a single row at a time, and selecting another one must unselect all the other rows.

This only applies to the user. Scripts can select multiple rows even when the multiple attribute is absent.

Whenever the selection of a datagrid changes, whether due to the user interacting with the element, or as a result of calls to methods of the selection object, a select event that bubbles but is not cancelable must be fired on the datagrid element. If changes are made to the selection via calls to the object's methods during the execution of a script, then the select events must be coalesced into one, which must then be fired when the script execution has completed.

The DataGridSelection interface has no relation to the Selection interface.

3.18.2.9. Columns and captions

This section only applies to interactive user agents.

Each datagrid element must keep track of which columns are currently being rendered. User agents should initially show all the columns except those with the initially-hidden class, but may allow users to hide or show columns. User agents should initially display the columns in the order given by the data provider, but may allow this order to be changed by the user.

If columns are not being used, as might be the case if the data grid is being presented in an icon view, or if an overview of data is being read in an aural context, then the text of the first column of each row should be used to represent the row.

If none of the columns have any captions (i.e. if the data provider does not provide a getCaptionText() method), then user agents may avoid showing the column headers at all. This may prevent the user from performing actions on the columns (such as reordering them, changing the sort column, and so on).

Whatever the order used for rendering, and irrespective of what columns are being shown or hidden, the "first column" as referred to in this specification is always the column with index zero, and the "last column" is always the column with the index one less than the value returned by the getColumnCount() method of the data provider.

If a column is sortable, then the user must be able to invoke it to sort the data. When the user does so, then the datagrid must invoke the data provider's toggleColumnSortState() method, with the column's index as the only argument. The datagrid must then act as if the datagrid's updateEverything() method had been invoked.

3.18.3. The `command` element

Metadata element, and strictly inline-level content.

Contexts in which this element may be used:

In a head element.

Where strictly inline-level content is allowed.

Content model:

Empty.

Element-specific attributes:

type

label

icon

hidden

disabled

checked

radiogroup

default

Also, the title attribute has special semantics on this element.

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLCommandElement : HTMLElement {
           attribute DOMString type;
           attribute DOMString label;
           attribute DOMString icon;
           attribute boolean hidden;
           attribute boolean disabled;
           attribute boolean checked;
           attribute DOMString radiogroup;
           attribute boolean default;
 void click(); // shadows HTMLElement.click()
};

The Command interface must also be implemented by this element.

The command element represents a command that the user can invoke.

The type attribute indicates the kind of command: either a normal command with an associated action, or a state or option that can be toggled, or a selection of one item from a list of items.

The attribute's value must be either "command", "checkbox", or "radio", denoting each of these three types of commands respectively. The attribute may also be omitted if the element is to represent the first of these types, a simple command.

The label attribute gives the name of the command, as shown to the user.

The title attribute gives a hint describing the command, which might be shown to the user to help him.

The icon attribute gives a picture that represents the command. If the attribute is specified, the attribute's value must contain a URI (or IRI).

The hidden attribute is a boolean attribute that, if present, indicates that the command is not relevant and is to be hidden.

The disabled attribute is a boolean attribute that, if present, indicates that the command is not available in the current state.

The distinction between Disabled State and Hidden State is subtle. A command should be Disabled if, in the same context, it could be enabled if only certain aspects of the situation were changed. A command should be marked as Hidden if, in that situation, the command will never be enabled. For example, in the context menu for a water faucet, the command "open" might be Disabled if the faucet is already open, but the command "eat" would be marked Hidden since the faucet could never be eaten.

The checked attribute is a boolean attribute that, if present, indicates that the command is selected.

The radiogroup attribute gives the name of the group of commands that will be toggled when the command itself is toggled, for commands whose type attribute has the value "radio". The scope of the name is the child list of the parent element.

If the command element is used when generating a context menu, then the default attribute indicates, if present, that the command is the one that would have been invoked if the user had directly activated the menu's subject instead of using its context menu. The default attribute is a boolean attribute.

Need an example that shows an element that, if double-clicked, invokes an action, but that also has a context menu, showing the various command attributes off, and that has a default command.

The type, label, icon, hidden, disabled, checked, radiogroup, and default DOM attributes must reflect their respective namesake content attributes.

The click() method's behaviour depends on the value of the type attribute of the element, as follows:

If the type attribute has the value checkbox

If the element has a checked attribute, the UA must remove that attribute. Otherwise, the UA must add a checked attribute, with the literal value checked. The UA must then fire a click event at the element.

If the type attribute has the value radio

If the element has a parent, then the UA must walk the list of child nodes of that parent element, and for each node that is a command element, if that element has a radiogroup attribute whose value exactly matches the current element's (treating missing radiogroup attributes as if they were the empty string), and has a checked attribute, must remove that attribute and fire a click event at the element.

Then, the element's checked attribute attribute must be set to the literal value checked and a click event must be fired at the element.

Otherwise

The UA must fire a click event at the element.

Firing a synthetic click event at the element does not cause any of the actions described above to happen.

should change all the above so it actually is just trigged by a click event, then we could remove the shadowing click() method and rely on actual events.

Need to define the command="" attribute

command elements are not rendered unless they form part of a menu.

3.18.4. The `menu` element

Block-level element, and structured inline-level element.

Contexts in which this element may be used:

Where block-level elements are expected.

Where structured inline-level elements are allowed.

Content model:

Zero or more li elements, or inline-level content (but not both).

Element-specific attributes:

type

label

autosubmit

Predefined classes that apply to this element:

None.

DOM interface:

interface HTMLMenuElement : HTMLElement {
           attribute DOMString type;
           attribute DOMString label;
           attribute boolean autosubmit;
};

The menu element represents a list of commands.

The type attribute indicates the kind of menu. It must have either the value popup (to declare a context menu) or the value toolbar (to define a tool bar). The attribute may also be omitted, to indicate that the element is merely a list of commands that is neither declaring a context menu nor defining a tool bar.

If a menu element has a type attribute with the value popup, then it represents the commands of a context menu, and the user can only interact with the commands if that context menu is activated.

If a menu element has a type attribute with the value toolbar, then it represents a list of active commands that the user can immediately interact with.

Otherwise, if a menu element has no type attribute, or if has a type attribute with a value other than popup or toolbar, then it either represents an unordered list of items (each represented by an li element), each of which represents a command that the user may perform or activate, or, if the element has no li element children, a paragraph describing available commands.

The label attribute gives the label of the menu. It is used by user agents to display nested menus in the UI. For example, a context menu containing another menu would use the nested menu's label attribute for the submenu's menu label.

The autosubmit attribute is a boolean attribute that, if present, indicates that selections made to form controls in this menu are to result in the control's form being immediately submitted.

If a change event bubbles through a menu element, then, in addition to any other default action that that event might have, the UA must act as if the following was an additional default action for that event: if (when it comes time to execute the default action) the menu element has an autosubmit attribute, and the target of the event is an input element, and that element has a type attribute whose value is either radio or checkbox, and the input element in question has a non-null form DOM attribute, then the UA must invoke the submit() method of the form element indicated by that DOM attribute.

3.18.4.1. Introduction

This section is non-normative.

...

3.18.4.2. Building menus

A menu consists of a list of zero or more of the following components:

Commands, which can be marked as default commands
Separators
Other menus (which allows the list to be nested)

The list corresponding to a particular menu element is built by iterating over its child nodes. For each child node in tree order, the required behaviour depends on what the node is, as follows:

An element that defines a command: Append the command to the menu. If the element is a command element with a default attribute, mark the command as being a default command.
An hr element
An option element that has a value attribute set to the empty string, and has a disabled attribute, and whose textContent consists of a string of one or more hyphens (U+002D HYPHEN-MINUS): Append a separator to the menu.
An li element: Iterate over the children of the li element.
A menu element with no label attribute
A select element: Append a separator to the menu, then iterate over the children of the menu or select element, then append another separator.
A menu element with a label attribute
An optgroup element: Append a submenu to the menu, using the value of the element's label attribute as the label of the menu. The submenu must be constructed by taking the element and creating a new menu for it using the complete process described in this section.
Any other node: Ignore the node.

Once all the nodes have been processed as described above, the user agent must the post-process the menu as follows:

Any menu item with no label, or whose label is the empty string, must be removed.
Any sequence of two or more separators in a row must be collapsed to a single separator.
Any separator at the start or end of the menu must be removed.

3.18.4.3. Context menus

The contextmenu attribute gives the element's context menu. The value must be the ID of a menu element in the DOM. If the node that would be obtained by the invoking the getElementById() method using the attribute's value as the only argument is null or not a menu element, then the element has no assigned context menu. Otherwise, the element's assigned context menu is the element so identified.

When an element's context menu is requested (e.g. by the user right-clicking the element, or pressing a context menu key), the UA must fire a contextmenu event on the element for which the menu was requested.

Typically, therefore, the firing of the contextmenu event will be the default action of a mouseup or keyup event. The exact sequence of events is UA-dependent, as it will vary based on platform conventions.

The default action of the contextmenu event depends on whether the element has a context menu assigned (using the contextmenu attribute) or not. If it does not, the default action must be for the user agent to show its default context menu, if it has one.

If the element does have a context menu assigned, then the user agent must fire a show event on the relevant menu element.

The default action of this event is that the user agent must show a context menu built from the menu element.

The user agent may also provide access to its default context menu, if any, with the context menu shown. For example, it could merge the menu items from the two menus together, or provide the page's context menu as a submenu of the default menu.

If the user dismisses the menu without making a selection, nothing in particular happens.

If the user selects a menu item that represents a command, then the UA must invoke that command's Action.

Context menus must not, while being shown, reflect changes in the DOM; they are constructed as the default action of the show event and must remain like that until dismissed.

User agents may provide means for bypassing the context menu processing model, ensuring that the user can always access the UA's default context menus. For example, the user agent could handle right-clicks that have the Shift key depressed in such a way that it does not fire the contextmenu event and instead always shows the default context menu.

The contextMenu attribute must reflect the contextmenu content attribute.

3.18.4.4. Toolbars

Toolbars are a kind of menu that is always visible.

When a menu element has a type attribute with the value toolbar, then the user agent must build the menu for that menu element and render it in the document in a position appropriate for that menu element.

The user agent must reflect changes made to the menu's DOM immediately in the UI.

3.18.5. Commands

A command is the abstraction behind menu items, buttons, and links. Once a command is defined, other parts of the interface can refer to the same command, allowing many access points to a single feature to share aspects such as the disabled state.

Type: The kind of command: "command", meaning it is a normal command; "radio", meaning that triggering the command will, amongst other things, set the Checked State to true (and probably uncheck some other commands); or "checkbox", meaning that triggering the command will, amongst other things, toggle the value of the Checked State.
ID: The name of the command, for referring to the command from the markup or from script. If a command has no ID, it is an anonymous command.
Label: The name of the command as seen by the user.
Hint: A helpful or descriptive string that can be shown to the user.
Icon: A graphical image that represents the action.
Hidden State: Whether the command is hidden or not (basically, whether it should be shown in menus).
Disabled State: Whether the command can be triggered or not. If the Hidden State is true (hidden) then the Disabled State will be true (disabled) regardless. We could make this into a string value that acts as a Hint for why the command is disabled.
Checked State: Whether the command is checked or not.
Action: The actual effect that triggering the command will have. This could be a scripted event handler, a URI to which to navigate, or a form submission.
Triggers: The list of elements that can trigger the command. The element defining a command is always in the list of elements that can trigger the command. For anonymous commands, only the element defining the command is on the list, since other elements have no way to refer to it.

Commands are represented by elements in the DOM. Any element that can define a command also implements the Command interface:

interface Command {
  readonly attribute DOMString commandType;          
  readonly attribute DOMString id;
  readonly attribute DOMString label;
  readonly attribute DOMString title;
  readonly attribute DOMString icon;
  readonly attribute boolean hidden;
  readonly attribute boolean disabled;              
  readonly attribute boolean checked;              
  void click();
  readonly attribute HTMLCollection triggers;
  readonly attribute Command command;
};

The Command interface is implemented by any element capable of defining a command. (If an element can define a command, its definition will list this interface explicitly.) All the attributes of the Command interface are read-only. Elements implementing this interface may implement other interfaces that have attributes with identical names but that are mutable; in bindings that simply flatten all supported interfaces on the object, the mutable attributes must shadow the readonly attributes defined in the Command interface.

The commandType attribute must return a string whose value is either "command", "radio", or "checked", depending on whether the Type of the command defined by the element is "command", "radio", or "checked" respectively. If the element does not define a command, it must return null.

The id attribute must return the command's ID, or null if the element does not define a command or defines an anonymous command. This attribute will be shadowed by the id DOM attribute on the HTMLElement interface.

The label attribute must return the command's Label, or null if the element does not define a command or does not specify a Label. This attribute will be shadowed by the label DOM attribute on option and command elements.

The title attribute must return the command's Hint, or null if the element does not define a command or does not specify a Hint. This attribute will be shadowed by the title DOM attribute on the HTMLElement interface.

The icon attribute must return an absolute URI to the command's Icon. If the element does not specify an icon, or if the element does not define a command, then the attribute must return null. This attribute will be shadowed by the icon DOM attribute on command elements.

The hidden attribute must return true if the command's Hidden State is that the command is hidden, and false if it is that the command is not hidden. If the element does not define a command, the attribute must return false. This attribute will be shadowed by the hidden DOM attribute on command elements.

The disabled attribute must return true if the command's Disabled State is that the command is disabled, and false if the command is not disabled. This attribute is not affected by the command's Hidden State. If the element does not define a command, the attribute must return false. This attribute will be shadowed by the disabled attribute on button, input, option, and command elements.

The checked attribute must return true if the command's Checked State is that the command is checked, and false if it is that the command is not checked. If the element does not define a command, the attribute must return false. This attribute will be shadowed by the checked attribute on input and command elements.

The click() method must trigger the Action for the command. If the element does not define a command, this method must do nothing. This method will be shadowed by the click() method on HTML elements, and is included only for completeness.

The triggers attribute must return a list containing the elements that can trigger the command (the command's Triggers). The list must be live. While the element does not define a command, the list must be empty.

The commands attribute of the document's HTMLDocument interface must return an HTMLCollection rooted at the Document node, whose filter matches only elements that define commands and have IDs.

The following elements can define commands: a, button, input, option, command.

3.18.5.1. Using the `a` element to define a command

An a element with an href attribute defines a command.

The Type of the command is "command".

The ID of the command is the value of the id attribute of the element, if the attribute is present and not empty. Otherwise the command is an anonymous command.

The Label of the command is the string given by the element's textContent DOM attribute.

The Hint of the command is the value of the title attribute of the a element. If the attribute is not present, the Hint is the empty string.

The Icon of the command is the absolute URI of the first image in the element. Specifically, in a depth-first search of the children of the element, the first element that is img element with a src attribute is the one that is used as the image. The URI must be taken from the element's src attribute. Relative URIs must be resolved relative to the base URI of the image element. If no image is found, then the Icon facet is left blank.

The Hidden State and Disabled State facets of the command are always false. (The command is always enabled.)

The Checked State of the command is always false. (The command is never checked.)

The Action of the command is to fire a click event at the element.

3.18.5.2. Using the `button` element to define a command

A button element always defines a command.

The Type, ID, Label, Hint, Icon, Hidden State, Checked State, and Action facets of the command are determined as for a elements (see the previous section).

The Disabled State of the command mirrors the disabled state of the button. Typically this is given by the element's disabled attribute, but certain button types become disabled at other times too (for example, the move-up button type is disabled when it would have no effect).

3.18.5.3. Using the `input` element to define a command

An input element whose type attribute is one of submit, reset, button, radio, checkbox, move-up, move-down, add, and remove defines a command.

The Type of the command is "radio" if the type attribute has the value radio, "checkbox" if the type attribute has the value checkbox, and "command" otherwise.

The ID of the command is the value of the id attribute of the element, if the attribute is present and not empty. Otherwise the command is an anonymous command.

The Label of the command depends on the Type of the command:

If the Type is "command", then it is the string given by the value attribute, if any, and a UA-dependent value that the UA uses to label the button itself if the attribute is absent.

Otherwise, the Type is "radio" or "checkbox". If the element has a label element associated with it, the textContent of the first such element is the Label (in DOM terms, this the string given by element.labels[0].textContent). Otherwise, the value of the value attribute, if present, is the Label. Otherwise, the Label is the empty string.

The Hint of the command is the value of the title attribute of the input element. If the attribute is not present, the Hint is the empty string.

There is no Icon for the command.

The Hidden State of the command is always false. (The command is never hidden.)

The Disabled State of the command mirrors the disabled state of the control. Typically this is given by the element's disabled attribute, but certain input types become disabled at other times too (for example, the move-up input type is disabled when it would have no effect).

The Checked State of the command is true if the command is of Type "radio" or "checkbox" and the element has a checked attribute, and false otherwise.

The Action of the command is to fire a click event at the element.

3.18.5.4. Using the `option` element to define a command

An option element with an ancestor select element and either no value attribute or a value attribute that is not the empty string defines a command.

The Type of the command is "radio" if the option's nearest ancestor select element has no multiple attribute, and "checkbox" if it does.

The ID of the command is the value of the id attribute of the element, if the attribute is present and not empty. Otherwise the command is an anonymous command.

The Label of the command is the value of the option element's label attribute, if there is one, or the value of the option element's textContent DOM attribute if it doesn't.

The Hint of the command is the string given by the element's title attribute, if any, and the empty string if the attribute is absent.

There is no Icon for the command.

The Hidden State of the command is always false. (The command is never hidden.)

The Disabled State of the command is true (disabled) if the element has a disabled attribute, and false otherwise.

The Checked State of the command is true (checked) if the element's selected DOM attribute is true, and false otherwise.

The Action of the command depends on its Type. If the command is of Type "radio" then this must set the selected DOM attribute of the option element to true, otherwise it must toggle the state of the selected DOM attribute (set it to true if it is false and vice versa). Then a change event must be fired on the option element's nearest ancestor select element (if there is one), as if the selection had been changed directly.

3.18.5.5. Using the `command` element to define a command

A command element defines a command.

The Type of the command is "radio" if the command's type attribute is "radio", "checkbox" if the attribute's value is "checkbox", and "command" otherwise.

The ID of the command is the value of the id attribute of the element, if the attribute is present and not empty. Otherwise the command is an anonymous command.

The Label of the command is the value of the element's label attribute, if there is one, or the empty string if it doesn't.

The Hint of the command is the string given by the element's title attribute, if any, and the empty string if the attribute is absent.

The Icon for the command is the absolute URI resulting from resolving the value of the element's icon attribute as a URI relative to the element's base URI. If the element has no icon attribute then the command has no Icon.

The Hidden State of the command is true (hidden) if the element has a hidden attribute, and false otherwise.

The Disabled State of the command is true (disabled) if the element has either a disabled attribute or a hidden attribute (or both), and false otherwise.

The Checked State of the command is true (checked) if the element has a checked attribute, and false otherwise.

The Action of the command is to invoke the behaviour described in the definition of the click() method of the HTMLCommandElement interface.

3.19. Miscellaneous elements

3.19.1. The `legend` element

Contexts in which this element may be used:: As the first child of a fieldset element.; As the first child of a details element.; As a child of a figure element, if there are no other legend element children of that element.
Content model:: If used as a child of a fieldset or details element: significant strictly inline-level content; If used as a child of a figure element: inline-level content.
Element-specific attributes:: None.
Predefined classes that apply to this element:: None.
DOM interface:: No difference from HTMLElement.

The legend element represents a title or explanatory caption for the rest of the contents of the legend element's parent element.

3.19.2. The `div` element

Generate implied end tags.

Contexts in which this element may be used:: Where block-level elements are expected.
Content model:: Zero or more style elements, followed by either zero or more block-level elements, or inline-level content (but not both).
Element-specific attributes:: None.
Predefined classes that apply to this element:: error, example, issue, note, search, warning
DOM interface:: No difference from HTMLElement.

The div element represents nothing at all. It can be used with the class, lang/xml:lang, and title attributes to mark up semantics common to a group of consecutive elements.

4. Web browsers

This section describes features that apply most directly to Web browsers. Having said that, unless specified elsewhere, the requirements defined in this section do apply to all user agents, whether they are Web browsers or not.

4.1. Browsing contexts

A browsing context is a collection of one or more Document objects, and one or more views.

At any one time, one of the Documents in a browsing context is the active document. The collection of Documents is the browsing context's session history.

A view is a user agent interface tied to a particular media used for the presentation of Document objects in some media. A view may be interactive. Each view is represented by an AbstractView object. Each view belongs to a browsing context. [DOM2VIEWS]

The document attribute of an AbstractView object representing a view gives the Document object of the view's browsing context's active document. [DOM2VIEWS]

Events that use the UIEvent interface are related to a specific view (the view in which the event happened); the AbstractView of that view is given in the event object's view attribute. [DOM3EVENTS]

A typical Web browser has one obvious view per browsing context: the browser's window (screen media). If a page is printed, however, a second view becomes evident, that of the print media. The two views always share the same underlying Document, but they have a different presentation of that document. A speech browser also establishes a browsing context, one with a view in the speech media.

A Document does not necessarily have a browsing context associated with it. In particular, data mining tools are likely to never instantiate browsing contexts.

The main view through which a user primarily interacts with a user agent is the default view.

The default view of a Document is given by the defaultView attribute on the Document object's DocumentView interface. [DOM3VIEWS]

When a browsing context is first created, it must be created with a single Document in its session history, whose address is about:blank, which is marked as being an HTML documents. The Document must have a single child html node, which itself has a single child body node. If the browsing context is created specifically to be immediately navigated, then that initial navigation will have replacement enabled.

4.1.1. Nested browsing contexts

Certain elements (for example, iframe elements) can instantiate further browsing contexts. These are called nested browsing contexts. If a browsing context P has an element in one of its Documents D that nests another browsing context C inside it, then P is said to be the parent browsing context of C, C is said to be a child browsing context of P, and C is said to be nested through D.

The browsing context with no parent browsing context is the top-level browsing context of all the browsing contexts nested within it (either directly or indirectly through other nested browsing contexts).

A Document is said to be fully active when it is the active document of its browsing context, and either its browsing context is a top-level browsing context, or the Document through which that browsing context is nested is itself fully active.

Because they are nested through an element, child browsing contexts are always tied to a specific Document in their parent browsing context. User agents must not allow the user to interact with child browsing contexts of elements that are in Documents that are not themselves fully active.

4.1.2. Auxiliary browsing contexts

It is possible to create new browsing contexts that are related to a top level browsing context without being nested through an element. Such browsing contexts are called auxiliary browsing contexts. Auxiliary browsing contexts are always top-level browsing contexts.

An auxiliary browsing context has an opener browsing context, which is the browsing context from which the auxiliary browsing context was created, and it has a furthest ancestor browsing context, which is the top-level browsing context of the opener browsing context when the auxiliary browsing context was created.

The opener DOM attribute on the Window object must return the Window object of the browsing context from which the current browsing context was created (its opener browsing context), if there is one and it is still available.

4.1.3. Secondary browsing contexts

User agents may support secondary browsing contexts, which are browsing contexts that form part of the user agent's interface, apart from the main content area.

4.1.4. Threads

Each browsing context is defined as having a list of zero or more directly reachable browsing contexts. These are:

All the browsing context's child browsing contexts.
The browsing context's parent browsing context.
All the browsing contexts that have the browsing context as their opener browsing context.
The browsing context's opener browsing context.

The transitive closure of all the browsing contexts that are directly reachable browsing contexts consists of a unit of related browsing contexts.

All the executable code in a unit of related browsing contexts must execute on a single conceptual thread. The dispatch of events fired by the user agent (e.g. in response to user actions or network activity) and the execution of any scripts associated with timers must be serialised so that for each unit of related browsing contexts there is only one script being executed at a time.

4.1.5. Browsing context names

Browsing contexts can be named.

A valid browsing context name is any string that does not start with a U+005F LOW LINE character, or, a string that case-insensitively matches one of: _self, _parent, or _top. (Names starting with an underscore are reserved for special keywords.)

The rules for chosing a browsing context given a browsing context name are as follows. The rules assume that they are being applied in the context of a browsing context.

If the given browsing context name is the empty string or _self, then the chosen browsing context must be the current one.
If the given browsing context name is _parent, then the chosen browsing context must be the parent browsing context of the current one, unless there isn't one, in which case the chosen browsing context must be the current browsing context.
If the given browsing context name is _top, then the chosen browsing context must be the most top-level browsing context of the current one.
If the given browsing context name is not _blank and there exists a browsing context whose name is the same as the given browsing context name, and the origin of that browsing context's active document is the same as the origin of the current browsing context's active document, and the user agent determines that the two browsing contexts are related enough that it is ok if they reach each other, then that browsing context must be the chosen one. If there are multiple matching browsing contexts, the user agent should select one in some arbitrary consistent manner, such as the most recently opened, most recently focused, or more closely related.
Otherwise, a new browsing context is being requested, and what happens depends on the user agent's configuration and/or abilities:

If the user agent has been configured such that in this instance it will create a new browsing context
A new auxiliary browsing context must be created, with the opener browsing context being the current one. If the given browsing context name is not _blank, then the new auxiliary browsing context's name must be the given browsing context name (otherwise, it has no name). The chosen browsing context must be this new browsing context. If it is immediately navigated, then the navigation will be done with replacement enabled.
If the user agent has been configured such that in this instance it will reuse the current browsing context
The chosen browsing context is the current browsing context.
If the user agent has been configured such that in this instance it will not find a browsing context
There must not be a chosen browsing context.

4.2. The default view

The AbstractView object of default views must also implement the Window object.

interface Window {
  // the current browsing context
  readonly attribute Window window;
  readonly attribute Window self;
  readonly attribute History history;
  readonly attribute UndoManager undoManager;
  Selection getSelection();

  // the user agent
  readonly attribute ClientInformation navigator; 
  readonly attribute Storage sessionStorage;
  readonly attribute StorageList globalStorage;

  // modal user prompts
  void alert(in DOMString message);
  boolean confirm(in DOMString message);
  DOMString prompt(in DOMString message);
  DOMString prompt(in DOMString message, in DOMString default);

  // other browsing contexts
  readonly attribute Window frames;
  readonly attribute unsigned long length;
  readonly attribute Window opener;
  Window open();
  Window open(in DOMString url);
  Window open(in DOMString url, in DOMString target);
  Window open(in DOMString url, in DOMString target, in DOMString features);
  Window open(in DOMString url, in DOMString target, in DOMString features, in DOMString replace);


  // event handler DOM attributes
           attribute EventListener onabort;
           attribute EventListener onbeforeunload;
           attribute EventListener onblur;
           attribute EventListener onchange;
           attribute EventListener onclick;
           attribute EventListener oncontextmenu;
           attribute EventListener ondblclick;
           attribute EventListener ondrag;
           attribute EventListener ondragend;
           attribute EventListener ondragenter;
           attribute EventListener ondragleave;
           attribute EventListener ondragover;
           attribute EventListener ondragstart;
           attribute EventListener ondrop;
           attribute EventListener onerror;
           attribute EventListener onfocus;
           attribute EventListener onkeydown;
           attribute EventListener onkeypress;
           attribute EventListener onkeyup;
           attribute EventListener onload;
           attribute EventListener onmessage;
           attribute EventListener onmousedown;
           attribute EventListener onmousemove;
           attribute EventListener onmouseout;
           attribute EventListener onmouseover;
           attribute EventListener onmouseup;
           attribute EventListener onmousewheel;
           attribute EventListener onresize;
           attribute EventListener onscroll;
           attribute EventListener onselect;
           attribute EventListener onsubmit;
           attribute EventListener onunload;
};

The window, frames, self DOM attributes must all return the Window object itself.

The Window object also provides the scope for script execution. Each Document in a browsing context has an associated list of added properties which, when a document is active, are available on the Document's default view Window object. A Document object's list of added properties must be empty when the Document object is created.

Objects implementing the Window interface must also implement the EventTarget interface.

Window objects also have an implicit [[Get]] method which returns nested browsing contexts.

4.2.1. Security

User agents must raise a security exception whenever any of the members of a Window object are accessed by scripts whose origin is not the same as the Window object's browsing context's active document's origin, with the following exceptions:

The location object

User agents must not allow scripts to override the location object's setter.

4.2.2. Constructors

All Window objects must provide the following constructors:

Audio()
Audio(src): When invoked as constructors, these must return a new HTMLAudioElement object (a new audio element). If the src argument is present, the object created must have its src content attribute set to the provided value.
Image()
Image(in unsigned long w)
Image(in unsigned long w, in unsigned long h): When invoked as corstructors, these must return a new HTMLImageElement object (a new img element). If the h argument is present, the new object's height content attribute must be set to h. If the w argument is present, the new object's width content attribute must be set to w.
Option()
Option(in DOMString name)
Option(in DOMString name, in DOMString value): When invoked as constructors, these must return a new HTMLOptionElement object (a new option element). need to define argument processing

And when constructors are invoked but without using the constructor syntax...?

4.2.3. APIs for creating and navigating browsing contexts by name

The open() method on Window objects provides a mechanism for navigating an existing browsing context or opening and navigating an auxiliary browsing context.

The method has four arguments, though they are all optional.

The first argument, url, gives a URI (or IRI) for a page to load in the browsing context. If no arguments are provided, then the url argument defaults to "about:blank". The argument must be resolved to an absolute URI by ...

The second argument, target, specifies the name of the browsing context that is to be navigated. It must be a valid browsing context name. If fewer than two arguments are provided, then the name argument defaults to the value "_blank".

The third argument, features, has no effect and is supported for historical reasons only.

The fourth argument, replace, specifies whether or not the new page will replace the page currently loaded in the browsing context, when target identifies an existing browsing context (as opposed to leaving the current page in the browsing context's session history). When three or fewer arguments are provided, replace defaults to false.

When the method is invoked, the user agent must first select a browsing context to navigate by applying the rules for chosing a browsing context given a browsing context name using the target argument as the name and the browsing context of the script as the context in which the algorithm is executed, unless the user has indicated a preference, in which case the browsing context to navigate may instead be the one indicated by the user.

For example, suppose there is a user agent that supports control-clicking a link to open it in a new tab. If a user clicks in that user agent on an element whose onclick handler uses the window.open() API to open a page in an iframe, but, while doing so, holds the control key down, the user agent could override the selection of the target browsing context to instead target a new tab.

Then, the user agent must navigate the selected browsing context to the URI given in url. If the replace is true, then replacement must be enabled; otherwise, it must not be enabled unless the browsing context was just created as part of the the rules for chosing a browsing context given a browsing context name.

4.2.4. Accessing other browsing contexts

In ECMAScript implementations, objects that implement the Window interface must have a [[Get]] method that, when invoked with a property name that is a number i, returns the ith child browsing context of the active Document, sorted in document order of the elements nesting those browsing contexts.

The length DOM attribute on the Window interface must return the number of child browsing contexts of the active Document.

4.3. Session history and navigation

4.3.1. The session history of browsing contexts

The sequence of Documents in a browsing context is its session history.

History objects provide a representation of the pages in the session history of browsing contexts. Each browsing context has a distinct session history.

Each Document object in a browsing context's session history is associated with a unique instance of the History object, although they all must model the same underlying session history.

The history attribute of the Window interface must return the object implementing the History interface for that Window object's active document.

History objects represent their browsing context's session history as a flat list of URIs and state objects. (This does not imply that the UI need be linear. See the notes below.)

Typically, the history list will consist of only URIs. However, a page can add state objects between its entry in the session history and the next ("forward") entry. These are then returned to the script when the user (or script) goes back in the history, thus enabling authors to use the "navigation" metaphor even in one-page applications.

Entries that consist of state objects share the same Document as the entry for the URI itself. Contiguous entries that differ just by fragment identifier must also share the same Document.

All entries that share the same Document (and that are therefore merely different states of one particular document) are contiguous by definition.

At any point, one of the entries in the session history is the current entry. This is the entry representing the active document of the browsing context. The current entry is usually an entry for the location of the Document. However, it can also be one of the entries for state objects added to the history by that document.

User agents may discard the DOMs of entries other than the current entry that are not referenced from any script, reloading the pages afresh when the user or script navigates back to such pages. This specification does not specify when user agents should discard pages' DOMs and when they should cache them. See the section on the load and unload events for more details.

Entries that have had their DOM discarded must, for the purposes of the algorithms given below, act as if they had not. When the user or script navigates back or forwards to a page which has no in-memory DOM objects, any other entries that shared the same Document object with it must share the new object as well.

When a user agent discards the DOM from an entry in the session history, it must also discard all the entries from the first state object entry for that Document object up to and including the last entry for that Document object (including any non-state-object entries in that range, such as entries where the user navigated using fragment identifiers). These entries are not recreated if the user or script navigates back to the page. If there are no state object entries for that Document object then no entries are removed.

4.3.2. The `History` interface

interface History {
  readonly attribute long length;
  void go(in long delta);
  void go();
  void back();
  void forward();
  void pushState(in DOMObject data);
  void clearState();
};

The length attribute of the History interface must return the number of entries in this session history.

The actual entries are not accessible from script.

The go(delta) method causes the UA to move the number of steps specified by delta in the session history.

If the index of the current entry plus delta is less than zero or greater than or equal to the number of items in the session history, then the user agent must do nothing.

If the delta is zero, then the user agent must act as if the location.reload() method was called instead.

Otherwise, the user agent must cause the current browsing context to traverse the history to the specified entry, as described below. The specified entry is the one whose index equals the index of the current entry plus delta.

When a user agent is required to traverse the history to an entry target, the user agent must act as follows:

If there is no longer a Document object for the entry in question, the user agent must navigate the browsing context to the location for that entry to preform an entry update of that entry, and abort these steps. The "navigate" algorithm reinvokes this "traverse" algorithm to complete the traversal, at which point there is a Document object and so this step gets skipped.
If appropriate, update the current entry in the browsing context's Document object's History object to reflect any state that the user agent wishes to persist.

For example, some user agents might want to persist the scroll position, or the values of form controls.
If there are any entries with state objects between the current entry and the target entry (not inclusive), then the user agent must iterate through every entry between the current entry and the target entry, starting with the entry closest to the current entry, and ending with the one closest to the target entry. For each entry, if the entry is a state object, the user agent must activate the state object.
If the specified entry has a different Document object than the current entry then the user agent must follow the following substeps:
1. The user agent must move any properties that have been added to the browsing context's default view's Window object to the active document's Document's list of added properties.
2. The user agent must make the specified entry's Document object the active document of the browsing context.
3. The user agent must move any properties that have been added to the active document's Document's list of added properties to browsing context's default view's Window object.
If the specified entry is a state object, the user agent must activate that state object.
User agents may also update other aspects of the document view when the location changes in this way, for instance the scroll position, values of form fields, etc.

how does the changing of the global attributes affect .watch() when seen from other Windows?

When the user navigates through a browsing context, e.g. using a browser's back and forward buttons, the user agent must translate this action into the equivalent invocations of the history.go(delta) method on the various affected window objects.

Some of the other members of the History interface are defined in terms of the go() method, as follows:

Member	Definition
`go()`	Must do the same as `go(0)`
`back()`	Must do the same as `go(-1)`
`forward()`	Must do the same as `go(1)`

The pushState(data) method adds a state object to the history.

When this method is invoked, the user agent must first remove from the session history any entries for that Document from the entry after the current entry up to the last entry in the session history that references the same Document object, if any. If the current entry is the last entry in the session history, or if there are no entries after the current entry that reference the same Document object, then no entries are removed.

Then, the user agent must add a state object entry to the session history, after the current entry, with the specified data as the state object.

Finally, the user agent must update the current entry to be the this newly added entry.

There has been a suggestion that pushState() should take a URI and a string; the URI to allow for the page to be bookmarked, and the string to allow the UA to give the page a meaningful title in the history state, if it shows history state.

User agents may limit the number of state objects added to the session history per page. If a page hits the UA-defined limit, user agents must remove the entry immediately after the first entry for that Document object in the session history after having added the new entry. (Thus the state history acts as a FIFO buffer for eviction, but as a LIFO buffer for navigation.)

The clearState() method removes all the state objects for the Document object from the session history.

When this method is invoked, the user agent must remove from the session history all the entries from the first state object entry for that Document object up to the last entry that references that same Document object, if any.

Then, if the current entry was removed in the previous step, the current entry must be set to the last entry for that Document object in the session history.

4.3.3. Activating state objects

When a state object in the session history is activated (which happens in the cases described above), the user agent must fire a popstate event in no namespace on the the body element using the PopStateEvent interface, with the state object in the state attribute. This event bubbles but is not cancelable and has no default action.

interface PopStateEvent : Event {
  readonly attribute DOMObject state;
  void initPopStateEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMObject statetArg);
  void initPopStateEventNS(in DOMString namespaceURIArg, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMObject stateArg);
};

The initPopStateEvent() and initPopStateEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The state attribute represents the context information for the event.

Should we coalesce these events if they occur while the page is away? (e.g. during traversal -- see above)

4.3.4. The `Location` interface

Each Document object in a browsing context's session history is associated with a unique instance of a Location object.

The location attribute of the Window interface must return the Location object for that Window object's active document.

For historical reasons, the location attribute of the HTMLDocument interface must return the same object as the location attribute on its associated Window object.

Location objects provide a representation of the URI of their document, and allow the current entry of the browsing context's session history to be changed, by adding or replacing entries in the history object.

interface Location {
  readonly attribute DOMString href;
  void assign(in DOMString url);
  void replace(in DOMString url);
  void reload();

  // URI decomposition attributes 
  readonly attribute DOMString protocol;
  readonly attribute DOMString host;
  readonly attribute DOMString hostname;
  readonly attribute DOMString port;
  readonly attribute DOMString pathname;
  readonly attribute DOMString search;
  readonly attribute DOMString hash;
};

In the ECMAScript DOM binding, objects implementing this interface must stringify to the same value as the href attribute.

In the ECMAScript DOM binding, the location members of the HTMLDocument and Window interfaces behave as if they had a setter: user agents must treats attempts to set these location attribute as attempts at setting the href attribute of the relevant Location object instead.

The href attribute returns the address of the page represented by the associated Document object, as an absolute IRI reference.

On setting, the user agent must act as if the assign() method had been called with the new value as its argument.

When the assign(url) method is invoked, the UA must navigate the browsing context to the specified url.

When the replace(url) method is invoked, the UA must navigate to the specified url with replacement enabled.

Relative url arguments for assign() and replace() must be resolved relative to the base URI of the script that made the method call.

The Location interface also has the complement of URI decomposition attributes, protocol, host, port, hostname, pathname, search, and hash. These must follow the rules given for URI decomposition attributes, with the input being the address of the page represented by the associated Document object, as an absolute IRI reference (same as the href attribute), and the common setter action being the same as setting the href attribute to the new output value.

4.3.4.1. Security

User agents must raise a security exception whenever any of the members of a Location object are accessed by scripts whose origin is not the same as the Location object's associated Document's origin, with the following exceptions:

The href setter

User agents must not allow scripts to override the href attribute's setter.

4.3.5. Implementation notes for session history

This section is non-normative.

The History interface is not meant to place restrictions on how implementations represent the session history to the user.

For example, session history could be implemented in a tree-like manner, with each page having multiple "forward" pages. This specification doesn't define how the linear list of pages in the history object are derived from the actual session history as seen from the user's perspective.

Similarly, a page containing two iframes has a history object distinct from the iframes' history objects, despite the fact that typical Web browsers present the user with just one "Back" button, with a session history that interleaves the navigation of the two inner frames and the outer page.

Security: It is suggested that to avoid letting a page "hijack" the history navigation facilities of a UA by abusing pushState(), the UA provide the user with a way to jump back to the previous page (rather than just going back to the previous state). For example, the back button could have a drop down showing just the pages in the session history, and not showing any of the states. Similarly, an aural browser could have two "back" commands, one that goes back to the previous state, and one that jumps straight back to the previous page.

In addition, a user agent could ignore calls to pushState() that are invoked on a timer, or from event handlers that do not represent a clear user action, or that are invoked in rapid succession.

4.4. Links

4.4.1. Hyperlink elements

The a, area, and link elements can, in certain situations described in the definitions of those elements, represent hyperlinks.

The href attribute on a hyperlink element must have a value that is a URI (or IRI). This URI is the destination resource of the hyperlink.

The href attribute on a and area elements is not required; when those elements do not have href attributes they do not represent hyperlinks.

The href attribute on the link element is required, but whether a link element represents a hyperlink or not depends on the value of the rel attribute of that element.

The target attribute, if present, must be a valid browsing context name. User agents use this name when following hyperlinks.

The ping attribute, if present, gives the URIs of the resources that are interested in being notified if the user follows the hyperlink. The value must be a space separated list of one or more URIs (or IRIs). The value is used by the user agent when following hyperlinks.

For a and area elements that represent hyperlinks, the relationship between the document containing the hyperlink and the destination resource indicated by the hyperlink is given by the value of the element's rel attribute, which must be an unordered set of space-separated tokens. The allowed values and their meanings are defined below. The rel attribute has no default value. If the attribute is omitted or if none of the values in the attribute are recognised by the UA, then the document has no particular relationship with the destination resource other than there being a hyperlink between the two.

The media attribute describes for which media the target document was designed. It is purely advisory. The value must be a valid media query. [MQ] The default, if the media attribute is omitted or has an invalid value, is all.

The hreflang attribute on hyperlink elements, if present, gives the language of the linked resource. It is purely advisory. The value must be a valid RFC 3066 language code. [RFC3066] User agents must not consider this attribute authoritative — upon fetching the resource, user agents must only use language information associated with the resource to determine its language, not metadata included in the link to the resource.

The type attribute, if present, gives the MIME type of the linked resource. It is purely advisory. The value must be a valid MIME type, optionally with parameters. [RFC2046] User agents must not consider the type attribute authoritative — upon fetching the resource, user agents must not use metadata included in the link to the resource to determine its type.

4.4.2. Following hyperlinks

When a user follows a hyperlink, the user agent must navigate a browsing context to the URI of the hyperlink.

The URI of the hyperlink is URI given by resolving the the href attribute of that hyperlink relative to the hyperlink's element. In the case of server-side image maps, the URI of the hyperlink must further have its hyperlink suffix appended to it.

If the user indicated a specific browsing context when following the hyperlink, or if the user agent is configured to follow hyperlinks by navigating a particular browsing context, then that must be the browsing context that is navigated.

Otherwise, if the hyperlink element is an a or area element that has a target attribute, then the browsing context that is navigated must be chosen by applying the rules for chosing a browsing context given a browsing context name, using the value of the target attribute as the browsing context name. If these rules result in the creation of a new browsing context, it must be navigated with replacement enabled.

Otherwise, if the hyperlink element is a sidebar hyperlink and the user agent implements a feature that can be considered a secondary browsing context, such a secondary browsing context may be selected as the browsing context to be navigated.

Otherwise, if the hyperlink element is an a or area element with no target attribute, but there is a base element with a target attribute in the document, then the browsing context that is navigated must be chosen by applying the rules for chosing a browsing context given a browsing context name, using the value of the target attribute of the first such base element as the browsing context name. If these rules result in the creation of a new browsing context, it must be navigated with replacement enabled.

Otherwise, the browsing context that must be navigated is the same browsing context as the one which the hyperlink element itself is in.

4.4.2.1. Hyperlink auditing

If an a or area hyperlink element has a ping attribute and the user follows the hyperlink, the user agent must take the ping attribute's value, strip leading and trailing spaces, split the value on sequences of spaces, treat each resulting part as a URI (resolving relative URIs according to element's base URI) and then should send a request to each of the resulting URIs. This may be done in parallel with the primary request, and is independent of the result of that request.

User agents should allow the user to adjust this behaviour, for example in conjunction with a setting that disables the sending of HTTP Referrer headers. Based on the user's preferences, UAs may either ignore the ping attribute altogether, or selectively ignore URIs in the list (e.g. ignoring any third-party URIs).

For URIs that are HTTP URIs, the requests must be performed using the POST method (with an empty entity body in the request). User agents must ignore any entity bodies returned in the responses, but must, unless otherwise specified by the user, honour the HTTP headers — in particular, HTTP cookie headers. [RFC2965]

To save bandwidth, implementors might wish to consider omitting optional headers such as Accept from these requests.

When the ping attribute is present, user agents should clearly indicate to the user that following the hyperlink will also cause secondary requests to be sent in the background, possibly including listing the actual target URIs.

The ping attribute is redundant with pre-existing technologies like HTTP redirects and JavaScript in allowing Web pages to track which off-site links are most popular or allowing advertisers to track click-through rates.

However, the ping attribute provides these advantages to the user over those alternatives:

It allows the user to see the final target URI unobscured.
It allows the UA to inform the user about the out-of-band notifications.
It allows the paranoid user to disable the notifications without losing the underlying link functionality.
It allows the UA to optimise the use of available network bandwidth so that the target page loads faster.

Thus, while it is possible to track users without this feature, authors are encouraged to use the ping attribute so that the user agent can improve the user experience.

4.4.3. Link types

The following table summarises the link types that are defined by this specification. This table is non-normative; the actual definitions for the link types are given in the next few sections.

In this section, the term referenced document refers to the resource identified by the element representing the link, and the term current document refers to the resource within which the element representing the link finds itself.

To determine which link types apply to a link, a, or area element, the element's rel attribute must be split on spaces. The resulting tokens are the link types that apply to that element.

Link type	Effect on...		Brief description
Link type	`link`	`a` and `area`	Brief description
`alternate`	Hyperlink	Hyperlink	Gives alternate representations of the current document.
`archives`	Hyperlink	Hyperlink	Provides a link to a collection of records, documents, or other materials of historical interest.
`author`	Hyperlink	Hyperlink	Gives a link to the current document's author.
`bookmark`	not allowed	Hyperlink	Gives the permalink for the nearest ancestor section.
`contact`	Hyperlink	Hyperlink	Gives a link to contact information for the current document.
`external`	not allowed	Hyperlink	Indicates that the referenced document is not part of the same site as the current document.
`feed`	Hyperlink	Hyperlink	Gives the address of a syndication feed for the current document.
`first`	Hyperlink	Hyperlink	Indicates that the current document is a part of a series, and that the first document in the series is the referenced document.
`help`	Hyperlink	Hyperlink	Provides a link to context-sensitive help.
`icon`	External Resource	not allowed	Imports an icon to represent the current document.
`index`	Hyperlink	Hyperlink	Gives a link to the document that provides a table of contents or index listing the current document.
`last`	Hyperlink	Hyperlink	Indicates that the current document is a part of a series, and that the last document in the series is the referenced document.
`license`	Hyperlink	Hyperlink	Indicates that the current document is covered by the copyright license described by the referenced document.
`next`	Hyperlink	Hyperlink	Indicates that the current document is a part of a series, and that the next document in the series is the referenced document.
`nofollow`	not allowed	Hyperlink	Indicates that the current document's original author or publisher does not endorse the referenced document.
`pingback`	External Resource	not allowed	Gives the address of the pingback server that handles pingbacks to the current document.
`prefetch`	External Resource	not allowed	Specifies that the target resource should be pre-emptively cached.
`prev`	Hyperlink	Hyperlink	Indicates that the current document is a part of a series, and that the previous document in the series is the referenced document.
`search`	Hyperlink	Hyperlink	Gives a link to a resource that can be used to search through the current document and its related pages.
`stylesheet`	External Resource	not allowed	Imports a stylesheet.
`sidebar`	Hyperlink	Hyperlink	Specifies that the referenced document, if retrieved, is intended to be shown in the browser's sidebar (if it has one).
`tag`	Hyperlink	Hyperlink	Gives a tag (identified by the given address) that applies to the current document.
`up`	Hyperlink	Hyperlink	Provides a link to a document giving the context for the current document.

Some of the types described below list synonyms for these values. These are to be handled as specified by user agents, but must not be used in documents.

4.4.3.1. Link type "`alternate`"

The alternate keyword may be used with link, a, and area elements. For link elements, if the rel attribute does not also contain the keyword stylesheet, it creates a hyperlink; but if it does also contains the keyword stylesheet, the alternate keyword instead modifies the meaning of the stylesheet keyword in the way described for that keyword, and the rest of this subsection doesn't apply.

The alternate keyword indicates that the referenced document is an alternate representation of the current document.

The nature of the referenced document is given by the media, hreflang, and type attributes.

If the alternate keyword is used with the media attribute, it indicates that the referenced document is intended for use with the media specified.

If the alternate keyword is used with the hreflang attribute, and that attribute's value differs from the root element's language, it indicates that the referenced document is a translation.

If the alternate keyword is used with the type attribute, it indicates that the referenced document is a reformulation of the current document in the specified format.

The media, hreflang, and type attributes can be combined when specified with the alternate keyword.

For example, the following link is a French translation that uses the PDF format:

<link rel=alternate type=application/pdf hreflang=fr href=manual-fr>

If the alternate keyword is used with the type attribute set to the value application/rss+xml or the value application/atom+xml, then the user agent must treat the link as it would if it had the feed keyword specified as well.

The alternate link relationship is transitive — that is, if a document links to two other documents with the link type "alternate", then, in addition to implying that those documents are alternative representations of the first document, it is also implying that those two documents are alternative representations of each other.

4.4.3.2. Link type "`archives`"

The archives keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The archives keyword indicates that the referenced document describes a collection of records, documents, or other materials of historical interest.

A blog's index page could link to an index of the blog's past posts with rel="archives".

Synonyms: For historical reasons, user agents must also treat the keyword "archive" like the archives keyword.

4.4.3.3. Link type "`author`"

The author keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

For a and area elements, the author keyword indicates that the referenced document provides further information about the author of the section that the element defining the hyperlink applies to.

For link elements, the author keyword indicates that the referenced document provides further information about the author for the page as a whole.

The "referenced document" can be, and often is, a mailto: URI giving the e-mail address of the author. [MAILTO]

Synonyms: For historical reasons, user agents must also treat link, a, and area elements that have a rev attribute with the value "made" as having the author keyword specified as a link relationship.

4.4.3.4. Link type "`bookmark`"

The bookmark keyword may be used with a and area elements.

The bookmark keyword gives a permalink for the nearest ancestor article element of the linking element in question, or of the section the linking element is most closely associated with, if there are no ancestor article elements.

The following snippet has three permalinks. A user agent could determine which permalink applies to which part of the spec by looking at where the permalinks are given.

 ...
 <body>
  <h1>Example of permalinks</h1>
  <div id="a">
   <h2>First example</h2>
   <p><a href="a.html" rel="bookmark">This</a> permalink applies to
   only the content from the first H2 to the second H2. The DIV isn't
   exactly that section, but it roughly corresponds to it.</p>
  </div>
  <h2>Second example</h2>
  <article id="b">
   <p><a href="b.html" rel="bookmark">This</a> permalink applies to
   the outer ARTICLE element (which could be, e.g., a blog post).</p>
   <article id="c">
    <p><a href="c.html" rel="bookmark">This</a> permalink applies to
    the inner ARTICLE element (which could be, e.g., a blog comment).</p>
   </article>
  </article>
 </body>
 ...

4.4.3.5. Link type "`contact`"

The contact keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

For a and area elements, the contact keyword indicates that the referenced document provides further contact information for the section that the element defining the hyperlink applies to.

User agents must treat any hyperlink in an address element as having the contact link type specified.

For link elements, the contact keyword indicates that the referenced document provides further contact information for the page as a whole.

4.4.3.6. Link type "`external`"

The external keyword may be used with a and area elements.

The external keyword indicates that the link is leading to a document that is not part of the site that the current document forms a part of.

4.4.3.7. Link type "`feed`"

The feed keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The feed keyword indicates that the referenced document is a syndication feed. If the alternate link type is also specified, then the feed is specifically the feed for the current document; otherwise, the feed is just a syndication feed, not necessarily associated with a particular Web page.

The first link, a, or area element in the document (in tree order) that creates a hyperlink with the link type feed must be treated as the default syndication feed for the purposes of feed autodiscovery.

The feed keyword is implied by the alternate link type in certain cases (q.v.).

The following two link elements are equivalent: both give the syndication feed for the current page:

<link rel="alternate" type="application/atom+xml" href="data.xml">

<link rel="feed alternate" href="data.xml">

The following extract offers various different syndication feeds:

 <p>You can access the planets database using Atom feeds:</p>
 <ul>
  <li><a href="recently-visited-planets.xml" rel="feed">Recently Visited Planets</a></li>
  <li><a href="known-bad-planets.xml" rel="feed">Known Bad Planets</a></li>
  <li><a href="unexplored-planets.xml" rel="feed">Unexplored Planets</a></li>
 </ul>

4.4.3.8. Link type "`help`"

The help keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

For a and area elements, the help keyword indicates that the referenced document provides further help information for the parent of the element defining the hyperlink, and its children.

In the following example, the form control has associated context-sensitive help. The user agent could use this information, for example, displaying the referenced document if the user presses the "Help" or "F1" key.

 <p><label> Topic: <input name=topic> <a href="help/topic.html" rel="help">(Help)</a></label></p>

For link elements, the help keyword indicates that the referenced document provides help for the page as a whole.

4.4.3.9. Link type "`icon`"

The icon keyword may be used with link elements, for which it creates an external resource link.

The specified resource is an icon representing the page or site, and should be used by the user agent when representing the page in the user interface.

Icons could be auditory icons, visual icons, or other kinds of icons. If multiple icons are provided, the user agent must select the most appropriate icon according to the media attribute.

4.4.3.10. Link type "`license`"

The license keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The license keyword indicates that the referenced document provides the copyright license terms under which the current document is provided.

Synonyms: For historical reasons, user agents must also treat the keyword "copyright" like the license keyword.

4.4.3.11. Link type "`nofollow`"

The nofollow keyword may be used with a and area elements.

The nofollow keyword indicates that the link is not endorsed by the original author or publisher of the page.

4.4.3.12. Link type "`pingback`"

The pingback keyword may be used with link elements, for which it creates an external resource link.

For the semantics of the pingback keyword, see the Pingback 1.0 specification. [PINGBACK]

4.4.3.13. Link type "`prefetch`"

The prefetch keyword may be used with link elements, for which it creates an external resource link.

The prefetch keyword indicates that preemptively fetching and caching the specified resource is likely to be beneficial, as it is highly likely that the user will require this resource.

4.4.3.14. Link type "`search`"

The search keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The search keyword indicates that the referenced document provides an interface specifically for searching the document and its related resources.

OpenSearch description documents can be used with link elements and the search link type to enable user agents to autodiscover search interfaces.

4.4.3.15. Link type "`stylesheet`"

The stylesheet keyword may be used with link elements, for which it creates an external resource link that contributes to the styling processing model.

The specified resource is a resource that describes how to present the document. Exactly how the resource is to be processed depends on the actual type of the resource.

If the alternate keyword is also specified on the link element, then the link is an alternative stylesheet.

4.4.3.16. Link type "`sidebar`"

The sidebar keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The sidebar keyword indicates that the referenced document, if retrieved, is intended to be shown in a secondary browsing context (if possible), instead of in the current browsing context.

A hyperlink element with with the sidebar keyword specified is a sidebar hyperlink.

4.4.3.17. Link type "`tag`"

The tag keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The tag keyword indicates that the tag that the referenced document represents applies to the current document.

4.4.3.18. Hierarchical link types

Some documents form part of a hierarchical structure of documents.

A hierarchical structure of documents is one where each document can have various subdocuments. A subdocument is said to be a child of the document it is a subdocument of. The document of which it is a subdocument is said to be its parent. The children of a document have a relative order; the subdocument that precedes another is its previous sibling, and the one that follows it is its next sibling. A document with no parent forms the top of the hierarchy.

4.4.3.18.1. Link type "`first`"

The first keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The first keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the first child of the current document's parent document.

Synonyms: For historical reasons, user agents must also treat the keywords "begin" and "start" like the first keyword.

4.4.3.18.2. Link type "`index`"

The index keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The index keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the top of the hierarchy.

Synonyms: For historical reasons, user agents must also treat the keywords "top", "contents", and "toc" like the index keyword.

4.4.3.18.3. Link type "`last`"

The last keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The last keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the last child of the current document's parent document.

Synonyms: For historical reasons, user agents must also treat the keyword "end" like the last keyword.

4.4.3.18.4. Link type "`next`"

The next keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The next keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the next sibling of the current document.

4.4.3.18.5. Link type "`prev`"

The prev keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The prev keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the previous sibling of the current document.

Synonyms: For historical reasons, user agents must also treat the keyword "previous" like the prev keyword.

4.4.3.18.6. Link type "`up`"

The up keyword may be used with link, a, and area elements. For link elements, it creates a hyperlink.

The up keyword indicates that the document is part of a hierarchical structure, and that the link is leading to the document that is the parent of the current document.

4.4.3.19. Other link types

Other than the types defined above, only types defined as extensions in the WHATWG Wiki RelExtensions page may be used with the rel attribute on link, a, and area elements. [WHATWGWIKI]

Anyone is free to edit the WHATWG Wiki RelExtensions page at any time to add a type. Extension types must be specified with the following information:

Keyword

The actual value being defined. The value should not be confusingly similar to any other defined value (e.g. differing only in case).

Effect on... link

One of the following:

not allowed: The keyword is not allowed to be specified on link elements.
Hyperlink: The keyword may be specified on a link element; it creates a hyperlink link.
External Resource: The keyword may be specified on a link element; it creates a external resource link.

Effect on... a and area

One of the following:

not allowed: The keyword is not allowed to be specified on a and area elements.
Hyperlink: The keyword may be specified on a and area elements; it creates a hyperlink.

Brief description

A short description of what the keyword's meaning is.

Link to more details

A link to a more detailed description of the keyword's semantics and requirements. It could be another page on the Wiki, or a link to an external page.

Synonyms

A list of other keyword values that have exactly the same processing requirements. Authors must not use the values defined to be synonyms, they are only intended to allow user agents to support legacy content.

Status

One of the following:

Proposal: The keyword has not received wide peer review and approval. It is included for completeness because pages use the keyword. Pages should not use the keyword.
Accepted: The keyword has received wide peer review and approval. It has a specification that unambiguously defines how to handle pages that use the keyword, including when they use them in incorrect ways. Pages may use the keyword.
Rejected: The keyword has received wide peer review and it has been found to have significant problems. Pages must not use the keyword. When a keyword has this status, the "Effect on... link" and "Effect on... a and area" information should be set to "not allowed".

If a keyword is added with the "proposal" status and found to be redundant with existing values, it should be removed and listed as a synonym for the existing value. If a keyword is added with the "proposal" status and found to be harmful, then it should be changed to "rejected" status, and its "Effect on..." information should be changed accordingly.

Conformance checkers must use the information given on the WHATWG Wiki RelExtensions page to establish if a value not explicitly defined in this specification is allowed or not. When an author uses a new type not defined by either this specification or the Wiki page, conformance checkers should offer to add the value to the Wiki, with the details described above, with the "proposal" status.

This specification does not define how new values will get approved. It is expected that the Wiki will have a community that addresses this.

4.5. Interfaces for URI manipulation

An interface that has a complement of URI decomposition attributes will have seven attributes with the following definitions:

  readonly attribute DOMString protocol;
  readonly attribute DOMString host;
  readonly attribute DOMString hostname;
  readonly attribute DOMString port;
  readonly attribute DOMString pathname;
  readonly attribute DOMString search;
  readonly attribute DOMString hash;

The attributes defined to be URI decomposition attributes must act as described for the attributes with the same corresponding names in this section.

In addition, an interface with a complement of URI decomposition attributes will define an input, which is a URI that the attributes act on, and a common setter action, which is a set of steps invoked when any of the attributes' setters are invoked.

The seven URI decomposition attributes have similar requirements.

On getting, if the input fulfills the condition given in the "getter condition" column corresponding to the attribute in the table below, the user agent must return the part of the input URI given in the "component" column, with any prefixes specified in the "prefix" column appropriately added to the start of the string and any suffixes specified in the "suffix" column appropriately added to the end of the string. Otherwise, the attribute must return the empty string.

On setting, the new value must first be mutated as described by the "setter preprocessor" column, then mutated by %-escaping any characters in the new value that are not valid in the relevant component as given by the "component" column. Then, if the resulting new value fulfills the condition given in the "setter condition" column, the user agent must make a new string output by replacing the component of the URI given by the "component" column in the input URI with the new value; otherwise, the user agent must let output be equal to the input. Finally, the user agent must invoke the common setter action with the value of output.

The rules for parsing and constructing URIs are described in RFC 2396 as modified by RFC 2732. [RFC2396] [RFC2732]

Attribute	Component	Getter Condition	Prefix	Suffix	Setter Preprocessor	Setter Condition
`protocol`	<scheme>	—	—	U+003A COLON ("`:`")	Remove all trailing U+003A COLON ("`:`") characters	The new value is not the empty string
`host`	<hostport>	input is hierarchical and uses a server-based naming authority	—	—	—	—
`hostname`	<host>	input is hierarchical and uses a server-based naming authority	—	—	Remove all leading U+002F SOLIDUS ("`/`") characters	—
`port`	<port>	input is hierarchical and uses a server-based naming authority	—	—	Remove any characters in the new value that are not in the range U+0030 DIGIT ZERO .. U+0039 DIGIT NINE	The new value is not the empty string
`pathname`	<abs_path>	input is hierarchical	—	—	If it has no leading U+002F SOLIDUS ("`/`") character, prepend a U+002F SOLIDUS ("`/`") character to the new value	—
`search`	<query>	input is hierarchical	U+003F QUESTION MARK ("`?`")	—	Remove one leading U+003F QUESTION MARK ("`?`") character, if any	—
`hash`	<fragment>	Fragment identifier is longer than zero characters	U+0023 NUMBER SIGN ("`#`")	—	Remove one leading U+0023 NUMBER SIGN ("`#`") character, if any	—

4.6. Navigating across documents

Certain actions cause the browsing context to navigate. For example, following a hyperlink, form submission, and the window.open() and location.assign() methods can all cause a browsing context to navigate. A user agent may also provide various ways for the user to explicitly cause a browsing context to navigate.

When a browsing context is navigated, the user agent must follow the following steps:

Cancel any preexisting attempt to navigate the browsing context.
If the new resource is the same as the current resource, but a fragment identifier has been specified, then scroll the document to the specified element and abort these steps.
If the new resource is to be handled by displaying some sort of inline content, e.g. an error message because the specified scheme is not one of the supported protocols, or an inline prompt to allow the user to select a registered handler for the given scheme, then display the inline content and abort these steps.
If the new resource is to be handled using a mechanism that does not affect the browsing context, then abort these steps and proceed with that mechanism instead.
Start fetching the specified resource, as appropriate (e.g. performing an HTTP GET or POST operation, or reading the file from disk, or executing script in the case of a javascript: URI). If this results in a redirect, return to step 2 with the new resource.
Wait for one or more bytes to be available or for the user agent to establish that the resource in question is empty. During this time, the user agent may allow the user to cancel this navigation attempt or start other navigation attempts.
If the document's out-of-band metadata (e.g. HTTP headers), not counting any type information (such as the Content-Type HTTP header), requires some sort of processing that will not affect the browsing context, then perform that processing and abort these steps.
Such processing might be triggered by, amongst other things, the following:
- HTTP status codes (e.g. 204 No Content or 205 Reset Content)
- HTTP Content-Disposition headers
- Network errors
Let type be the sniffed type of the resource.
If the user agent has been configured to process resources of the given type using some mechanism other than rendering the content in a browsing context, then skip this step. Otherwise, if the type is one of the following types, jump to the appropriate entry in the following list, and process the resource as described there:

"text/html"
Follow the steps given in the HTML document section, and abort these steps.
Any type ending in "+xml"
"application/xml"
"text/xml"
Follow the steps given in the XML document section. If that section determines that the content is not to be displayed as a generic XML document, then proceed to the next step in this overall set of steps. Otherwise, abort these steps.
"text/plain"
Follow the steps given in the plain text file section, and abort these steps.
A supported image type
Follow the steps given in the image section, and abort these steps.
A type that will use an external application to render the content in the browsing context
Follow the steps given in the plugin section, and abort these steps.
If, given type, the new resource is to be handled by displaying some sort of inline content, e.g. a native rendering of the content, an error message because the specified type is not supported, or an inline prompt to allow the user to select a registered handler for the given type, then display the inline content and abort these steps.
Otherwise, the document's type is such that the resource will not affect the browsing context, e.g. because the resource is to be handed to an external application. Process the resource appropriately.

Some of the sections below, to which the above algorithm defers in certain cases, require the user agent to update the session history with the new page. When a user agent is required to do this, it must follows the set of steps given below that is appropriate for the situation at hand. From the point of view of any script, these steps must occur atomically.

pause for scripts
onbeforeunload
onunload
If the navigation was initiated for entry update of an entry
1. Replace the entry being updated with a new entry representing the new resource and its Document object and related state. The user agent may propagate state from the old entry to the new entry (e.g. scroll position).
2. Traverse the history to the new entry.
Otherwise
1. Remove all the entries after the current entry in the browsing context's Document object's History object.
  
  This doesn't necessarily have to affect the user agent's user interface.
2. Append a new entry at the end of the History object representing the new resource and its Document object and related state.
3. Traverse the history to the new entry.
4. If the navigation was initiated with replacement enabled, remove the entry immediately before the new current entry in the session history.

4.6.1. Page load processing model for HTML files

When an HTML document is to be loaded in a browsing context, the user agent must create a Document object, mark it as being an HTML document, create an HTML parser, associate it with the document, and begin to use the bytes provided for the document as the input stream for that parser.

The input stream converts bytes into characters for use in the tokeniser. This process relies, in part, on character encoding information found in the real Content-Type metadata of the resource; the "sniffed type" is not used for this purpose.

When no more bytes are available, an EOF character is implied, which eventually causes a load event to be fired.

After creating the Document object, but potentially before the page has finished parsing, the user agent must update the session history with the new page.

4.6.2. Page load processing model for XML files

When faced with displaying an XML file inline, user agents must first create a Document object, following the requirements of the XML and Namespaces in XML recommendations, RFC 3023, DOM3 Core, and other relevant specifications. [XML] [XMLNS] [RFC3023] [DOM3CORE]

The actual HTTP headers and other metadata, not the headers as mutated or implied by the algorithms given in this specification, are the ones that must be used when determining the character encoding according to the rules given in the above specifications.

User agents may examine the namespace of the root Element node of this Document object to perform namespace-based dispatch to alternative processing tools, e.g. determining that the content is actually a syndication feed and passing it to a feed handler. If such processing is to take place, abort the steps in this section, and jump to step 10 in the navigate steps above.

Otherwise, then, with the newly created Document, the user agents must update the session history with the new page. User agents may do this before the complete document has been parsed (thus achieving incremental rendering).

Error messages from the parse process (e.g. namespace well-formedness errors) may be reported inline by mutating the Document.

4.6.3. Page load processing model for text files

When a plain text document is to be loaded in a browsing context, the user agent should create a Document object, mark it as being an HTML document, create an HTML parser, associate it with the document, act as if the tokeniser had emitted a start tag token with the tag name "pre", set the tokenisation stage's content model flag to PLAINTEXT, and begin to pass the stream of characters in the plain text document to that tokeniser.

The rules for how to convert the bytes of the plain text document into actual characters are defined in RFC 2046, RFC 2646, and subsequent versions thereof. [RFC2046] [RFC2646]

When no more character are available, an EOF character is implied, which eventually causes a load event to be fired.

After creating the Document object, but potentially before the page has finished parsing, the user agent must update the session history with the new page.

User agents may add content to the head element of the Document, e.g. linking to stylesheet or an XBL binding, providing script, giving the document a title, etc.

4.6.4. Page load processing model for images

When an image resource is to be loaded in a browsing context, the user agent should create a Document object, mark it as being an HTML document, append an html element to the Document, append a head element and a body element to the html element, append an img to the body element, and set the src attribute of the img element to the address of the image.

Then, the user agent must act as if it had stopped parsing.

After creating the Document object, but potentially before the page has finished fully loading, the user agent must update the session history with the new page.

User agents may add content to the head element of the Document, or attributes to the img element, e.g. to link to stylesheet or an XBL binding, to provide a script, to give the document a title, etc.

4.6.5. Page load processing model for content that uses plugins

When a resource that requires an external resource to be rendered is to be loaded in a browsing context, the user agent should create a Document object, mark it as being an HTML document, append an html element to the Document, append a head element and a body element to the html element, append an embed to the body element, and set the src attribute of the img element to the address of the image.

Then, the user agent must act as if it had stopped parsing.

After creating the Document object, but potentially before the page has finished fully loading, the user agent must update the session history with the new page.

User agents may add content to the head element of the Document, or attributes to the embed element, e.g. to link to stylesheet or an XBL binding, or to give the document a title.

4.6.6. Page load processing model for inline content that doesn't have a DOM

When the user agent is to display a user agent page inline in a browsing context, the user agent should create a Document object, mark it as being an HTML document, and then either associate that Document with a custom rendering that is not rendered using the normal Document rendering rules, or mutate that Document until it represents the content the user agent wants to render.

Once the page has been set up, the user agent must act as if it had stopped parsing.

After creating the Document object, but potentially before the page has been completely set up, the user agent must update the session history with the new page.

4.6.7. Scrolling to a fragment identifier

When a user agent is supposed to scroll to a particular element, it may change the scrolling position of the document as desired, or perform any other relevant action.

how to get a "particular element" from a frag id -- id="", name="", XPointer, etc; missing IDs (e.g. the infamous "#top")

Then, the user agent must update the session history with the new page, where "the new page" has the same Document as before, but potentially has a different scroll position.

4.7. Determining the type of a new resource in a browsing context

It is imperative that the rules in this section be followed exactly. When two user agents use different heuristics for content type detection, security problems can occur. For example, if a server believes a contributed file to be an image (and thus benign), but a Web browser believes the content to be HTML (and thus capable of executing script), the end user can be exposed to malicious content, making the user vulnerable to cookie theft attacks and other cross-site scripting attacks.

The sniffed type of a resource must be found as follows:

If the resource was fetched over an HTTP protocol, and there is no HTTP Content-Encoding header, but there is an HTTP Content-Type header and it has a value whose bytes exactly match one of the following three lines:

Bytes in Hexadecimal	Textual representation
74 65 78 74 2f 70 6c 61 69 6e	`text/plain`
74 65 78 74 2f 70 6c 61 69 6e 3b 20 63 68 61 72 73 65 74 3d 49 53 4f 2d 38 38 35 39 2d 31	`text/plain; charset=ISO-8859-1`
74 65 78 74 2f 70 6c 61 69 6e 3b 20 63 68 61 72 73 65 74 3d 69 73 6f 2d 38 38 35 39 2d 31	`text/plain; charset=iso-8859-1`

...then jump to the text or binary section below.

Let official type be the type given by the Content-Type metadata for the resource (in lowercase, ignoring any parameters). If there is no such type, jump to the unknown type step below.
If official type ends in "+xml", or if it is either "text/xml" or "application/xml", then the the sniffed type of the resource is official type; return that and abort these steps.
If official type is an image type supported by the user agent (e.g. "image/png", "image/gif", "image/jpeg", etc), then jump to the images section below.
If official type is "text/html", then jump to the feed or HTML section below.
Otherwise, the sniffed type of the resource is official type.

4.7.1. Content-Type sniffing: text or binary

The user agent may wait for 512 or more bytes of the resource to be available.
Let n be the smaller of either 512 or the number of bytes already available.

If n is 4 or more, and the first bytes of the file match one of the following byte sets:

Bytes in Hexadecimal	Description
FE FF	UTF-16BE BOM or UTF-32LE BOM
FF FE	UTF-16LE BOM
00 00 FE FF	UTF-32BE BOM
EF BB BF	UTF-8 BOM

...then the sniffed type of the resource is "text/plain".

Otherwise, if any of the first n bytes of the resource are in one of the following byte ranges:
- 0x00 - 0x08
- 0x0E - 0x1A
- 0x1C - 0x1F
...then the sniffed type of the resource is "application/octet-stream".
Otherwise, the sniffed type of the resource is "text/plain".

4.7.2. Content-Type sniffing: unknown type

The user agent may wait for 512 or more bytes of the resource to be available.
Let stream length be the smaller of either 512 or the number of bytes already available.
For each row in the table below:
1. Let pattern length be the length of the pattern (number of bytes described by the cell in the second column of the row).
2. If pattern length is smaller than stream length then skip this row.
3. Apply the "and" operator to the first pattern length bytes of the resource and the given mask (the bytes in the cell of first column of that row), and let the result be the data.
4. If the bytes of the data matches the given pattern bytes exactly, then the sniffed type of the resource is the type given in the cell of the third column in that row; abort these steps.
As a last-ditch effort, jump to the text or binary section.

Bytes in Hexadecimal		Sniffed type	Comment
Mask	Pattern	Sniffed type	Comment
FF FF DF DF DF DF DF DF DF FF DF DF DF DF	3C 21 44 4F 43 54 59 50 45 20 48 54 4D 4C	text/html	The string "`<!DOCTYPE HTML`" in US-ASCII or compatible encodings, case-insensitively.
FF DF DF DF DF	3C 48 54 4D 4C	text/html	The string "`<HTML`" in US-ASCII or compatible encodings, case-insensitively.
FF FF FF FF FF	25 50 44 46 2D	application/pdf	The string "`%PDF-`", the PDF signature.
FF FF FF FF FF FF FF FF FF FF FF	25 21 50 53 2D 41 64 6F 62 65 2D	application/postscript	The string "`%!PS-Adobe-`", the PostScript signature.
FF FF FF FF FF FF	47 49 46 38 37 61	image/gif	The string "`GIF87a`", a GIF signature.
FF FF FF FF FF FF	47 49 46 38 39 61	image/gif	The string "`GIF89a`", a GIF signature.
FF FF FF FF FF FF FF FF	89 50 4E 47 0D 0A 1A 0A	image/png	The PNG signature.
FF FF FF	FF D8 FF	image/jpeg	A JPEG SOI marker followed by the first byte of another marker.

User agents may support further types if desired, by implicitly adding to the above table. However, user agents should not use any other patterns for types already mentioned in the table above, as this could then be used for privilege escalation (where, e.g., a server uses the above table to determine that content is not HTML and thus safe from XSS attacks, but then a user agent detects it as HTML anyway and allows script to execute).

4.7.3. Content-Type sniffing: image

If the first bytes of the file match one of the byte sequences in the first columns of the following table, then the sniffed type of the resource is the type given in the corresponding cell in the second column on the same row:

Bytes in Hexadecimal	Sniffed type	Comment
47 49 46 38 37 61	image/gif	The string "`GIF87a`", a GIF signature.
47 49 46 38 39 61	image/gif	The string "`GIF89a`", a GIF signature.
89 50 4E 47 0D 0A 1A 0A	image/png	The PNG signature.
FF D8 FF	image/jpeg	A JPEG SOI marker followed by the first byte of another marker.

User agents must ignore any rows for image types that they do not support.

Otherwise, the sniffed type of the resource is the same as its official type.

4.7.4. Content-Type sniffing: feed or HTML

The user agent may wait for 512 or more bytes of the resource to be available.
Let s be the stream of bytes, and let s[i] represent the byte in s with position i, treating s as zero-indexed (so the first byte is at i=0).
If at any point this algorithm requires the user agent to determine the value of a byte in s which is not yet available, or which is past the first 512 bytes of the resource, or which is beyond the end of the resource, the user agent must stop this algorithm, and assume that the sniffed type of the resource is "text/html".

User agents are allowed, by the first step of this algorithm, to wait until the first 512 bytes of the resource are available.
Initialise pos to 0.
Examine s[pos].

If it is 0x09 (ASCII tab), 0x20 (ASCII space), 0x0A (ASCII LF), or 0x0D (ASCII CR)
Increase pos by 1 and repeat this step.
If it is 0x3C (ASCII "<")
Increase pos by 1 and go to the next step.
If it is anything else
The sniffed type of the resource is "text/html". Abort these steps.
If the bytes with positions pos to pos+2 in s are exactly equal to 0x21, 0x2D, 0x2D respectively (ASCII for "!--"), then:
1. Increase pos by 3.
2. If the bytes with positions pos to pos+2 in s are exactly equal to 0x2D, 0x2D, 0x3E respectively (ASCII for "-->"), then increase pos by 3 and jump back to the previous step (step 5) in the overall algorithm in this section.
3. Otherwise, increase pos by 1.
4. Otherwise, return to step 2 in these substeps.
If s[pos] is 0x21 (ASCII "!"):
1. Increase pos by 1.
2. If s[pos] equal 0x3E, then increase pos by 1 and jump back to step 5 in the overall algorithm in this section.
3. Otherwise, return to step 1 in these substeps.
If s[pos] is 0x3F (ASCII "?"):
1. Increase pos by 1.
2. If s[pos] and s[pos+1] equal 0x3F and 0x3E respectively, then increase pos by 1 and jump back to step 5 in the overall algorithm in this section.
3. Otherwise, return to step 1 in these substeps.

Otherwise, if the bytes in s starting at pos match any of the sequences of bytes in the first column of the following table, then the user agent must follow the steps given in the corresponding cell in the second column of the same row.

Bytes in Hexadecimal	Requirement	Comment
72 73 73	The sniffed type of the resource is "application/rss+xml"; abort these steps	The three ASCII characters "`rss`"
66 65 65 64	The sniffed type of the resource is "application/atom+xml"; abort these steps	The four ASCII characters "`feed`"
72 64 66 3A 52 44 46	Continue to the next step in this algorithm	The ASCII characters "`rdf:RDF`"

If none of the byte sequences above match the bytes in s starting at pos, then the sniffed type of the resource is "text/html". Abort these steps.

If, before the next ">", you find two xmlns* attributes with http://www.w3.org/1999/02/22-rdf-syntax-ns# and http://purl.org/rss/1.0/ as the namespaces, then the sniffed type of the resource is "application/rss+xml", abort these steps. (maybe we only need to check for http://purl.org/rss/1.0/ actually)
Otherwise, the sniffed type of the resource is "text/html".

For efficiency reaons, implementations may wish to implement this algorithm and the algorithm for detecting the character encoding of HTML documents in parallel.

4.7.5. Content-Type metadata

What explicit Content-Type metadata is associated with the resource (the resource's type information) depends on the protocol that was used to fetch the resource.

For HTTP resources, only the Content-Type HTTP header contributes any data; the explicit type of the resource is then the value of that header, interpreted as described by the HTTP specifications. [HTTP]

For resources fetched from the filesystem, user agents should use platform-specific conventions, e.g. operating system extension/type mappings.

Extensions must not be used for determining resource types for resources fetched over HTTP.

For resources fetched over most other protocols, e.g. FTP, there is no type information.

4.8. User prompts

The alert(message) method, when invoked, must show the given message to the user. The user agent may make the method wait for the user to acknowledge the message before returning; if so, the user agent must pause while the method is waiting.

The confirm(message) method, when invoked, must show the given message to the user, and ask the user to respond with a positive or negative response. The user agent must then pause as the the method waits for the user's response. If the user response positively, the method must return true, and if the user response negatively, the method must return false.

The prompt(message, default) method, when invoked, must show the given message to the user, and ask the user to either respond with a string value or abort. The user agent must then pause as the the method waits for the user's response. The second argument is optional. If the second argument (default) is present, then the response must be defaulted to the value given by default. If the user aborts, then the method must return null; otherwise, the method must return the string that the user responded with.

4.9. Scripting

4.9.1. Running executable code

Various mechanisms can cause author-provided executable code to run in the context of a document. These mechanisms include, but are probably not limited to:

Processing of script elements.
Processing of inline javascript: URIs (e.g. the src attribute of img elements, or an @import rule in a CSS style element block).
Event handlers, whether registered through the DOM using addEventListener(), by explicit event handler content attributes, by event handler DOM attributes, or otherwise.
Processing of technologies like XBL or SVG that have their own scripting features.

User agents may provide a mechanism to enable or disable the execution of author-provided code. When the user agent is configured such that author-provided code does not execute, or if the user agent is implemented so as to never execute author-provided code, it is said that scripting is disabled. When author-provided code does execute, scripting is enabled. A user agent with scripting disabled is a user agent with no scripting support for the purposes of conformance.

4.9.2. Origin

Access to certain APIs is granted or denied to scripts based on the origin of the script and the API being accessed.

If a script is in a script element: The origin of the script is the origin of the Document to which the script element belongs.
If a script is a function or other code reference created by another script: The origin of the script is the origin of the script that created it.
If a script is a javascript: URI in an attribute: The origin is the origin of the Document of the element on which the attribute is found.
If a script is a javascript: URI in a style sheet: The origin is the origin of the Document to which the style sheet applies.
If a script is a javascript: URI to which a browsing context is being navigated, the URI having been declared in markup: The origin is the origin of the Document of the element (e.g. an a or area element) that declared the URI.
If a script is a javascript: URI to which a browsing context is being navigated, the URI having been provided by script: The origin is the origin of the script that provided the URI.

The origin of scripts thus comes down to finding the origin of Document objects.

The origin of a Document or image that was served over the network and whose address uses a URI scheme with a server-based naming authority is the tuple consisting of the <scheme>, <host>, and <port> parts of the Document's full URI. [RFC2396] [RFC2732]

The origin of a Document or image that was generated from a data: URI found in another Document or in a script is the origin of the that Document or script.

The origin of a Document or image that was generated from a data: URI from another source is a globally unique identifier assigned when the document is created.

The origin of a Document or image that was generated from a javascript: URI is the same as the origin of that javascript: URI.

The string representing the script's domain in IDNA format is obtained as follows: take the domain part of the script's origin tuple and apply the IDNA ToASCII algorithm and then the IDNA ToUnicode algorithm to each component of the domain name (with both the AllowUnassigned and UseSTD3ASCIIRules flags set both times). [RFC3490]

If ToASCII fails to convert one of the components of the string, e.g. because it is too long or because it contains invalid characters, or if the origin of the script has no domain part, then the string representing the script's domain in IDNA format cannot be obtained. (ToUnicode is defined to never fail.)

4.9.3. Security exceptions

Define security exception.

4.9.4. The `javascript:` protocol

A URI using the javascript: protocol must, if evaluated, be evaluated using the in-context evaluation operation defined for javascript: URIs. [JSURI]

When a browsing context is navigated to a javascript: URI, and the active document of that browsing context has the same origin as the URI, the dereference context must be the browsing context being navigated.

When a browsing context is navigated to a javascript: URI, and the active document of that browsing context has a different origin than the URI, the dereference context must be an empty object.

Otherwise, the dereference context must the browsing context of the Document to which belongs the element for which the URI is being dereferenced, or to which the style sheet for which the URI is being dereferenced applies, whichever is appropriate.

URIs using the javascript: protocol should be evaluated when the resource for that URI is needed, unless scripting is disabled or the Document corresponding to the dereference context (as defined above), if any, has designMode enabled.

If the dereference by-product is void (there is no return value), then the URI must be treated in a manner equivalent to an HTTP resource with an HTTP 204 No Content response.

Otherwise, the URI must be treated in a manner equivalent to an HTTP resource with a 200 OK response whose Content-Type metadata is text/html and whose response body is the dereference by-product, converted to a string value.

Certain contexts, in particular img elements, ignore the Content-Type metadata.

So for example a javascript: URI for a src attribute of an img element would be evaluated in the context of the page as soon as the attribute is set; it would then be sniffed to determine the image type and decoded as an image.

A javascript: URI in an href attribute of an a element would only be evaluated when the link was followed.

The src attribute of an iframe element would be evaluated in the context of the iframe's own browsing context; once evaluated, its return value (if it was not void) would replace that browsing context's document, thus changing the variables visible in that browsing context.

4.9.5. Events

4.9.5.1. Event handler attributes

HTML elements can have event handler attributes specified. These act as bubbling event listeners for the element on which they are specified.

Each event handler attribute has two parts, an event handler content attribute and an event handler DOM attribute. Event handler attributes must initially be set to null. When their value changes (through the changing of their event handler content attribute or their event handler DOM attribute), they will either be null, or have an EventListener object assigned to them.

Objects other than Element objects, in particular Window, only have event handler DOM attribute (since they have no content attributes).

In the ECMAScript DOM binding, the ECMAScript native Function type must implement the EventListener interface such that invoking the handleEvent() method of that interface on the object from another language binding invokes the function itself, with the event argument as its only argument. In the ECMAScript binding itself, however, the handleEvent() method of the interface is not directly accessible on Function objects. Such functions, when invoked, must be called in the scope of the browsing context that they were created in. If the function returns false, the event's preventDefault() method must then invoked. Exception: for historical reasons, for the HTML mouseover event, the preventDefault() method must be called when the function returns true instead.

Event handler content attributes, when specified, must contain valid ECMAScript code matching the ECMAScript FunctionBody production. [ECMA262]

When an event handler content attribute is set, its new value must be interpreted as the body of an anonymous function with a single argument called event, with the new function's scope chain being linked from the activation object of the handler, to the element, to the element's form element if it is a form control, to the Document object, to the browsing context of that Document. The function's this parameter must be the Element object representing the element. The resulting function must then be set as the value of the corresponding event handler attribute, and the new value must be set as the value of the content attribute. If the given function body fails to compile, then the corresponding event handler attribute must be set to null instead (the content attribute must still be updated to the new value, though).

See ECMA262 Edition 3, sections 10.1.6 and 10.2.3, for more details on activation objects. [ECMA262]

How do we allow non-JS event handlers?

Event handler DOM attributes, on setting, must set the corresponding event handler attribute to their new value, and on getting, must return whatever the current value of the corresponding event handler attribute is (possibly null).

The following are the event handler attributes that must be supported by all HTML elements, as both content attributes and DOM attributes, and on Window objects, as DOM attributes:

onabort

Must be invoked whenever an abort event is targeted at or bubbles through the element.

onbeforeunload

Must be invoked whenever a beforeunload event is targeted at or bubbles through the element.

onblur

Must be invoked whenever a blur event is targeted at or bubbles through the element.

onchange

Must be invoked whenever a change event is targeted at or bubbles through the element.

onclick

Must be invoked whenever a click event is targeted at or bubbles through the element.

oncontextmenu

Must be invoked whenever a contextmenu event is targeted at or bubbles through the element.

ondblclick

Must be invoked whenever a dblclick event is targeted at or bubbles through the element.

ondrag

Must be invoked whenever a drag event is targeted at or bubbles through the element.

ondragend

Must be invoked whenever a dragend event is targeted at or bubbles through the element.

ondragenter

Must be invoked whenever a dragenter event is targeted at or bubbles through the element.

ondragleave

Must be invoked whenever a dragleave event is targeted at or bubbles through the element.

ondragover

Must be invoked whenever a dragover event is targeted at or bubbles through the element.

ondragstart

Must be invoked whenever a dragstart event is targeted at or bubbles through the element.

ondrop

Must be invoked whenever a drop event is targeted at or bubbles through the element.

onerror

Must be invoked whenever an error event is targeted at or bubbles through the element.

The onerror handler is also used for reporting script errors.

onfocus

Must be invoked whenever a focus event is targeted at or bubbles through the element.

onkeydown

Must be invoked whenever a keydown event is targeted at or bubbles through the element.

onkeypress

Must be invoked whenever a keypress event is targeted at or bubbles through the element.

onkeyup

Must be invoked whenever a keyup event is targeted at or bubbles through the element.

onload

Must be invoked whenever a load event is targeted at or bubbles through the element.

onmessage

Must be invoked whenever a message event is targeted at or bubbles through the element.

onmousedown

Must be invoked whenever a mousedown event is targeted at or bubbles through the element.

onmousemove

Must be invoked whenever a mousemove event is targeted at or bubbles through the element.

onmouseout

Must be invoked whenever a mouseout event is targeted at or bubbles through the element.

onmouseover

Must be invoked whenever a mouseover event is targeted at or bubbles through the element.

onmouseup

Must be invoked whenever a mouseup event is targeted at or bubbles through the element.

onmousewheel

Must be invoked whenever a mousewheel event is targeted at or bubbles through the element.

onresize

Must be invoked whenever a resize event is targeted at or bubbles through the element.

onscroll

Must be invoked whenever a scroll event is targeted at or bubbles through the element.

onselect

Must be invoked whenever a select event is targeted at or bubbles through the element.

onsubmit

Must be invoked whenever a submit event is targeted at or bubbles through the element.

onunload

Must be invoked whenever an unload event is targeted at or bubbles through the element.

When scripting is disabled, event handler attributes must do nothing.

When scripting is enabled, all event handler attributes on an element, whether set to null or to a function, must be registered as event listeners on the element, as if the addEventListenerNS() method on the Element object's EventTarget interface had been invoked when the element was created, with the event type (type argument) equal to the type described for the event handler attribute in the list above, the namespace (namespaceURI argument) set to null, the listener set to be a target and bubbling phase listener (useCapture argument set to false), the event group set to the default group (evtGroup argument set to null), and the event listener itself (listener argument) set to do nothing while the event handler attribute is null, and set to invoke the function associated with the event handler attribute otherwise.

4.9.5.2. Event firing

maybe this should be moved higher up (terminology? conformance? DOM?)

Certain operations and methods are defined as firing events on elements. For example, the click() method on the HTMLElement interface is defined as firing a click event on the element. [DOM3EVENTS]

Firing a click event means that a click event with no namespace, which bubbles and is cancelable, and which uses the MouseEvent interface, must be dispatched at the given element. The event object must have its screenX, screenY, clientX, clientY, and button attributes set to 0, its ctrlKey, shiftKey, altKey, and metaKey attributes set according to the current state of the key input device, if any (false for any keys that are not available), its detail attribute set to 1, and its relatedTarget attribute set to null. The getModifierState() method on the object must return values appropriately describing the state of the key input device at the time the event is created.

Firing a change event means that a change event with no namespace, which bubbles but is not cancelable, and which uses the Event interface, must be dispatched at the given element. The event object must have its detail attribute set to 0.

Firing a contextmenu event means that a contextmenu event with no namespace, which bubbles and is cancelable, and which uses the Event interface, must be dispatched at the given element. The event object must have its detail attribute set to 0.

Firing a simple event called e means that an event with the name e, with no namespace, which does not bubble but is cancelable, and which uses the Event interface, must be dispatched at the given element.

Firing a show event means firing a simple event called show. Firing a load event means firing a simple event called load. Firing an error event means firing a simple event called error.

Firing a progress event called e means something that hasn't yet been defined, in the [PROGRESS] spec.

The default action of these event is to do nothing unless otherwise stated.

If you dispatch a custom "click" event at an element that would normally have default actions, should they get triggered? If so, we need to go through the entire spec and make sure that any default actions are defined in terms of any event of the right type on that element, not those that are dispatched in expected ways.

4.9.5.3. Events and the `Window` object

When an event is dispatched at a DOM node in a Document in a browsing context, if the event is not a load event, the user agent must also dispatch the event to the Window, as follows:

In the capture phase, the event must be dispatched to the Window object before being dispatched to any of the nodes.
In the bubble phase, the event must be dispatched to the Window object at the end of the phase, unless bubbling has been prevented.

4.9.5.4. Runtime script errors

This section only applies to user agents that support scripting in general and ECMAScript in particular.

Whenever a runtime script error occurs in one of the scripts associated with the document, the value of the onerror event handler DOM attribute of the Window object must be processed, as follows:

If the value is a function

The function referenced by the onerror attribute must be invoked with three arguments, before notifying the user of the error.

The three arguments passed to the function are all DOMStrings; the first must give the message that the UA is considering reporting, the second must give the URI to the resource in which the error occured, and the third must give the line number in that resource on which the error occured.

If the function returns false, then the error should not be reported to the user. Otherwise, if the function returns another value (or does not return at all), the error should be reported to the user.

Any exceptions thrown or errors caused by this function must be reported to the user immediately after the error that the function was called for, without calling the function again.

If the value is null

The error should not reported to the user.

If the value is anything else

The error should be reported to the user.

The initial value of onerror must be undefined.

4.10. Browser state

The navigator attribute of the Window interface must return an instance of the ClientInformation interface, which represents the identity and state of the user agent (the client), and allows Web pages to register themselves as potential protocol and content handlers:

interface ClientInformation {
  readonly attribute boolean onLine;
  void registerProtocolHandler(in DOMString protocol, in DOMString uri, in DOMString title);
  void registerContentHandler(in DOMString mimeType, in DOMString uri, in DOMString title);
};

4.10.1. Offline Web applications

The navigator.onLine attribute must return false if the user agent will not contact the network when the user follows links or when a script requests a remote page (or knows that such an attempt would fail), and must return true otherwise.

The offline event must be fired when the value of the navigator.onLine attribute of the Window changes from true to false.

The online event must be fired when the value of the navigator.onLine attribute of the Window changes from false to true.

These events are in no namespace, do bubble, are not cancelable, have no default action, and use the normal Event interface. They must be fired on the body element. (As the events bubble, they will reach the Window object.)

4.10.2. Custom protocol and content handlers

The registerProtocolHandler() method allows Web sites to register themselves as possible handlers for particular protocols. For example, an online fax service could register itself as a handler of the fax: protocol ([RFC2806]), so that if the user clicks on such a link, he is given the opportunity to use that Web site. Analogously, the registerContentHandler() method allows Web sites to register themselves as possible handlers for content in a particular MIME type. For example, the same online fax service could register itself as a handler for image/g3fax files ([RFC1494]), so that if the user has no native application capable of handling G3 Facsimile byte streams, his Web browser can instead suggest he use that site to view the image.

User agents may, within the constraints described in this section, do whatever they like when the methods are called. A UA could, for instance, prompt the user and offer the user the opportunity to add the site to a shortlist of handlers, or make the handlers his default, or cancel the request. UAs could provide such a UI through modal UI or through a non-modal transient notification interface. UAs could also simply silently collect the information, providing it only when relevant to the user.

There is an example of how these methods could be presented to the user below.

The arguments to the methods have the following meanings:

protocol (registerProtocolHandler() only)

A scheme, such as ftp or fax. The scheme must be treated case-insensitively by user agents for the purposes of comparing with the scheme part of URIs that they consider against the list of registered handlers.

The protocol value, if it contains a colon (as in "ftp:"), will never match anything, since schemes don't contain colons.

mimeType (registerContentHandler() only)

A MIME type, such as model/vrml or text/richtext. The MIME type must be treated case-insensitively by user agents for the purposes of comparing with MIME types of documents that they consider against the list of registered handlers.

User agents must compare the given values only to the MIME type/subtype parts of content types, not to the complete type including parameters. Thus, if mimeType values passed to this method include characters such as commas or whitespace, or include MIME parameters, then the handler being registered will never be used.

uri

The URI of the page that will handle the requests. When the user agent uses this URI, it must replace the first occurrence of the exact literal string "%s" with an escaped version of the URI of the content in question (as defined below), and then fetch the resulting URI using the GET method (or equivalent for non-HTTP URIs).

To get the escaped version of the URI, first, the domain part of the URI (if any) must be converted to its punycode representation, and then, every character in the URI that is not in the ranges given in the next paragraph must be replaced by its UTF-8 byte representation, each byte being represented by a U+0025 (%) character and two digits in the range U+0030 (0) to U+0039 (9) and U+0041 (A) to U+0046 (F) giving the hexadecimal representation of the byte.

The ranges of characters that must not be escaped are: U+002D (-), U+002E (.), U+0030 (0) to U+0039 (9), U+0041 (A) to U+005A (Z), U+005F (_), U+0061 (a) to U+007A (z), and U+007E (~).

If the user had visited a site that made the following call:

navigator.registerContentHandler('application/x-soup', 'http://example.com/soup?url=%s', 'SoupWeb™')

...and then clicked on a link such as:

<a href="http://www.example.net/chickenkïwi.soup">Download our Chicken Kiwi soup!</a>

...then, assuming this chickenkiwi.soup file was served with the MIME type application/x-soup, the UA might instead navigate to the following URI:

http://example.com/soup?url=http%3A%2F%2Fwww.example.net%2Fchickenk%C3%AFwi.soup

This site could then fetch the chickenkiwi.soup file and do whatever it is that it does with soup (synthesise it and ship it to the user, or whatever).

title

A descriptive title of the handler, which the UA might use to remind the user what the site in question is.

User agents should raise security exceptions if the methods are called with protocol or mimeType values that the UA deems to be "privileged". For example, a site attempting to register a handler for http URIs or text/html content in a Web browser would likely cause an exception to be raised.

User agents must raise a SYNTAX_ERR exception if the uri argument passed to one of these methods does not contain the exact literal string "%s".

User agents must not raise any other exceptions (other than binding-specific exceptions, such as for an incorrect number of arguments in an ECMAScript implementation).

This section does not define how the pages registered by these methods are used, beyond the requirements on how to process the uri value (see above). To some extent, the processing model for navigating across documents defines some cases where these methods are relevant, but in general UAs may use this information wherever they would otherwise consider handing content to native plugins or helper applications.

UAs must not use registered content handlers to handle content that was returned as part of a non-GET transaction (or rather, as part of any non-idempotent transaction), as the remote site would not be able to fetch the same data.

4.10.2.1. Security and privacy

These mechanisms can introduce a number of concerns, in particular privacy concerns.

Hijacking all Web usage. User agents should not allow protocols that are key to its normal operation, such as http or https, to be rerouted through third-party sites. This would allow a user's activities to be trivially tracked, and would allow user information, even in secure connections, to be collected.

Hijacking defaults. It is strongly recommended that user agents do not automatically change any defaults, as this could lead the user to send data to remote hosts that the user is not expecting. New handlers registering themselves should never automatically cause those sites to be used.

Registration spamming. User agents should consider the possibility that a site will attempt to register a large number of handlers, possibly from multiple domains (e.g. by redirecting through a series of pages each on a different domain, and each registering a handler for video/mpeg — analogous practices abusing other Web browser features have been used by pornography Web sites for many years). User agents should gracefully handle such hostile attempts, protecting the user.

Misleading titles. User agents should not rely wholy on the title argument to the methods when presenting the registered handlers to the user, since sites could easily lie. For example, a site hostile.example.net could claim that it was registering the "Cuddly Bear Happy Content Handler". User agents should therefore use the handler's domain in any UI along with any title.

Hostile handler metadata. User agents should protect against typical attacks against strings embedded in their interface, for example ensuring that markup or escape characters in such strings are not executed, that null bytes are properly handled, that over-long strings do not cause crashes or buffer overruns, and so forth.

Leaking Intranet URIs. The mechanism described in this section can result in secret Intranet URIs being leaked, in the following manner:

The user registers a third-party content handler as the default handler for a content type.
The user then browses his corporate Intranet site and accesses a document that uses that content type.
The user agent contacts the third party and hands the third party the URI to the Intranet content.

No actual confidential file data is leaked in this manner, but the URIs themselves could contain confidential information. For example, the URI could be https://www.corp.example.com/upcoming-aquisitions/samples.egf, which might tell the third party that Example Corporation is intending to merge with Samples LLC. Implementors might wish to consider allowing administrators to disable this feature for certain subdomains, content types, or protocols.

Leaking secure URIs. User agents should not send HTTPS URIs to third-party sites registered as content handlers, in the same way that user agents do not send Referer headers from secure sites to third-party sites.

Leaking credentials. User agents must never send username or password information in the URIs that are escaped and included sent to the handler sites. User agents may even avoid attempting to pass to Web-based handlers the URIs of resources that are known to require authentication to access, as such sites would be unable to access the resources in question without prompting the user for credentials themselves (a practice that would require the user to know whether to trust the third-party handler, a decision many users are unable to make or even understand).

4.10.2.2. Sample user interface

This section is non-normative.

A simple implementation of this feature for a desktop Web browser might work as follows.

The registerProtocolHandler() method could display a modal dialog box:

||[ Protocol Handler Registration ]|||||||||||||||||||||||||||
|                                                            |
| This Web page:                                             |
|                                                            |
|    Kittens at work                                         |
|    http://kittens.example.org/                             |
|                                                            |
| ...would like permission to handle the protocol "x-meow:"  |
| using the following Web-based application:                 |
|                                                            |
|    Kittens-at-work displayer                               |
|    http://kittens.example.org/?show=%s                     |
|                                                            |
| Do you trust the administrators of the "kittens.example.   |
| org" domain?                                               |
|                                                            |
|              ( Trust kittens.example.org )  (( Cancel ))   |
|____________________________________________________________|

...where "Kittens at work" is the title of the page that invoked the method, "http://kittens.example.org/" is the URI of that page, "x-meow" is the string that was passed to the registerProtocolHandler() method as its first argument (protocol), "http://kittens.example.org/?show=%s" was the second argument (uri), and "Kittens-at-work displayer" was the third argument (title).

If the user clicks the Cancel button, then nothing further happens. If the user clicks the "Trust" button, then the handler is remembered.

When the user then attempts to fetch a URI that uses the "x-meow:" scheme, then it might display a dialog as follows:

||[ Unknown Protocol ]||||||||||||||||||||||||||||||||||||||||
|                                                            |
| You have attempted to access:                              |
|                                                            |
|    x-meow:S2l0dGVucyBhcmUgdGhlIGN1dGVzdCE%3D               |
|                                                            |
| How would you like FerretBrowser to handle this resource?  |
|                                                            |
|  (o) Contact the FerretBrowser plugin registry to see if   |
|      there is an official way to handle this resource.     |
|                                                            |
|  ( ) Pass this URI to a local application:                 |
|      [ /no application selected/             ] ( Choose )  |
|                                                            |
|  ( ) Pass this URI to the "Kittens-at-work displayer"      |
|      application at "kittens.example.org".                 |
|                                                            |
|  [ ] Always do this for resources using the "x-meow"       |
|      protocol in future.                                   |
|                                                            |
|                                     ( Ok )  (( Cancel ))   |
|____________________________________________________________|

...where the third option is the one that was primed by the site registering itself earlier.

If the user does select that option, then the browser, in accordance with the requirements described in the previous two sections, will redirect the user to "http://kittens.example.org/?show=x-meow%3AS2l0dGVucyBhcmUgdGhlIGN1dGVzdCE%253D".

The registerContentHandler() method would work equivalently, but for unknown MIME types instead of unknown protocols.

4.11. Client-side session and persistent storage

4.11.1. Introduction

This section is non-normative.

This specification introduces two related mechanisms, similar to HTTP session cookies [RFC2965], for storing structured data on the client side.

The first is designed for scenarios where the user is carrying out a single transaction, but could be carrying out multiple transactions in different windows at the same time.

Cookies don't really handle this case well. For example, a user could be buying plane tickets in two different windows, using the same site. If the site used cookies to keep track of which ticket the user was buying, then as the user clicked from page to page in both windows, the ticket currently being purchased would "leak" from one window to the other, potentially causing the user to buy two tickets for the same flight without really noticing.

To address this, this specification introduces the sessionStorage DOM attribute. Sites can add data to the session storage, and it will be accessible to any page from that domain opened in that window.

For example, a page could have a checkbox that the user ticks to indicate that he wants insurance:

<label>
 <input type="checkbox" onchange="sessionStorage.insurance = checked">
 I want insurance on this trip.
</label>

A later page could then check, from script, whether the user had checked the checkbox or not:

if (sessionStorage.insurance) { ... }

If the user had multiple windows opened on the site, each one would have its own individual copy of the session storage object.

The second storage mechanism is designed for storage that spans multiple windows, and lasts beyond the current session. In particular, Web applications may wish to store megabytes of user data, such as entire user-authored documents or a user's mailbox, on the clientside for performance reasons.

Again, cookies do not handle this case well, because they are transmitted with every request.

The globalStorage DOM attribute is used to access the global storage areas.

The site at example.com can display a count of how many times the user has loaded its page by putting the following at the bottom of its page:

<p>
  You have viewed this page
  <span id="count">an untold number of</span>
  time(s).
</p>
<script>
  var storage = globalStorage['example.com'];
  if (!storage.pageLoadCount)
    storage.pageLoadCount = 0;
  storage.pageLoadCount = parseInt(storage.pageLoadCount, 10) + 1;
  document.getElementById('count').textContent = storage.pageLoadCount;
</script>

Each domain and each subdomain has its own separate storage area. Subdomains can access the storage areas of parent domains, and domains can access the storage areas of subdomains.

globalStorage[''] is accessible to all domains.
globalStorage['com'] is accessible to all .com domains
globalStorage['example.com'] is accessible to example.com and any of its subdomains
globalStorage['www.example.com'] is accessible to www.example.com and example.com, but not www2.example.com.

Storage areas (both session storage and global storage) store strings. To store structured data in a storage area, you must first convert it to a string.

4.11.2. The `Storage` interface

interface Storage {
  readonly attribute unsigned long length;
  DOMString key(in unsigned long index);
  StorageItem getItem(in DOMString key);
  void setItem(in DOMString key, in DOMString data);
  void removeItem(in DOMString key);
};

Each Storage object provides access to a list of key/value pairs, which are sometimes called items. Keys are strings, and any string (including the empty string) is a valid key. Values are strings with associated metadata, represented by StorageItem objects.

Each Storage object is associated with a list of key/value pairs when it is created, as defined in the sections on the sessionStorage and globalStorage attributes. Multiple separate objects implementing the Storage interface can all be associated with the same list of key/value pairs simultaneously.

Key/value pairs have associated metadata. In particular, a key/value pair can be marked as either "safe only for secure content", or as "safe for both secure and insecure content".

A key/value pair is accessible if either it is marked as "safe for both secure and insecure content", or it is marked as "safe only for secure content" and the script in question is running in a secure browsing context.

The length attribute must return the number of key/value pairs currently present and accessible in the list associated with the object.

The key(n) method must return the name of the nth accessible key in the list. The order of keys is user-agent defined, but must be consistent within an object between changes to the number of keys. (Thus, adding or removing a key may change the order of the keys, but merely changing the value of an existing key must not.) If n is less than zero or greater than or equal to the number of key/value pairs in the object, then this method must raise an INDEX_SIZE_ERR exception.

The getItem(key) method must return the StorageItem object representing the key/value pair with the given key. If the given key does not exist in the list associated with the object, or is not accessible, then this method must return null. Subsequent calls to this method with the same key from scripts running in the same security context must return the same instance of the StorageItem interface. (Such instances must not be shared across security contexts, though.)

The setItem(key, value) method must first check if a key/value pair with the given key already exists in the list associated with the object.

If it does not, then a new key/value pair must be added to the list, with the given key and value, such that any current or future StorageItem objects referring to this key/value pair will return the value given in the value argument. If the script setting the value is running in a secure browsing context, then the key/value pair must be marked as "safe only for secure content", otherwise it must be marked as "safe for both secure and insecure content".

If the given key does exist in the list, then, if the key/value pair with the given key is accessible, it must have its value updated so that any current or future StorageItem objects referring to this key/value pair will return the value given in the value argument. If it is not accessible, the method must raise a security exception.

When the setItem() method is successfully invoked (i.e. when it doesn't raise an exception), events are fired on other HTMLDocument objects that can access the newly stored data, as defined in the sections on the sessionStorage and globalStorage attributes.

The removeItem(key) method must cause the key/value pair with the given key to be removed from the list associated with the object, if it exists and is accessible. If no item with that key exists, the method must do nothing. If an item with that key exists but is not accessible, the method must raise a security exception.

The setItem() and removeItem() methods must be atomic with respect to failure. That is, changes to the data storage area must either be successful, or the data storage area must not be changed at all.

In the ECMAScript DOM binding, enumerating a Storage object must enumerate through the currently stored and accessible keys in the list the object is associated with. (It must not enumerate the values or the actual members of the interface). In the ECMAScript DOM binding, Storage objects must support dereferencing such that getting a property that is not a member of the object (i.e. is neither a member of the Storage interface nor of Object) must invoke the getItem() method with the property's name as the argument, and setting such a property must invoke the setItem() method with the property's name as the first argument and the given value as the second argument.

4.11.3. The `StorageItem` interface

Items in Storage objects are represented by objects implementing the StorageItem interface.

interface StorageItem {
           attribute boolean secure;
           attribute DOMString value;
};

In the ECMAScript DOM binding, StorageItem objects must stringify to their value attribute's value.

The value attribute must return the current value of the key/value pair represented by the object. When the attribute is set, the user agent must invoke the setItem() method of the Storage object that the StorageItem object is associated with, with the key that the StorageItem object is associated with as the first argument, and the new given value of the attribute as the second argument.

StorageItem objects must be live, meaning that as the underlying Storage object has its key/value pairs updated, the StorageItem objects must always return the actual value of the key/value pair they represent.

If the key/value pair has been deleted, the StorageItem object must act as if its value was the empty string. On setting, the key/value pair will be recreated.

The secure attribute must raise an INVALID_ACCESS_ERR exception when accessed or set from a script whose browsing context is not considered secure. (Basically, if the page is not an SSL page.)

If the browsing context is secure, then the secure attribute must return true if the key/value pair is considered "safe only for secure content", and false if it is considered "safe for both secure and insecure content". If it is set to true, then the key/value pair must be flagged as "safe only for secure content". If it is set to false, then the key/value pair must be flagged as "safe for both secure and insecure content".

If a StorageItem object is obtained by a script that is not running in a secure browsing context, and the item is then marked with the "safe only for secure content" flag by a script that is running in a secure context, the StorageItem object must continue to be available to the first script, who will be able to read the value of the object. However, any attempt to set the value would then start raising exceptions as described in the previous section, and the key/value pair would no longer appear in the appropriate Storage object.

4.11.4. The `sessionStorage` attribute

The sessionStorage attribute represents the storage area specific to the current top-level browsing context.

Each top-level browsing context has a unique set of session storage areas, one for each domain.

User agents should not expire data from a browsing context's session storage areas, but may do so when the user requests that such data be deleted, or when the UA detects that it has limited storage space, or for security reasons. User agents should always avoid deleting data while a script that could access that data is running. When a top-level browsing context is destroyed (and therefore permanently inaccessible to the user) the data stored in its session storage areas can be discarded with it, as the API described in this specification provides no way for that data to ever be subsequently retrieved.

The lifetime of a browsing context can be unrelated to the lifetime of the actual user agent process itself, as the user agent may support resuming sessions after a restart.

When a new HTMLDocument is created, the user agent must check to see if the document's top-level browsing context has allocated a session storage area for that document's domain. If it has not, a new storage area for that document's domain must be created.

The Storage object for the document's associated Window object's sessionStorage attribute must then be associated with the domain's session storage area.

When a new top-level browsing context is created by cloning an existing browsing context, the new browsing context must start with the same session storage areas as the original, but the two sets must from that point on be considered separate, not affecting each other in any way.

When a new top-level browsing context is created by a script in an existing browsing context, or by the user following a link in an existing browsing context, or in some other way related to a specific HTMLDocument, then, if the new context's first HTMLDocument has the same domain as the HTMLDocument from which the new context was created, the new browsing context must start with a single session storage area. That storage area must be a copy of that domain's session storage area in the original browsing context, which from that point on must be considered separate, with the two storage areas not affecting each other in any way.

When the setItem() method is called on a Storage object x that is associated with a session storage area, then, if the method does not raise a security exception, in every HTMLDocument object whose Window object's sessionStorage attribute's Storage object is associated with the same storage area, other than x, a storage event must be fired, as described below.

4.11.5. The `globalStorage` attribute

interface StorageList {
  Storage namedItem(in DOMString domain);
};

The globalStorage object provides a Storage object for each domain.

In the ECMAScript DOM binding, StorageList objects must support dereferencing such that getting a property that is not a member of the object (i.e. is neither a member of the StorageList interface nor of Object) must invoke the namedItem() method with the property's name as the argument.

User agents must have a set of global storage areas, one for each domain.

User agents should only expire data from the global storage areas for security reasons or when requested to do so by the user. User agents should always avoid deleting data while a script that could access that data is running. Data stored in global storage areas should be considered potentially user-critical. It is expected that Web applications will use the global storage areas for storing user-written documents.

The namedItem(domain) method tries to returns a Storage object associated with the given domain, according to the rules that follow.

The domain must first be split into an array of strings, by splitting the string at "." characters (U+002E FULL STOP). If the domain argument is the empty string, then the array is empty as well. If the domain argument is not empty but has no dots, then the array has one item, which is equal to the domain argument. If the domain argument contains consecutive dots, there will be empty strings in the array (e.g. the string "hello..world" becomes split into the three strings "hello", "", and "world", with the middle one being the empty string).

Each component of the array must then have the IDNA ToASCII algorithm applied to it, with both the AllowUnassigned and UseSTD3ASCIIRules flags set. [RFC3490] If ToASCII fails to convert one of the components of the string, e.g. because it is too long or because it contains invalid characters, then the user agent must raise a SYNTAX_ERR exception. [DOM3CORE] The components after this step consist of only US-ASCII characters.

The components of the array must then be converted to lowercase. Since only US-ASCII is involved at this step, this only requires converting characters in the range A-Z to the corresponding characters in the range a-z.

The resulting array is used in a comparison with another array, as described below. In addition, its components are concatenated together, each part separated by a dot (U+002E), to form the normalised requested domain.

If the original domain was "Åsgård.Example.Com", then the resulting array would have the three items "xn--sgrd-poac", "example", and "com", and the normalised requested domain would be "xn--sgrd-poac.example.com".

Next, the domain part of the tuple forming the calling script's origin is processed to find if it is allowed to access the requested domain.

If the script's origin has no domain part, e.g. if only the server's IP address is known, and the normalised requested domain is not the empty string, then the user agent must raise a security exception.

If the normalised requested domain is the empty string, then the rest of this algorithm can be skipped. This is because in that situation, the comparison of the two arrays below will always find them to be the same — the first array in such a situation is also empty and so permission to access that storage area will always be given.

If the domain part of the script's origin contains no dots (U+002E) then the string ".localdomain" must be appended to the script's domain.

Then, the domain part of the script's origin must be turned into an array, being split, converted to ASCII, and lowercased as described for the domain argument above.

Of the two arrays, the longest one must then be shortened to the length of the shorter one, by dropping items from the start of the array.

If the domain argument is "www.example.com" and the script origin's domain part is "example.com" then the first array will be a three item array ("www", "example", "com"), and the second will be a two item array ("example", "com"). The first array is therefore shortened, dropping the leading parts, making both into the same array ("example", "com").

If the two arrays are not component-for-component identical in literal string comparisons, then the user agent must then raise a security exception.

Otherwise, the user agent must check to see if it has allocated global storage area for the normalised requested domain. If it has not, a new storage area for that domain must be created.

The user agent must then create a Storage object associated with that domain's global storage area, and return it.

When the requested domain is a top level domain, or the empty string, or a country-specific sub-domain like "co.uk" or "ca.us", the associated global storage area is known as public storage area

The setItem() method might be called on a Storage object that is associated with a global storage area for a domain d, created by a StorageList object associated with a Window object x. Whenever this occurs, if the method didn't raise an exception, a storage event must be fired, as described below, in every HTMLDocument object that matches the following conditions:

Its Window object is not x, and
Its Window object's globalStorage attribute's StorageList object's namedItem() method would not raise a security exception according to the rules above if it was invoked with the domain d.

In other words, every other document that has access to that domain's global storage area is notified of the change.

4.11.6. The `storage` event

The storage event is fired in an HTMLDocument when a storage area changes, as described in the previous two sections (for session storage, for global storage).

When this happens, a storage event in no namespace, which bubbles, is not cancelable, has no default action, and which uses the StorageEvent interface described below, must be fired on the body element.

However, it is possible (indeed, for session storage areas, likely) that the target HTMLDocument object is not active at that time. For example, it might not be the current entry in the session history; user agents typically stop scripts from running in pages that are in the history. In such cases, the user agent must instead delay the firing of the event until such time as the HTMLDocument object in question becomes active again.

When there are multiple delayed storage events for the same HTMLDocument object, user agents should coalesce events with the same domain value (dropping duplicates).

If the DOM of a page that has delayed storage events queued up is discarded, then the delayed events are dropped as well.

interface StorageEvent : Event {
  readonly attribute DOMString domain;
  void initStorageEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString domainArg);
  void initStorageEventNS(in DOMString namespaceURIArg, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString domainArg);
};

The initStorageEvent() and initStorageEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The domain attribute of the StorageEvent event object must be set to the name of the domain associated with the storage area that changed if that storage area is a global storage area, or the string "#session" if it was a session storage area.

4.11.7. Miscellaneous implementation requirements for storage areas

4.11.7.1. Disk space

User agents should limit the total amount of space allowed for a domain based on the domain of the page setting the value.

User agents should not limit the total amount of space allowed on a per-storage-area basis, otherwise a site could just store data in any number of subdomains, e.g. storing up to the limit in a1.example.com, a2.example.com, a3.example.com, etc, circumventing per-domain limits.

User agents should consider additional quota mechanisms (for example limiting the amount of space provided to a domain's subdomains as a group) so that hostile authors can't run scripts from multiple subdomains all adding data to the global storage area in an attempted denial-of-service attack.

User agents may prompt the user when per-domain space quotas are reached, allowing the user to grant a site more space. This enables sites to store many user-created documents on the user's computer, for instance.

User agents should allow users to see how much space each domain is using.

If the storage area space limit is reached during a setItem() call, the user agent should raise an exception.

A mostly arbitrary limit of five megabytes per domain is recommended. Implementation feedback is welcome and will be used to update this suggestion in future.

4.11.7.2. Threads

Multiple browsing contexts must be able to access the global storage areas simultaneously in a predictable manner. Scripts must not be able to detect any concurrent script execution.

This is required to guarentee that the length attribute of a Storage object never changes while a script is executing, other than in a way that is predictable by the script itself.

There are various ways of implementing this requirement. One is that if a script running in one browsing context accesses a global storage area, the UA blocks scripts in other browsing contexts when they try to access any global storage area until the first script has executed to completion. (Similarly, when a script in one browsing context accesses its session storage area, any scripts that have the same top level browsing context and the same domain would block when accessing their session storage area until the first script has executed to completion.) Another (potentially more efficient but probably more complex) implementation strategy is to use optimistic transactional script execution. This specification does not require any particular implementation strategy, so long as the requirement above is met.

4.11.8. Security and privacy

4.11.8.1. User tracking

A third-party advertiser (or any entity capable of getting content distributed to multiple sites) could use a unique identifier stored in its domain's global storage area to track a user across multiple sessions, building a profile of the user's interests to allow for highly targeted advertising. In conjunction with a site that is aware of the user's real identity (for example an e-commerce site that requires authenticated credentials), this could allow oppressive groups to target individuals with greater accuracy than in a world with purely anonymous Web usage.

The globalStorage object also introduces a way for sites to cooperate to track users over multiple domains, by storing identifying data in "public" top-level domain storage area, accessible by any domain.

There are a number of techniques that can be used to mitigate the risk of user tracking:

Blocking third-party storage: user agents may restrict access to the globalStorage object to scripts originating at the domain of the top-level document of the browsing context.

This blocks a third-party site from using its private storage area for tracking a user, but top-level sites could still cooperate with third parties to perferm user tracking by using the "public" storage area.
Expiring stored data: user agents may automatically delete stored data after a period of time.

For example, a user agent could treat the global storage area as session-only storage, deleting the data once the user had closed all the browsing contexts that could access it.

This can restrict the ability of a site to track a user, as the site would then only be able to track the user across multiple sessions when he authenticates with the site itself (e.g. by making a purchase or logging in to a service).
Blocking access to the top-level domain ("public") storage areas: user agents may prevent domains from storing data in and reading data from the top-level domain entries in the globalStorage object.

In practice this requires a detailed list of all the "public" second-level (and third-level) domains. For example, content at the domain www.example.com would be allowed to access example.com data but not com data; content at the domain example.co.uk would be allowed access to example.co.uk but not co.uk or uk; and content at example.chiyoda.tokyo.jp would be allowed access to example.chiyoda.tokyo.jp but not chiyoda.tokyo.jp, tokyo.jp, or jp, while content at example.metro.tokyo.jp would be allowed access to both example.metro.tokyo.jp and metro.tokyo.jp but not tokyo.jp or jp. The problem is even more convoluted when one considers private domains with third-party subdomains such as dyndns.org or uk.com.

Blocking access to the "public" storage areas can also prevent innocent sites from cooperating to provide services beneficial to the user.
Treating persistent storage as cookies: user agents may present the persistent storage feature to the user in a way that does not distinguish it from HTTP session cookies. [RFC2965]

This might encourage users to view persistent storage with healthy suspicion.
Site-specific white-listing of access to "public" storage area: user agents may allow sites to access persistent storage for their own domain and subdomains in an unrestricted manner, but require the user to authorise access to the storage area of higher-level domains.

For example, code at example.com would be always allowed to read and write data for www.example.com and example.com, but if it tried to access com, the user agent could display a non-modal message informing the user that the page requested access to com and offering to allow it.
Origin-tracking of persistent storage data: user agents may record the domain of the script that caused data to be stored.

If this information is then used to present the view of data currently in persistent storage, it would allow the user to make informed decisions about which parts of the persistent storage to prune. Combined with a blacklist ("delete this data and prevent this domain from ever storing data again"), the user can restrict the use of persistent storage to sites that he trusts.
Shared blacklists: user agents may allow users to share their persistent storage domain blacklists.

This would allow communities to act together to protect their privacy.

While these suggestions prevent trivial use of this API for user tracking, they do not block it altogether. Within a single domain, a site can continue to track the user across multiple sessions, and can then pass all this information to the third party along with any identifying information (names, credit card numbers, addresses) obtained by the site. If a third party cooperates with multiple sites to obtain such information, a profile can still be created.

However, user tracking is to some extent possible even with no cooperation from the user agent whatsoever, for instance by using session identifiers in URIs, a technique already commonly used for innocuous purposes but easily repurposed for user tracking (even retroactively). This information can then be shared with other sites, using using visitors' IP addresses and other user-specific data (e.g. user-agent headers and configuration settings) to combine separate sessions into coherent user profiles.

If the user interface for persistent storage presents data in the persistent storage feature separately from data in HTTP session cookies, then users are likely to delete data in one and not the other. This would allow sites to use the two features as redundant backup for each other, defeating a user's attempts to protect his privacy.

4.11.8.3. Integrity of "public" storage areas

Since the "public" global storage areas are accessible by content from many different parties, it is possible for third-party sites to delete or change information stored in those areas in ways that the originating sites may not expect.

Authors must not use the "public" global storage areas for storing sensitive data. Authors must not trust information stored in "public" global storage areas.

4.11.8.4. Cross-protocol and cross-port attacks

This API makes no distinction between content served over HTTP, FTP, or other host-based protocols, and does not distinguish between content served from different ports at the same host.

Thus, for example, data stored in the global persistent storage for domain "www.example.com" by a page served from HTTP port 80 will be available to a page served in http://example.com:18080/, even if the latter is an experimental server under the control of a different user.

Since the data is not sent over the wire by the user agent, this is not a security risk in its own right. However, authors must take proper steps to ensure that all hosts that have fully qualified host names that are subsets of hosts dealing with sensitive information are as secure as the originating hosts themselves.

Similarly, authors must ensure that all Web servers on a host, regardless of the port, are equally trusted if any of them are to use persistent storage. For instance, if a Web server runs a production service that makes use of the persistent storage feature, then other users that have access to that machine and that can run a Web server on another port will be able to access the persistent storage added by the production service (assuming they can trick a user into visiting their page).

However, if one is able to trick users into visiting a Web server with the same host name but on a different port as a production service used by these users, then one could just as easily fake the look of the site and thus trick users into authenticating with the fake site directly, forwarding the request to the real site and stealing the credentials in the process. Thus, the persistent storage feature is considered to only minimally increase the risk involved.

What about if someone is able to get a server up on a port, and can then send people to that URI? They could steal all the data with no further interaction. How about putting the port number at the end of the string being compared? (Implicitly.)

4.11.8.5. DNS spoofing attacks

Because of the potential for DNS spoofing attacks, one cannot guarentee that a host claiming to be in a certain domain really is from that domain. The secure attribute is provided to mark certain key/value pairs as only being accessible to pages that have been authenticated using secure certificates (or similar mechanisms).

Authors must ensure that they do not mark sensitive items as "safe for both secure and insecure content". (To prevent the risk of a race condition, data stored by scripts in secure contexts default to being marked as "safe only for secure content".)

4.11.8.6. Cross-directory attacks

Different authors sharing one host name, for example users hosting content on geocities.com, all share one persistent storage object. There is no feature to restrict the access by pathname. Authors on shared hosts are therefore recommended to avoid using the persistent storage feature, as it would be trivial for other authors to read from and write to the same storage area.

Even if a path-restriction feature was made available, the usual DOM scripting security model would make it trivial to bypass this protection and access the data from any path.

4.11.8.7. Public storage areas corresponding to hosts

If a "public" global storage area corresponds to a host, as it typically does if for private domains with third-party subdomains such as dyndns.org or uk.com, the host corresponding to the "public" domain has access to all the storage areas of its third-party subdomains. In general, authors are discouraged from using the globalStorage API for sensitive data unless the operators of all the domains involved are trusted.

User agents may mitigate this problem by preventing hosts corresponding to "public" global storage areas from accessing any storage areas other than their own.

4.11.8.8. Storage areas in the face of untrusted higher-level domains that do not correspond to public storage areas

Authors should not store sensitive data using the global storage APIs if there are hosts with fully-qualified domain names that are subsets of their own which they do not trust. For example, an author at finance.members.example.net should not store sensitive financial user data in the finance.members.example.net storage area if he does not trust the host that runs example.net.

4.11.8.9. Storage areas in the face of untrusted subdomains

If an author publishing content on one host, e.g. example.com, wishes to use the globalStorage API but does not wish any content on the host's subdomains to access the data, the author should use an otherwise non-existent subdomain name, e.g., private.example.com, to store the data. This will be accessible only to that host (and its parent domains), and not to any of the real subdomains (e.g. upload.example.com).

4.11.8.10. Implementation risks

The two primary risks when implementing this persistent storage feature are letting hostile sites read information from other domains, and letting hostile sites write information that is then read from other domains.

Letting third-party sites read data that is not supposed to be read from their domain causes information leakage, For example, a user's shopping wishlist on one domain could be used by another domain for targeted advertising; or a user's work-in-progress confidential documents stored by a word-processing site could be examined by the site of a competing company.

Letting third-party sites write data to the storage areas of other domains can result in information spoofing, which is equally dangerous. For example, a hostile site could add items to a user's wishlist; or a hostile site could set a user's session identifier to a known ID that the hostile site can then use to track the user's actions on the victim site.

A risk is also presented by servers on local domains having host names matching top-level domain names, for instance having a host called "com" or "net". Such hosts might, if implementations fail to correctly implement the .localdomain suffixing, have full access to all the data stored in a UA's persistent storage for that top level domain.

Thus, strictly following the model described in this specification is important for user security.

In addition, a number of optional restrictions related to the "public" global storage areas are suggested in the previous sections. The design of this API is intended to be such that not supporting these restrictions, or supporting them less than perfectly, does not result in critical security problems. However, implementations are still encouraged to create and maintain a list of "public" domains, and apply the restrictions described above.

5. Editing

This section describes various features that allow authors to enable users to edit documents and parts of documents interactively.

5.1. Introduction

This section is non-normative.

Would be nice to explain how these features work together.

5.2. The `contenteditable` attribute

The contenteditable attribute is a common attribute. User agents must support this attribute on all HTML elements.

The contenteditable attribute is an enumerated attribute whose keywords are the empty string, true, and false. The empty string and the true keyword map to the true state. The false keyword maps to the false state, which is also the invalid value default. There is no missing value default.

If an HTML element has a contenteditable attribute set to the true state, or if its nearest ancestor with the contenteditable attribute set has its attribute set to the true state, or if it has no ancestors with the contenteditable attribute set but the Document has designMode enabled, then the UA must treat the element as editable (as described below).

Otherwise, either the HTML element has a contenteditable attribute set to the false state, or its nearest ancestor with the contenteditable attribute set is not editable, or it has no ancestor with the contenteditable attribute set and the Document itself has designMode disabled, and the element is thus not editable.

The contentEditable DOM attribute, on getting, must return the string "inherit" if the content attribute isn't set, "true" if the attribute is set and has the true state, and "false" otherwise. On setting, if the new value is case-insensitively equal to the string "inherit" then the content attribute must be removed, if the new value is case-insensitively equal to the string "true then the content attribute must be set to the string "true, if the new value is case-insensitively equal to the string "false then the content attribute must be set to the string "false, and otherwise the attribute setter must raise a SYNTAX_ERR exception.

If an element is editable and its parent element is not, or if an element is editable and it has no parent element, then the element is an editing host. Editable elements can be nested. User agents must make editing hosts focusable (which typicially means they enter the tab order). An editing host can contain non-editable sections, these are handled as described below. An editing host can contain non-editable sections that contain further editing hosts.

When an editing host has focus, it must have a caret position that specifies where the current editing position is. It may also have a selection.

How the caret and selection are represented depends entirely on the UA.

5.2.1. User editing actions

There are several actions that the user agent should allow the user to perform while the user is interacting with an editing host. How exactly each action is triggered is not defined for every action, but when it is not defined, suggested key bindings are provided to guide implementors.

Move the caret

User agents must allow users to move the caret to any position within an editing host, even into nested editable elements. This could be triggered as the default action of keydown events with various key identifiers and as the default action of mouseydown events.

Change the selection

User agents must allow users to change the selection within an editing host, even into nested editable elements. This could be triggered as the default action of keydown events with various key identifiers and as the default action of mouseydown events.

Insert text

This action must be triggered as the default action of a textInput event, and may be triggered by other commands as well. It must cause the user agent to insert the specified text (given by the event object's data attribute in the case of the textInput event) at the caret.

If the caret is positioned somewhere where inline-level content is not allowed (e.g. because the element accepts "both block-level and inline-level content but not both", and the element already contains block-level content), then the user agent must not insert the text directly at the caret position. In such cases the behaviour is UA-dependent, but user agents must not, in response to a request to insert text, generate a DOM that is less conformant than the DOM prior to the request.

User agents should allow users to insert new paragraphs into elements that only contain block-level content.

For example, given the markup:

<section>
 <dl>
  <dt> Ben </dt>
  <dd> Goat </dd>
 </dl>
</section>

...the user agent should allow the user to insert p elements before and after the dl element, as children of the section element.

Break block

UAs should offer a way for the user to request that the current block be broken at the caret, e.g. as the default action of a keydown event whose identifier is the "Enter" key and that has no modifiers set.

The exact behaviour is UA-dependent, but user agents must not, in response to a request to break a block, generate a DOM that is less conformant than the DOM prior to the request.

Insert a line separator

UAs should offer a way for the user to request an explicit line break at the caret position without breaking the block, for example as in a poem verse or an address. To insert a line break, the user agent must insert a br element.

If the caret is positioned somewhere where inline-level content is not allowed (e.g. because the element accepts "both block-level and inline-level content but not both", and the element already contains block-level content), then the user agent must not insert the br element directly at the caret position. In such cases the behaviour is UA-dependent, but user agents must not, in response to a request to insert a line separator, generate a DOM that is less conformant than the DOM prior to the request.

Delete

UAs should offer a way for the user to delete text and elements, e.g. as the default action of keydown events whose identifiers are "U+0008" or "U+007F".

Five edge cases in particular need to be considered carefully when implementing this feature: backspacing at the start of an element, backspacing when the caret is immediately after an element, forward-deleting at the end of an element, forward-deleting when the caret is immediately before an element, and deleting a selection whose start and end points do not share a common parent node.

In any case, the exact behaviour is UA-dependent, but user agents must not, in response to a request to delete text or an element, generate a DOM that is less conformant than the DOM prior to the request.

Insert, and wrap text in, semantic elements

UAs should offer a way for the user to mark text as having stress emphasis and as being important, and may offer the user the ability to mark text and blocks with other semantics.

UAs should similarly offer a way for the user to insert empty semantic elements (such as, again, em, strong, and others) to subsequently fill by entering text manually.

UAs should also offer a way to remove those semantics from marked up text, and to remove empty semantic element that have been inserted.

The exact behaviour is UA-dependent, but user agents must not, in response to a request to wrap semantics around some text or to insert or remove a semantic element, generate a DOM that is less conformant than the DOM prior to the request.

Select and move non-editable elements nested inside editing hosts

UAs should offer a way for the user to move images and other non-editable parts around the content within an editing host. This may be done using the drag and drop mechanism. User agents must not, in response to a request to move non-editable elements nested inside editing hosts, generate a DOM that is less conformant than the DOM prior to the request.

Edit form controls nested inside editing hosts

When an editable form control is edited, the changes must be reflected in both its current value and its default value. For input elements this means updating the defaultValue DOM attribute as well as the value DOM attribute; for select elements it means updating the option elements' defaultSelected DOM attribute as well as the selected DOM attribute; for textarea elements this means updating the defaultValue DOM attribute as well as the value DOM attribute. (Updating the default* DOM attributes causes content attributs to be updated as well.)

User agents may perform several commands per user request; for example if the user selects a block of text and hits Enter, the UA might interpret that as a request to delete the content of the selection followed by a request to break the block at that position.

5.2.2. Making entire documents editable

Documents have a designMode, which can be either enabled or disabled.

The designMode DOM attribute on the Document object takes takes two values, "on" and "off". When it is set, the new value must be case-insensitively compared to these two values. If it matches the "on" value, then designMode must be enabled, and if it matches the "off" value, then designMode must be disabled. Other values must be ignored.

When designMode is enabled, the DOM attribute must return the value "on", and when it is disabled, it must return the value "off".

The last state set must persist until the document is destroyed or the state is changed. Initially, documents must have their designMode disabled.

Enabling designMode causes scripts in general to be disabled and the document to become editable.

When the Document has designMode enabled, event listeners registered on the document or any elements owned by the document must do nothing.

5.3. Drag and drop

This section defines an event-based drag-and-drop mechanism.

This specification does not define exactly what a drag-and-drop operation actually is.

On a visual medium with a pointing device, a drag operation could be the default action of a mousedown event that is followed by a series of mousemove events, and the drop could be triggered by the mouse being released.

On media without a pointing device, the user would probably have to explicitly indicate his intention to perform a drag-and-drop operation, stating what he wishes to drag and what he wishes to drop, respectively.

However it is implemented, drag-and-drop operations must have a starting point (e.g. where the mouse was clicked, or the start of the selection or element that was selected for the drag), may have any number of intermediate steps (elements that the mouse moves over during a drag, or elements that the user picks as possible drop points as he cycles through possibilities), and must either have an end point (the element above which the mouse button was released, or the element that was finally selected), or be canceled. The end point must be the last element selected as a possible drop point before the drop occurs (so if the operation is not canceled, there must be at least one element in the middle step).

5.3.1. The `DragEvent` and `DataTransfer` interfaces

The drag-and-drop processing model involves several events. They all use the DragEvent interface.

interface DragEvent : Event {
  readonly attribute DataTransfer dataTransfer;
  void initDragEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg);
  void initDragEventNS(in DOMString namespaceURIArg, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg);
};

The initDragEvent() and initDragEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The dataTransfer attribute of the DragEvent interface represents the context information for the event.

When a DragEvent object is created, a new DataTransfer object must be created and assigned to the dataTransfer context information field of the event object.

interface DataTransfer {
           attribute DOMString dropEffect;
           attribute DOMString effectAllowed;
  void clearData(in DOMString format);
  void setData(in DOMString format, in DOMString data);
  DOMString getData(in DOMString format);
  void setDragImage(in Element image, in long x, in long y);
  void addElement(in Element element);
};

DataTransfer objects can conceptually contain various kinds of data.

When a DragEvent event object is initialised, the DataTransfer object created for the event's dataTransfer member must be initialised as follows:

The DataTransfer object must initially contain no data, no elements, and have no associated image.
The DataTransfer object's effectAllowed attribute must be set to "uninitialized".
The dropEffect attribute must be set to "none".

The dropEffect attribute controls the drag-and-drop feedback that the user is given during a drag-and-drop operation.

The attribute must ignore any attempts to set it to a value other than none, copy, link, and move. On getting, the attribute must return the last of those four values that it was set to.

The effectAllowed attribute is used in the drag-and-drop processing model to initialise the dropEffect attribute during the dragenter and dragover events.

The attribute must ignore any attempts to set it to a value other than none, copy, copyLink, copyMove, link, linkMove, move, all, and uninitialized. On getting, the attribute must return the last of those values that it was set to.

DataTransfer objects can hold pieces of data, each associated with a unique format. Formats are generally given by MIME types, with some values special-cased for legacy reasons.

The clearData(format) method must clear the DataTransfer object of any data associated with the given format. If format is the value "Text", then it must be treated as "text/plain". If the format is "URL", then it must be treated as "text/uri-list".

The setData(format, data) method must add data to the data stored in the DataTransfer object, labelled as being of the type format. This must replace any previous data that had been set for that format. If format is the value "Text", then it must be treated as "text/plain". If the format is "URL", then it must be treated as "text/uri-list".

The getData(format) method must return the data that is associated with the type format, if any, and must return the empty string otherwise. If format is the value "Text", then it must be treated as "text/plain". If the format is "URL", then the data associated with the "text/uri-list" format must be parsed as appropriate for text/uri-list data, and the first URI from the list must be returned. If there is no data with that format, or if there is but it has no URIs, then the method must return the empty string. [RFC2483]

The setDragImage(element, x, y) method sets which element to use to generate the drag feedback. The element argument can be any Element; if it is an img element, then the user agent should use the element's image (at its intrinsic size) to generate the feedback, otherwise the user agent should base the feedback on the given element (but the exact mechanism for doing so is not specified).

The addElement(element) method is an alternative way of specifying how the user agent is to render the drag feedback. It adds an element to the DataTransfer object.

5.3.2. Events fired during a drag-and-drop action

The following events are involved in the drag-and-drop model. Whenever the processing model described below causes one of these events to be fired, the event fired must use the DragEvent interface defined above, must have the bubbling and cancelable behaviours given in the table below, and must have the context information set up as described after the table.

Event Name	Target	Bubbles?	Cancelable?	`dataTransfer`	`effectAllowed`	`dropEffect`	Default Action
`dragstart`	Source node	✓ Bubbles	✓ Cancelable	Contains source node unless a selection is being dragged, in which case it is empty	`uninitialized`	`none`	Initiate the drag-and-drop operation
`drag`	Source node	✓ Bubbles	✓ Cancelable	Empty	Same as last event	`none`	Continue the drag-and-drop operation
`dragenter`	Immediate user selection or the body element	✓ Bubbles	✓ Cancelable	Empty	Same as last event	Based on `effectAllowed` value	Reject immediate user selection as potential target element
`dragleave`	Previous target element	✓ Bubbles	—	Empty	Same as last event	`none`	None
`dragover`	Current target element	✓ Bubbles	✓ Cancelable	Empty	Same as last event	Based on `effectAllowed` value	Reset the current drag operation to "none"
`drop`	Current target element	✓ Bubbles	✓ Cancelable	`getData()` returns data set in `dragstart` event	Same as last event	Current drag operation	Varies
`dragend`	Source node	✓ Bubbles	—	Empty	Same as last event	Current drag operation	Varies

The dataTransfer object's contents are empty except for dragstart events and drop events, for which the contents are set as described in the processing model, below.

The effectAllowed attribute must be set to "uninitialized" for dragstart events, and to whatever value the field had after the last drag-and-drop event was fired for all other events (only counting events fired by the user agent for the purposes of the drag-and-drop model described below).

The dropEffect attribute must be set to "none" for dragstart, drag, dragleave, and dragend events (except when stated otherwise in the algorithms given in the sections below), to the value corresponding to the current drag operation for drop events, and to a value based on the effectAllowed attribute's value and to the drag-and-drop source, as given by the following table, for the remaining events (dragenter and dragover):

`effectAllowed`	`dropEffect`
`none`	`none`
`copy`, `copyLink`, `copyMove`, `all`	`copy`
`link`, `linkMove`	`link`
`move`	`move`
`uninitialized` when what is being dragged is a selection from a text field	`move`
`uninitialized` when what is being dragged is a selection	`copy`
`uninitialized` when what is being dragged is an `a` element with an `href` attribute	`link`
Any other case	`copy`

5.3.3. Drag-and-drop processing model

When the user attempts to begin a drag operation, the user agent must first determine what is being dragged. If the drag operation was invoked on a selection, then it is the selection that is being dragged. Otherwise, it is the first element, going up the ancestor chain, starting at the node that the user tried to drag, that has the DOM attribute draggable set to true. If there is no such element, then nothing is being dragged, the drag-and-drop operation is never started, and the user agent must not continue with this algorithm.

img elements and a elements with an href attribute have their draggable attribute set to true by default.

If the user agent determines that something can be dragged, a dragstart event must then be fired.

If it is a selection that is being dragged, then this event must be fired on the node that the user started the drag on (typically the text node that the user originally clicked). If the user did not specify a particular node, for example if the user just told the user agent to begin a drag of "the selection", then the event must be fired on the deepest node that is a common ancestor of all parts of the selection.

If it is not a selection that is being dragged, then the event must be fired on the element that is being dragged.

The node on which the event is fired is the source node. Multiple events are fired on this node during the course of the drag-and-drop operation.

If it is a selection that is being dragged, the dataTransfer member of the event must be created with no nodes. Otherwise, it must be created containing just the source node. Script can use the addElement() method to add further elements to the list of what is being dragged.

If it is a selection that is being dragged, the dataTransfer member of the event must have the text of the selection added to it as the data associated with the text/plain format. Otherwise, if it is an img element being dragged, then the value of the element's src DOM attribute must be added, associated with the text/uri-list format. Otherwise, if it is an a element being dragged, then the value of the element's href DOM attribute must be added, associated with the text/uri-list format. Otherwise, no data is added to the object by the user agent.

If the event is canceled, then the drag-and-drop operation must not occur; the user agent must not continue with this algorithm.

If it is not canceled, then the drag-and-drop operation must be initiated.

Since events with no event handlers registered are, almost by definition, never canceled, drag-and-drop is always available to the user if the author does not specifically prevent it.

The drag-and-drop feedback must be generated from the first of the following sources that is available:

The element specified in the last call to the setDragImage() method of the dataTransfer object of the dragstart event, if the method was called. In visual media, if this is used, the x and y arguments that were passed to that method should be used as hints for where to put the cursor relative to the resulting image. The values are expressed as distances in CSS pixels from the left side and from the top side of the image respectively. [CSS21]
The elements that were added to the dataTransfer object, both before the event was fired, and during the handling of the event using the addElement() method, if any such elements were indeed added.
The selection that the user is dragging.

The user agent must take a note of the data that was placed in the dataTransfer object. This data will be made available again when the drop event is fired.

From this point until the end of the drag-and-drop operation, device input events (e.g. mouse and keyboard events) must be suppressed. In addition, the user agent must track all DOM changes made during the drag-and-drop operation, and add them to its undo history as one atomic operation once the drag-and-drop operation has ended.

During the drag operation, the element directly indicated by the user as the drop target is called the immediate user selection. (Only elements can be selected by the user; other nodes must not be made available as drop targets.) However, the immediate user selection is not necessarily the current target element, which is the element currently selected for the drop part of the drag-and-drop operation. The immediate user selection changes as the user selects different elements (either by pointing at them with a pointing device, or by selecting them in some other way). The current target element changes when the immediate user selection changes, based on the results of event handlers in the document, as described below.

Both the current target element and the immediate user selection can be null, which means no target element is selected. They can also both be elements in other (DOM-based) documents, or other (non-Web) programs altogether. (For example, a user could drag text to a word-processor.) The current target element is initially null.

In addition, there is also a current drag operation, which can take on the values "none", "copy", "link", and "move". Initially it has the value "none". It is updated by the user agent as described in the steps below.

User agents must, every 350ms (±200ms), perform the following steps in sequence. (If the user agent is still performing the previous iteration of the sequence when the next iteration becomes due, the user agent must not execute the overdue iteration, effectively "skipping missed frames" of the drag-and-drop operation.)

First, the user agent must fire a drag event at the source node. If this event is canceled, the user agent must set the current drag operation to none (no drag operation).

Next, if the drag event was not canceled and the user has not ended the drag-and-drop operation, the user agent must check the state of the drag-and-drop operation, as follows:

First, if the user is indicating a different immediate user selection than during the last iteration (or if this is the first iteration), and if this immediate user selection is not the same as the current target element, then the current target element must be updated, as follows:
1. If the new immediate user selection is null, or is in a non-DOM document or application, then set the current target element to the same value.
2. Otherwise, the user agent must fire a dragenter event at the immediate user selection.
3. If the event is canceled, then the current target element must be set to the immediate user selection.
4. Otherwise, if the current target element is not the body element, the user agent must fire a dragenter event at the body element, and the current target element must be set to the body element, regardless of whether that event was canceled or not. (If the body element is null, then the current target element would be set to null too in this case, it wouldn't be set to the Document object.)
If the previous step caused the current target element to change, and if the previous target element was not null or a part of a non-DOM document, the user agent must fire a dragleave event at the previous target element.

If the current target element is a DOM element, the user agent must fire a dragover event at this current target element.

If the dragover event is canceled, the current drag operation must be reset to "none".

Otherwise, the current drag operation must be set based on the values the effectAllowed and dropEffect attributes of the dataTransfer object had after the event was handled, as per the following table:

`effectAllowed`	`dropEffect`	Drag operation
`uninitialized`, `copy`, `copyLink`, `copyMove`, or `all`	`copy`	"copy"
`uninitialized`, `link`, `copyLink`, `linkMove`, or `all`	`link`	"link"
`uninitialized`, `move`, `copyMove`, `linkMove`, or `all`	`move`	"move"
Any other case		"none"

Then, regardless of whether the dragover event was canceled or not, the drag feedback (e.g. the mouse cursor) must be updated to match the current drag operation, as follows:

Drag operation	Feedback
"copy"	Data will be copied if dropped here.
"link"	Data will be linked if dropped here.
"move"	Data will be moved if dropped here.
"none"	No operation allowed, dropping here will cancel the drag-and-drop operation.

Otherwise, if the current target element is not a DOM element, the user agent must use platform-specific mechanisms to determine what drag operation is being performed (none, copy, link, or move). This sets the current drag operation.

Otherwise, if the user ended the drag-and-drop operation (e.g. by releasing the mouse button in a mouse-driven drag-and-drop interface), or if the drag event was canceled, then this will be the last iteration. The user agent must follow the following steps, then stop looping.
1. If the current drag operation is none (no drag operation), or, if the user ended the drag-and-drop operation by canceling it (e.g. by hitting the Escape key), or if the current target element is null, then the drag operation failed. If the current target element is a DOM element, the user agent must fire a dragleave event at it; otherwise, if it is not null, it must use platform-specific conventions for drag cancellation.
2. Otherwise, the drag operation was as success. If the current target element is a DOM element, the user agent must fire a drop event at it; otherwise, it must use platform-specific conventions for indicating a drop.
  
  When the target is a DOM element, the dropEffect attribute of the event's dataTransfer object must be given the value representing the current drag operation (copy, link, or move), and the object must be set up so that the getData() method will return the data that was added during the dragstart event.
  
  If the event is canceled, the current drag operation must be set to the value of the dropEffect attribute of the event's dataTransfer object as it stood after the event was handled.
  
  Otherwise, the event is not canceled, and the user agent must perform the event's default action, which depends on the exact target as follows:
  
  If the current target element is a text field (e.g. textarea, or an input element with type="text")
  The user agent must insert the data associated with the text/plain format, if any, into the text field in a manner consistent with platform-specific conventions (e.g. inserting it at the current mouse cursor position, or inserting it at the end of the field).
  Otherwise
  Reset the current drag operation to "none".
3. Finally, the user agent must fire a dragend event at the source node, with the dropEffect attribute of the event's dataTransfer object being set to the value corresponding to the current drag operation.
  
  The current drag operation can change during the processing of the drop event, if one was fired.
  
  The event is not cancelable. After the event has been handled, the user agent must act as follows:
  
  If the current target element is a text field (e.g. textarea, or an input element with type="text"), and a drop event was fired in the previous step, and the current drag operation is "move", and the source of the drag-and-drop operation is a selection in the DOM
  The user agent should delete the range representing the dragged selection from the DOM.
  If the current target element is a text field (e.g. textarea, or an input element with type="text"), and a drop event was fired in the previous step, and the current drag operation is "move", and the source of the drag-and-drop operation is a selection in a text field
  The user agent should delete the dragged selection from the relevant text field.
  Otherwise
  The event has no default action.

5.3.3.1. When the drag-and-drop operation starts or ends in another document

The model described above is independent of which Document object the nodes involved are from; the events must be fired as described above and the rest of the processing model must be followed as described above, irrespective of how many documents are involved in the operation.

5.3.3.2. When the drag-and-drop operation starts or ends in another application

If the drag is initiated in another application, the source node is not a DOM node, and the user agent must use platform-specific conventions instead when the requirements above involve the source node. User agents in this situation must act as if the dragged data had been added to the DataTransfer object when the drag started, even though no dragstart event was actually fired; user agents must similarly use platform-specific conventions when deciding on what drag feedback to use.

If a drag is started in a document but ends in another application, then the user agent must instead replace the parts of the processing model relating to handling the target according to platform-specific conventions.

In any case, scripts running in the context of the document must not be able to distinguish the case of a drag-and-drop operation being started or ended in another application from the case of a drag-and-drop operation being started or ended in another document from another domain.

5.3.4. The `draggable` attribute

All elements may have the draggable content attribute set. The draggable attribute is an enumerated attribute. It has three states. The first state is true and it has the keyword true. The second state is false and it has the keyword false. The third state is auto; it has no keywords but it is the missing value default.

The draggable DOM attribute, whose value depends on the content attribute's in the way described below, controls whether or not the element is draggable. Generally, only text selections are draggable, but elements whose draggable DOM attribute is true become draggable as well.

If an element's draggable content attribute has the state true, the draggable DOM attribute must return true.

Otherwise, if the element's draggable content attribute has the state false, the draggable DOM attribute must return false.

Otherwise, the element's draggable content attribute has the state auto. If the element is an img element, or, if the element is an a element with an href content attribute, the draggable DOM attribute must return true.

Otherwise, the draggable DOM must return false.

If the draggable DOM attribute is set to the value false, the draggable content attribute must be set to the literal value false. If the draggable DOM attribute is set to the value true, the draggable content attribute must be set to the literal value true.

5.3.5. Copy and paste

Copy-and-paste is a form of drag-and-drop: the "copy" part is equivalent to dragging content to another application (the "clipboard"), and the "paste" part is equivalent to dragging content from another application.

Select-and-paste (a model used by mouse operations in the X Window System) is equivalent to a drag-and-drop operation where the source is the selection.

5.3.5.1. Copy to clipboard

When the user invokes a copy operation, the user agent must act as if the user had invoked a drag on the current selection. If the drag-and-drop operation initiates, then the user agent must act as if the user had indicated (as the immediate user selection) a hypothetical application representing the clipbroad. Then, the user agent must act as if the user had ended the drag-and-drop operation without canceling it. If the drag-and-drop operation didn't get canceled, the user agent should then follow the relevant platform-specific conventions for copy operations (e.g. updating the clipboard).

5.3.5.2. Cut to clipboard

When the user invokes a cut operation, the user agent must act as if the user had invoked a copy operation (see the previous section), followed, if the copy was completed successfully, by a selection delete operation.

5.3.5.3. Paste from clipboard

When the user invokes a clipboard paste operation, the user agent must act as if the user had invoked a drag on a hypothetical application representing the clipboard, setting the data associated with the drag as the text from the keyboard (either as text/plain or text/uri-list). If the contents of the clipboard cannot be represented as text or URIs, then the paste operation must not have any effect.

Then, the user agent must act as if the user had indicated (as the immediate user selection) the element with the keyboard focus, and then ended the drag-and-drop operation without canceling it.

5.3.5.4. Paste from selection

When the user invokes a selection paste operation, the user agent must act as if the user had invoked a drag on the current selection, then indicated (as the immediate user selection) the element with the keyboard focus, and then ended the drag-and-drop operation without canceling it.

If the contents of the selection cannot be represented as text or URIs, then the paste operation must not have any effect.

5.3.6. Security risks in the drag-and-drop model

User agents must not make the data added to the DataTransfer object during the dragstart event available to scripts until the drop event, because otherwise, if a user were to drag sensitive information from one document to a second document, crossing a hostile third document in the process, the hostile document could intercept the data.

For the same reason, user agents must only consider a drop to be successful if the user specifically ended the drag operation — if any scripts end the drag operation, it must be considered unsuccessful (canceled) and the drop event must not be fired.

User agents should take care to not start drag-and-drop operations in response to script actions. For example, in a mouse-and-window environment, if a script moves a window while the user has his mouse button depressed, the UA would not consider that to start a drag. This is important because otherwise UAs could cause data to be dragged from sensitive sources and dropped into hostile documents without the user's consent.

5.4. Undo history

There has got to be a better way of doing this, surely.

The user agent must associate an undo transaction history with each HTMLDocument object.

The undo transaction history is a list of entries. The entries are of two type: DOM changes and undo objects.

Each DOM changes entry in the undo transaction history consists of batches of one or more of the following:

Changes to the content attributes of an Element node.
Changes to the DOM attributes of a Node.
Changes to the DOM hierarchy of nodes that are descendants of the HTMLDocument object (parentNode, childNodes).

Undo object entries consist of objects representing state that scripts running in the document are managing. For example, a Web mail application could use an undo object to keep track of the fact that a user has moved an e-mail to a particular folder, so that the user can undo the action and have the e-mail return to its former location.

Broadly speaking, DOM changes entries are handled by the UA in response to user edits of form controls and editing hosts on the page, and undo object entries are handled by script in response to higher-level user actions (such as interactions with server-side state, or in the implementation of a drawing tool).

5.4.1. The `UndoManager` interface

This API sucks. Seriously. It's a terrible API. Really bad. I hate it. Here are the requirements:

Has to cope with cases where the server has undo state already when the page is loaded, that can be stuffed into the undo buffer onload.
Has to support undo/redo.
Has to cope with the "undo" action being "contact the server and tell it to undo", rather than it being the opposite of the "redo" action.
Has to cope with some undo states expiring from the undo history (e.g. server can only remember one undelete action) but other states not expiring (e.g. client can undo arbitrary amounts of local edits).

To manage undo object entries in the undo transaction history, the UndoManager interface can be used:

interface UndoManager {
  unsigned long add(in DOMObject data, in DOMStrong title);
  void remove(in unsigned long index);
  void clearUndo();
  void clearRedo();
  DOMObject item(in unsigned long index);
  readonly attribute unsigned long length;
  readonly attribute unsigned long position;
};

The undoManager attribute of the Window interface must return the object implementing the UndoManager interface for that Window object's associated HTMLDocument object.

In the ECMAScript DOM binding, objects implementing this interface must also support being dereferenced using the square bracket notation, such that dereferencing with an integer index is equivalent to invoking the item() method with that index (e.g. undoManager[1] returns the same as undoManager.item(1)).

UndoManager objects represent their document's undo transaction history. Only undo object entries are visible with this API, but this does not mean that DOM changes entries are absent from the undo transaction history.

The length attribute must return the number of undo object entries in the undo transaction history.

The item(n) method must return the nth undo object entry in the undo transaction history.

The undo transaction history has a current position. This is the position between two entries in the undo transaction history's list where the previous entry represents what needs to happen if the user invokes the "undo" command (the "undo" side, lower numbers), and the next entry represents what needs to happen if the user invokes the "redo" command (the "redo" side, higher numbers).

The position attribute must return the index of the undo object entry nearest to the undo position, on the "redo" side. If there are no undo object entries on the "redo" side, then the attribute must return the same as the length attribute. If there are no undo object entries on the "undo" side of the undo position, the position attribute returns zero.

Since the undo transaction history contains both undo object entries and DOM changes entries, but the position attribute only returns indices relative to undo object entries, it is possible for several "undo" or "redo" actions to be performed without the value of the position attribute changing.

The add(data, title) method's behaviour depends on the current state. Normally, it must insert the data object passed as an argument into the undo transaction history immediately before the undo position, optionally remembering the given title to use in the UI. If the method is called during an undo operation, however, the object must instead be added immediately after the undo position.

If the method is called and there is neither an undo operation in progress nor a redo operation in progress then any entries in the undo transaction history after the undo position must be removed (as if clearRedo() had been called).

We could fire events when someone adds something to the undo history -- one event per undo object entry before the position (or after, during redo addition), allowing the script to decide if that entry should remain or not. Or something. Would make it potentially easier to expire server-held state when the server limitations come into play.

The remove(index) method must remove the undo object entry with the specified index. If the index is less than zero or greater than or equal to length then the method must raise an INDEX_SIZE_ERR exception. DOM changes entries are unaffected by this method.

The clearUndo() method must remove all entries in the undo transaction history before the undo position, be they DOM changes entries or undo object entries.

The clearRedo() method must remove all entries in the undo transaction history after the undo position, be they DOM changes entries or undo object entries.

Another idea is to have a way for scripts to say "startBatchingDOMChangesForUndo()" and after that the changes to the DOM go in as if the user had done them.

5.4.2. Undo: moving back in the undo transaction history

When the user invokes an undo operation, or when the execCommand() method is called with the undo command, the user agent must perform an undo operation.

If the undo position is at the start of the undo transaction history, then the user agent must do nothing.

If the entry immediately before the undo position is a DOM changes entry, then the user agent must remove that DOM changes entry, reverse the DOM changes that were listed in that entry, and, if the changes were reversed with no problems, add a new DOM changes entry (consisting of the opposite of those DOM changes) to the undo transaction history on the other side of the undo position.

If the DOM changes cannot be undone (e.g. because the DOM state is no longer consistent with the changes represented in the entry), then the user agent must simply remove the DOM changes entry, without doing anything else.

If the entry immediately before the undo position is an undo object entry, then the user agent must first remove that undo object entry from the undo transaction history, and then must fire an undo event on the Document object, using the undo object entry's associated undo object as the event's data.

Any calls to add() while the event is being handled will be used to populate the redo history, and will then be used if the user invokes the "redo" command to undo his undo.

5.4.3. Redo: moving forward in the undo transaction history

When the user invokes a redo operation, or when the execCommand() method is called with the redo command, the user agent must perform a redo operation.

This is mostly the opposite of an undo operation, but the full definition is included here for completeness.

If the undo position is at the end of the undo transaction history, then the user agent must do nothing.

If the entry immediately after the undo position is a DOM changes entry, then the user agent must remove that DOM changes entry, reverse the DOM changes that were listed in that entry, and, if the changes were reversed with no problems, add a new DOM changes entry (consisting of the opposite of those DOM changes) to the undo transaction history on the other side of the undo position.

If the DOM changes cannot be redone (e.g. because the DOM state is no longer consistent with the changes represented in the entry), then the user agent must simply remove the DOM changes entry, without doing anything else.

If the entry immediately after the undo position is an undo object entry, then the user agent must first remove that undo object entry from the undo transaction history, and then must fire a redo event on the Document object, using the undo object entry's associated undo object as the event's data.

5.4.4. The `UndoManagerEvent` interface and the `undo` and `redo` events

interface UndoManagerEvent : Event {
  readonly attribute DOMObject data;
  void initUndoManagerEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMObject dataArg);
  void initUndoManagerEventNS(in DOMString namespaceURIArg, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMObject dataArg);
};

The initUndoManagerEvent() and initUndoManagerEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The data attribute represents the undo object for the event.

The undo and redo events do not bubble, cannot be canceled, and have no default action. When the user agent fires one of these events it must use the UndoManagerEvent interface, with the data field containing the relevant undo object.

5.4.5. Implementation notes

How user agents present the above conceptual model to the user is not defined. The undo interface could be a filtered view of the undo transaction history, it could manipulate the undo transaction history in ways not described above, and so forth. For example, it is possible to design a UA that appears to have separate undo transaction histories for each form control; similarly, it is possible to design systems where the user has access to more undo information than is present in the offical (as described above) undo transaction history (such as providing a tree-based approach to document state). Such UI models should be based upon the single undo transaction history described in this section, however, such that to a script there is no detectable difference.

5.5. Command APIs

The execCommand(commandID, doShowUI, value) method on the HTMLDocument interface allows scripts to perform actions on the current selection or at the current caret position. Generally, these commands would be used to implement editor UI, for example having a "delete" button on a toolbar.

There are three variants to this method, with one, two, and three arguments respectively. The doShowUI and value parameters, even if specified, are ignored unless otherwise stated.

In this specification, in fact, the doShowUI parameter is always ignored, regardless of its value. It is included for historical reasons only.

When any of these methods are invoked, user agents must act as described in the list below.

For actions marked "editing hosts only", if the selection is not entirely within an editing host, of if there is no selection and the caret is not inside an editing host, then the user agent must do nothing.

If the commandID is undo

The user agent must move back one step in its undo transaction history, restoring the associated state. If there is no further undo information the user agent must do nothing. See the undo history.

If the commandID is redo

The user agent must move forward one step in its undo transaction history, restoring the associated state. If there is no further undo (well, "redo") information the user agent must do nothing. See the undo history.

If the commandID is selectAll

The user agent must change the selection so that all the content in the currently focused editing host is selected. If no editing host is focused, then the content of the entire document must be selected.

If the commandID is unselect

The user agent must change the selection so that nothing is selected.

We need some sort of way in which the user can make a selection without risk of script clobbering it.

If the commandID is superscript

Editing hosts only. The user agent must act as if the user had requested that the selection be wrapped in the semantics of the sup element (or unwrapped, or, if there is no selection, have that semantic inserted or removed — the exact behaviour is UA-defined).

If the commandID is subscript

Editing hosts only. The user agent must act as if the user had requested that the selection be wrapped in the semantics of the sub element (or, again, unwrapped, or have that semantic inserted or removed, as defined by the UA).

If the commandID is formatBlock

Editing hosts only. This command changes the semantics of the blocks containing the selection.

If there is no selection, then, where in the description below refers to the selection, the user agent must act as if the selection was an empty range at the caret position.

If the value parameter is not specified or has a value other than one of the following literal strings:

<address>
<aside>
<h1>
<h2>
<h3>
<h4>
<h5>
<h6>
<nav>
<p>
<pre>

...then the user agent must do nothing.

Otherwise, the user agent must, for every position in the selection, take the furthest block-level element ancestor of that position that contains only inline-level content and is not being used as a structured inline-level element, and, if that element is a descendant of the editing host, rename it according to the value, by stripping the leading < character and the trailing > character and using the rest as the new tag name.

If the commandID is delete

Editing hosts only. The user agent must act as if the user had performed a backspace operation.

If the commandID is forwardDelete

Editing hosts only. The user agent must act as if the user had performed a forward delete operation.

If the commandID is insertLineBreak

Editing hosts only. The user agent must act as if the user had requested a line separator.

If the commandID is insertParagraph

Editing hosts only. The user agent must act as if the user had performed a break block editing action.

If the commandID is insertText

Editing hosts only. The user agent must act as if the user had inserted text corresponding to the value parameter.

If the commandID is vendorID-customCommandID

User agents may implement vendor-specific extensions to this API. Vendor-specific extensions to the list of commands should use the syntax vendorID-customCommandID so as to prevent clashes between extensions from different vendors and future additions to this specification.

If the commandID is something else

User agents must do nothing.

5.6. The text selection APIs

Every browsing context has a selection. The selection may be empty, and the selection may have more than one range (a disjointed selection). The user should be able to change the selection. User agents are not required to let the user select more than one range, and may collapse multiple ranges in the selection to a single range when the user interacts with the selection. (But, of course, the user agent may let the user create selections with multiple ranges.)

This one selection must be shared by all the content of the browsing context (though not by nested browsing contexts), including any editing hosts in the document. (Editing hosts that are not inside a document cannot have a selection.)

If the selection is empty (collapsed, so that it has only one segment and that segment's start and end points are the same) then the selection's position should equal the caret position. When the selection is not empty, this specification does not define the caret position; user agents should follow platform conventions in deciding whether the caret is at the start of the selection, the end of the selection, or somewhere else.

On some platforms (such as those using Wordstar editing conventions), the caret position is totally independent of the start and end of the selection, even when the selection is empty. On such platforms, user agents may ignore the requirement that the cursor position be linked to the position of the selection altogether.

Mostly for historical reasons, in addition to the browsing context's selection, each textarea and input element has an independent selection. These are the text field selections.

The datagrid and select elements also have selections, indicating which items have been picked by the user. These are not discussed in this section.

This specification does not specify how selections are presented to the user. The Selectors specification, in conjunction with CSS, can be used to style text selections using the ::selection pseudo-element. [SELECTORS] [CSS21]

5.6.1. APIs for the browsing context selection

The getSelection() method on the Window interface must return the Selection object representing the selection of that Window object's browsing context.

For historical reasons, the getSelection() method on the HTMLDocument interface must return the same Selection object.

interface Selection {
  readonly attribute Node anchorNode;
  readonly attribute long anchorOffset;
  readonly attribute Node focusNode;
  readonly attribute long focusOffset;
  readonly attribute boolean isCollapsed;
  void collapse(in Node parentNode, in long offset);
  void collapseToStart();
  void collapseToEnd();
  void selectAllChildren(in Node parentNode);
  void deleteFromDocument();
  readonly attribute long rangeCount;
  Range getRangeAt(in long index);
  void addRange(in Range range);
  void removeRange(in Range range);
  void removeAllRanges();
  DOMString toString();
};

The Selection interface is represents a list of Range objects. The first item in the list has index 0, and the last item has index count-1, where count is the number of ranges in the list. [DOM2RANGE]

All of the members of the Selection interface are defined in terms of operations on the Range objects represented by this object. These operations can raise exceptions, as defined for the Range interface; this can therefore result in the members of the Selection interface raising exceptions as well, in addition to any explicitly called out below.

The anchorNode attribute must return the value returned by the startContainer attribute of the last Range object in the list, or null if the list is empty.

The anchorOffset attribute must return the value returned by the startOffset attribute of the last Range object in the list, or 0 if the list is empty.

The focusNode attribute must return the value returned by the endContainer attribute of the last Range object in the list, or null if the list is empty.

The focusOffset attribute must return the value returned by the endOffset attribute of the last Range object in the list, or 0 if the list is empty.

The isCollapsed attribute must return true if there are zero ranges, or if there is exactly one range and its collapsed attribute is itself true. Otherwise it must return false.

The collapse(parentNode, offset) method must raise a WRONG_DOCUMENT_ERR DOM exception if parentNode's ownerDocument is not the HTMLDocument object with which the Selection object is associated. Otherwise it is, and the method must remove all the ranges in the Selection list, then create a new Range object, add it to the list, and invoke its setStart() and setEnd() methods with the parentNode and offset values as their arguments.

The collapseToStart() method must raise an INVALID_STATE_ERR DOM exception if there are no ranges in the list. Otherwise, it must invoke the collapse() method with the startContainer and startOffset values of the first Range object in the list as the arguments.

The collapseToEnd() method must raise an INVALID_STATE_ERR DOM exception if there are no ranges in the list. Otherwise, it must invoke the collapse() method with the endContainer and endOffset values of the last Range object in the list as the arguments.

The selectAllChildren(parentNode) method must invoke the collapse() method with the parentNode value as the first argument and 0 as the second argument, and must then invoke the selectNodeContents() method on the first (and only) range in the list with the parentNode value as the argument.

The deleteFromDocument() method must invoke the deleteContents() method on each range in the list, if any, from first to last.

The rangeCount attribute must return the number of ranges in the list.

The getRangeAt(index) method must return the indexth range in the list. If index is less than zero or greater or equal to the value returned by the rangeCount attribute, then the method must raise an INDEX_SIZE_ERR DOM exception.

The addRange(range) method must add the given range Range object to the list of selections, at the end (so the newly added range is the new last range). Duplicates are not prevented; a range may be added more than once in which case it appears in the list more than once, which (for example) will cause toString() to return the range's text twice.

The removeRange(range) method must remove the first occurrence of range in the list of ranges, if it appears at all.

The removeAllRanges() method must remove all the ranges from the list of ranges, such that the rangeCount attribute returns 0 after the removeAllRanges() method is invoked (and until a new range is added to the list, either through this interface or via user interaction).

The toString() method must return a concatenation of the results of invoking the toString() method of the Range object on each of the ranges of the selection, in the order they appear in the list (first to last).

In language bindings where this is supported, objects implementing the Selection interface must stringify to the value returned by the object's toString() method.

In the following document fragment, the emphasised parts indicate the selection.

<p>The cute girl likes the <cite>Oxford English Dictionary</cite>.</p>

If a script invoked window.getSelection().toString(), the return value would be "the Oxford English".

The Selection interface has no relation to the DataGridSelection interface.

5.6.2. APIs for the text field selections

When we define HTMLTextAreaElement and HTMLInputElement we will have to add the IDL given below to both of their IDLs.

The input and textarea elements define four members in their DOM interfaces for handling their text selection:

  void select();
           attribute unsigned long selectionStart;
           attribute unsigned long selectionEnd;
  void setSelectionRange(in unsigned long start, in unsigned long end);

These methods and attributes expose and control the selection of input and textarea text fields.

The select() method must cause the contents of the text field to be fully selected.

The selectionStart attribute must, on getting, return the offset (in logical order) to the character that immediately follows the start of the selection. If there is no selection, then it must return the offset (in logical order) to the character that immediately follows the text entry cursor.

On setting, it must act as if the setSelectionRange() method had been called, with the new value as the first argument, and the current value of the selectionEnd attribute as the second argument, unless the current value of the selectionEnd is less than the new value, in which case the second argument must also be the new value.

The selectionEnd attribute must, on getting, return the offset (in logical order) to the character that immediately follows the end of the selection. If there is no selection, then it must return the offset (in logical order) to the character that immediately follows the text entry cursor.

On setting, it must act as if the setSelectionRange() method had been called, with the current value of the selectionStart attribute as the first argument, and new value as the second argument.

The setSelectionRange(start, end) method must set the selection of the text field to the sequence of characters starting with the character at the startth position (in logical order) and ending with the character at the (end-1)th position. Arguments greater than the length of the value in the text field must be treated as pointing at the end of the text field. If end is less than or equal to start then the start of the selection and the end of the selection must both be placed immediately before the character with offset end. In UAs where there is no concept of an empty selection, this must set the cursor to be just before the character with offset end.

To obtain the currently selected text, the following JavaScript suffices:

var selectionText = control.value.substring(control.selectionStart, control.selectionEnd);

...where control is the input or textarea element.

Characters with no visible rendering, such as U+200D ZERO WIDTH JOINER, still count as characters. Thus, for instance, the selection can include just an invisible character, and the text insertion cursor can be placed to one side or another of such a character.

When these methods and attributes are used with input elements that are not displaying simple text fields, they must raise an INVALID_STATE_ERR exception.

6. Communication

6.1. Event definitions

Messages in cross-document messaging and, by default, in server-sent DOM events, use the message event.

The following interface is defined for this event:

interface MessageEvent : Event {
  readonly attribute DOMString data;
  readonly attribute DOMString domain;
  readonly attribute DOMString uri;
  readonly attribute Document source;
  void initMessageEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString dataArg, in DOMString domainArg, in DOMString uriArg, in Document documentArg);
  void initMessageEventNS(in DOMString namespaceURI, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString dataArg, in DOMString domainArg, in DOMString uriArg, in Document documentArg);
};

The initMessageEvent() and initMessageEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The data attribute represents the message being sent.

The domain attribute represents, in cross-document messaging, the domain of the document from which the message came.

The uri attribute represents, in cross-document messaging, the address of the document from which the message came.

The source attribute represents, in cross-document messaging, the Document from which the message came.

6.2. Server-sent DOM events

This section describes a mechanism for allowing servers to dispatch DOM events into documents that expect it. The event-source element provides a simple interface to this mechanism.

6.2.1. The `RemoteEventTarget` interface

Any object that implements the EventTarget interface must also implement the RemoteEventTarget interface.

interface RemoteEventTarget {
  void addEventSource(in DOMString src);
  void removeEventSource(in DOMString src);
};

When the addEventSource(src) method is invoked, the user agent must add the URI specified in src to the list of event sources for that object. The same URI can be registered multiple times.

When the removeEventSource(src) method is invoked, the user agent must remove the URI specified in src from the list of event sources for that object. If the same URI has been registered multiple times, removing it must only remove one instance of that URI for each invocation of the removeEventSource() method.

Relative URIs must be resolved relative to ....

6.2.2. Connecting to an event stream

Each object implementing the EventTarget and RemoteEventTarget interfaces has a list of event sources that are registered for that object.

When a new URI is added to this list, the user agent should, as soon as all currently executing scripts (if any) have finished executing, and if the specified URI isn't removed from the list before they do so, fetch the resource identified by that URI.

When an event source is removed from the list of event sources for an object, if that resource is still being fetched, then the relevant connection must be closed.

Since connections established to remote servers for such resources are expected to be long-lived, UAs should ensure that appropriate buffering is used. In particular, while line buffering may be safe if lines are defined to end with a single U+000A LINE FEED character, block buffering or line buffering with different expected line endings can cause delays in event dispatch.

In general, the semantics of the transport protocol specified by the URIs for the event sources must be followed, including HTTP caching rules.

For HTTP connections, the Accept header may be included; if included, it must only contain formats of event framing that are supported by the user agent (one of which must be application/x-dom-event-stream, as described below).

Other formats of event framing may also be supported in addition to application/x-dom-event-stream, but this specification does not define how they are to be parsed or processed.

Such formats could include systems like SMS-push; for example servers could use Accept headers and HTTP redirects to an SMS-push mechanism as a kind of protocol negotiation to reduce network load in GSM environments.

User agents should use the Cache-Control: no-cache header in requests to bypass any caches for requests of event sources.

For connections to domains other than the document's domain, the semantics of the Access-Control HTTP header must be followed. [ACCESSCONTROL]

HTTP 200 OK responses with a Content-Type header specifying the type application/x-dom-event-stream that are either from the document's domain or explicitly allowed by the Access-Control HTTP headers must be processed line by line as described below.

For the purposes of such successfully opened event streams only, user agents should ignore HTTP cache headers, and instead assume that the resource indicates that it does not wish to be cached.

If such a resource completes loading (i.e. the entire HTTP response body is received or the connection itself closes), the user agent should request the event source resource again after a delay of approximately five seconds.

HTTP 200 OK responses that have a Content-Type other than application/x-dom-event-stream (or some other supported type), and HTTP responses whose Access-Control headers indicate that the resource are not to be used, must be ignored and must prevent the user agent from refetching the resource for that event source.

HTTP 201 Created, 202 Accepted, 203 Non-Authoritative Information, and 206 Partial Content responses must be treated like HTTP 200 OK responses for the purposes of reopening event source resources. They are, however, likely to indicate an error has occurred somewhere and may cause the user agent to emit a warning.

HTTP 204 No Content, and 205 Reset Content responses must be treated as if they were 200 OK responses with the right MIME type but no content, and should therefore cause the user agent to refetch the resource after a short delay.

HTTP 300 Multiple Choices responses should be handled automatically if possible (treating the responses as if they were 302 Found responses pointing to the appropriate resource), and otherwise must be treated as HTTP 404 responses.

HTTP 301 Moved Permanently responses must cause the user agent to reconnect using the new server specified URI instead of the previously specified URI for all subsequent requests for this event source. (It doesn't affect other event sources with the same URI unless they also receive 301 responses, and it doesn't affect future sessions, e.g. if the page is reloaded.)

HTTP 302 Found, 303 See Other, and 307 Temporary Redirect responses must cause the user agent to connect to the new server-specified URI, but if the user agent needs to again request the resource at a later point, it must return to the previously specified URI for this event source.

HTTP 304 Not Modified responses should be handled like HTTP 200 OK responses, with the content coming from the user agent cache. A new request should then be made after a short delay of approximately five seconds.

HTTP 305 Use Proxy, HTTP 401 Unauthorized, and 407 Proxy Authentication Required should be treated transparently as for any other subresource.

Any other HTTP response code not listed here should cause the user agent to stop trying to process this event source.

DNS errors must be considered fatal, and cause the user agent to not open any connection for that event source.

For non-HTTP protocols, UAs should act in equivalent ways.

6.2.3. Parsing an event stream

This event stream format's MIME type is application/x-dom-event-stream.

The event stream format is (in pseudo-BNF):

<stream>          ::= <bom>? <event>*
<event>           ::= [ <comment> | <command> | <field> ]* <newline>
<comment>         ::= ';' <any-char>* <newline>
<command>         ::= ':' <any-char>* <newline>
<field>           ::= <name> [ ':' <space>? <any-char>* ]? <newline>
<name>            ::= <name-start-char> <name-char>*

# characters:
<bom>             ::= a single U+FEFF BYTE ORDER MARK character
<space>           ::= a single U+0020 SPACE character (' ')
<newline>         ::= a U+000D CARRIAGE RETURN character
                      followed by a U+000A LINE FEED character
                      | a single U+000D CARRIAGE RETURN character
                      | a single U+000A LINE FEED character
                      | the end of the file
<name-start-char> ::= a single Unicode character other than
                      ':', ';', U+000D CARRIAGE RETURN and U+000A LINE FEED
<name-char>       ::= a single Unicode character other than
                      ':', U+000D CARRIAGE RETURN and U+000A LINE FEED
<any-char>        ::= a single Unicode character other than
                      U+000D CARRIAGE RETURN and U+000A LINE FEED

Event streams in this format must always be encoded as UTF-8. Lines must be separated by either a U+000D CARRIAGE RETURN U+000A LINE FEED (CRLF) character pair, a single U+000A LINE FEED (LF) character, or a single U+000D CARRIAGE RETURN (CR) character. User agents must treat those three variants as equivalent line terminators.

Bytes that are not valid UTF-8 sequences must be interpreted as the U+FFFD REPLACEMENT CHARACTER.

A leading U+FEFF BYTE ORDER MARK character must be ignored if present.

The stream must then be parsed by reading everything line by line, in blocks separated by blank lines. Comment lines (those starting with the character ';') and command lines (those starting with the character ':') must be ignored.

Command lines are reserved for future extensions.

For each non-blank, non-comment, non-command line, the field name must first be taken. This is everything on the line up to but not including the first colon (':') or the line terminator, whichever comes first. Then, if there was a colon, the data for that line must be taken. This is everything after the colon, ignoring a single space after the colon if there is one, up to the end of the line. If there was no colon the data is the empty string.

Examples:

Field name: Field data

This is a blank field

1. These two lines: have the same data
2. These two lines:have the same data

1. But these two lines:  do not
2. But these two lines: do not

If a field name occurs multiple times in a block, the value for that field in that black must consist of the data parts for each of those lines, concatenated with U+000A LINE FEED characters between them (regardless of what the line terminators used in the stream actually are).

For example, the following block:

Test: Line 1
Foo:  Bar
Test: Line 2

...is treated as having two fields, one called Test with the value "Line 1\nLine 2" (where \n represents a newline), and one called Foo with the value " Bar" (note the leading space character).

A block thus consists of all the name-value pairs for its fields. Command lines have no effect on blocks and are not considered part of a block.

Since any random stream of characters matches the above format, there is no need to define any error handling.

6.2.4. Interpreting an event stream

Once the fields have been parsed, they are interpreted as follows (these are case-sensitive exact comparisons):

Event field

This field gives the name of the event. For example, load, DOMActivate, updateTicker. If there is no field with this name, the name message must be used.

Namespace field

This field gives the DOM3 namespace for the event. (For normal DOM events this would be null.) If it isn't specified the event namespace is null.

Class field

This field gives is the interface used for the event, for instance Event, UIEvent, MutationEvent, KeyboardEvent, etc. For compatibility with DOM3 Events, the values UIEvents, MouseEvents, MutationEvents, and HTMLEvents are valid values and must be treated respectively as meaning the interfaces UIEvent, MouseEvent, MutationEvent, and Event. (This value can therefore be used as the argument to createEvent().)

If the value is not specified but the Namespace is null and the Event field exactly matches one of the events specified by DOM3 Events in section 1.4.2 "Complete list of event types", then the interface used must default to the interface relevant for that event type. [DOM3EVENTS]

For example:

Event: click

...would cause Class to be treated as MouseEvent.

If the Namespace is null and the Event field is message (including if it was not specified explicitly), then the MessageEvent interface must be used.

Otherwise, the Event interface must be used.

It is quite possible to give the wrong class for an event. This is equivalent to creating an event in the DOM using the DOM Event APIs, but using the wrong interface for it.

Bubbles field

This field specifies whether the event is to bubble. If it is specified and has the value No, the event must not bubble. If it is specified and has any other value (including no or NO) then the event must bubble.

If it is not specified but the Namespace field is null and the Event field exactly matches one of the events specified by DOM3 Events in section 1.4.2 "Complete list of event types", then the event must bubble if the DOM3 Events spec specifies that that event bubbles, and musn't bubble if it specifies it does not. [DOM3EVENTS]

For example:

Event: load

...would cause Bubbles to be treated as No.

Otherwise, the event must bubble.

Cancelable field

This field specifies whether the event can have its default action prevented. If it is specified and has the value No, the event must not be cancelable. If it is specified and has any other value (including no or NO) then the event must be cancelable.

If it is not specified, but the Namespace field is null and the Event field exactly matches one of the events specified by DOM3 Events in section 1.4.2 "Complete list of event types", then the event must be cancelable if the DOM3 Events specification specifies that it is, and must not be cancelable otherwise. [DOM3EVENTS]

For example:

Event: load

...would cause Cancelable to be treated as No.

Otherwise, the event must be cancelable.

Target field

This field gives the node that the event is to be dispatched on.

If the object for which the event source is being processed is not a Node, but the Target field is nonetheless specified, then the event must be dropped.

Otherwise, if field is specified and its value starts with a # character, then the remainder of the value represents an ID, and the event must be dispatched on the same node as would be obtained by the getElementById() method on the ownerDocument of the node whose event source is being processed.

For example,

Target: #test

...would target the element with ID test.

Otherwise, if the field is specified and its value is the literal string "Document", then the event must be dispatched at the ownerDocument of the node whose event source is being processed.

Otherwise, the field (whether specified or not) is ignored and the event must be dispatched at the object itself.

Other fields depend on the interface specified (or possibly implied) by the Class field. If the specified interface has an attribute that exactly matches the name of the field, and the value of the field can be converted (using the type conversions defined in ECMAScript) to the type of the attribute, then it must be used. Any attributes (other than the Event interface attributes) that do not have matching fields are initialised to zero, null, false, or the empty string.

For example:

Event: click
Class: MouseEvent
button: 2

...would result in a 'click' event using the MouseEvent interface that has button set to 2 but screenX, screenY, etc, set to 0, false, or null as appropriate.

If a field does not match any of the attributes on the event, it must be ignored.

For example:

Event: keypress
Class: MouseEvent
keyIdentifier: 0

...would result in a MouseEvent event with its fields all at their default values, with the event name being keypress. The keyIdentifier field would be ignored. (If the author had not included the Class field explicitly, it would have just worked, since the class would have defaulted as described above.)

Once a blank line or the end of the file is reached, an event of the type and namespace given by the Event and Namespace fields respectively must be synthesized and dispatched to the appropriate node as described by the fields above. No event must be dispatched until a blank line has been received or the end of the file reached.

The event must be dispatched as if using the DOM dispatchEvent() method. Thus, if the Event field was omitted, leaving the name as the empty string, or if the name had invalid characters, then the dispatching of the event fails.

Events fired from event sources do not have user-agent default actions.

The following event stream, once followed by a blank line:

data: YHOO
data: -2
data: 10

...would cause an event message with the interface MessageEvent to be dispatched on the event-source element, which would then bubble up the DOM, and whose data attribute would contain the string YHOO\n-2\n10 (where \n again represents a newline).

This could be used as follows:

<event-source src="http://stocks.example.com/ticker.php"
              onmessage="var data = event.data.split('\n'); updateStocks(data[0], data[1], data[2]);">

...where updateStocks() is a function defined as:

function updateStocks(symbol, delta, value) { ... }

...or some such.

The following stream contains four blocks and therefore fires four events. The first block has just a comment, and will fire a message event with all the fields set to the empty string or null. The second block has two fields with names "load" and "Target" respectively; since there is no "load" member on the MessageEvent object that field is ignored, leaving the event as a second message event with all the fields set to the empty string or null, but this time the event is targetted at an element with ID "image1". The third block is empty (no lines between two blank lines), and the fourth block has only two comments, so they both yet again fire message events with all the fields set to the empty string or null.

; test

load
Target: #image1


; if any more events follow this block, they will not be affected by
; the "Target" and "load" fields above.

6.2.5. Notes

Legacy proxy servers are known to, in certain cases, drop HTTP connections after a short timeout. To protect against such proxy servers, authors can include a comment line (one starting with a ';' character) every 15 seconds or so.

Authors wishing to relate event source connections to each other or to specific documents previously served might find that relying on IP addresses doesn't work, as individual clients can have multiple IP addresses (due to having multiple proxy servers) and individual IP addresses can have multiple clients (due to sharing a proxy server). It is better to include a unique identifier in the document when it is served and then pass that identifier as part of the URI in the src attribute of the event-source element.

Implementations that support HTTP's per-server connection limitation might run into trouble when opening multiple pages from a site if each page has an event-source to the same domain. Authors can avoid this using the relatively complex mechanism of using unique domain names per connection, or by allowing the user to enable or disable the event-source functionality on a per-page basis.

6.3. Network connections

To enable Web applications to communicate with each other in local area networks, and to maintain bidirectional communications with their originating server, this specification introduces the Connection interface.

The Window interface provides three constructors for creating Connection objects: TCPConnection(), for creating a direct (possibly encrypted) link to another node on the Internet using TCP/IP; LocalBroadcastConnection(), for creating a connection to any listening peer on a local network (which could be a local TCP/IP subnet using UDP, a Bluetooth PAN, or another kind of network infrastructure); and PeerToPeerConnection(), for a direct peer-to-peer connection (which could again be over TCP/IP, Bluetooth, IrDA, or some other type of network).

This interface does not allow for raw access to the underlying network. For example, this interface could not be used to implement an IRC client without proxying messages through a custom server.

6.3.1. Introduction

This section is non-normative.

An introduction to the client-side and server-side of using the direct connection APIs.

An example of a party-line implementation of a broadcast service, and direct peer-to-peer chat for direct local connections.

6.3.2. The `Connection` interface

interface Connection {
  readonly attribute DOMString network;
  readonly attribute DOMString peer;
  readonly attribute int readyState;
           attribute EventListener onopen;
           attribute EventListener onread;
           attribute EventListener onclose;
  void send(in DOMString data);
  void disconnect();
};

Connection objects must also implement the EventTarget interface. [DOM3EVENTS]

When a Connection object is created, the UA must try to establish a connection, as described in the sections below describing each connection type.

The network attribute represents the name of the network connection (the value depends on the kind of connection being established). The peer attribute identifies the remote host for direct (non-broadcast) connections.

The network attribute must be set as soon as the Connection object is created, and keeps the same value for the lifetime of the object. The peer attribute must initially be set to the empty string and must be updated once, when the connection is established, after which point it must keep the same value for the lifetime of the object.

The readyState attribute represents the state of the connection. When the object is created it must be set to 0. It can have the following values:

0 Connecting: The connection has not yet been established.
1 Connected: The connection is established and communication is possible.
2 Closed: The connection has been closed.

Once a connection is established, the readyState attribute's value must be changed to 1, and the open event must be fired on the Connection object.

When data is received, the read event will be fired on the Connection object.

When the connection is closed, the readyState attribute's value must be changed to 2, and the close event must be fired on the Connection object.

The onopen, onread, and onclose attributes must, when set, register their new value as an event listener for their respective events (namely open, read, and close), and unregister their previous value if any.

The send() method transmits data using the connection. If the connection is not yet established, it must raise an INVALID_STATE_ERR exception. If the connection is established, then the behaviour depends on the connection type, as described below.

The disconnect() method must close the connection, if it is open. If the connection is already closed, it must do nothing. Closing the connection causes a close event to be fired and the readyState attribute's value to change, as described above.

6.3.3. Connection Events

All the events described in this section are events in no namespace, which do not bubble, are not cancelable, and have no default action.

The open event is fired when the connection is established. UAs must use the normal Event interface when firing this event.

The close event is fired when the connection is closed (whether by the author, calling the disconnect() method, or by the server, or by a network error). UAs must use the normal Event interface when firing this event as well.

No information regarding why the connection was closed is passed to the application in this version of this specification.

The read event is fired when when data is received for a connection. UAs must use the ConnectionReadEvent interface for this event.

interface ConnectionReadEvent : Event {
  readonly attribute DOMString data;
  readonly attribute DOMString source;
  void initConnectionReadEvent(in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString dataArg);
  void initConnectionReadEventNS(in DOMString namespaceURI, in DOMString typeArg, in boolean canBubbleArg, in boolean cancelableArg, in DOMString dataArg);
};

The initConnectionReadEvent() and initConnectionReadEventNS() methods must initialise the event in a manner analogous to the similarly-named methods in the DOM3 Events interfaces. [DOM3EVENTS]

The data attribute represents the data that was transmitted from the peer.

The source attribute represents the name of the peer. This is primarily useful on broadcast connections; on direct connections it is equal to the peer attribute on the Connection object.

Events that would be fired during script execution (e.g. between the connection object being created — and thus the connection being established — and the current script completing; or, during the execution of a read event handler) must be buffered, and those events queued up and each one individually fired after the script has completed.

6.3.4. TCP connections

The TCPConnection(subdomain, port, secure) constructor on the Window interface returns a new object implementing the Connection interface, set up for a direct connection to a specified host on the page's domain.

When this constructor is invoked, the following steps must be followed.

First, if the domain part of the script's origin is not a host name (e.g. it is an IP address) then the UA must raise a security exception. We currently don't allow connections to be set up back to an originating IP address, but we could, if the subdomain is the empty string.

Then, if the subdomain argument is null or the empty string, the target host is the domain part of the script's origin. Otherwise, the subdomain argument is prepended to the domain part of the script's origin with a dot separating the two strings, and that is the target host.

If either:

the target host is not a valid host name, or
the port argument is neither equal to 80, nor equal to 443, nor greater than or equal to 1024 and less than or equal to 65535,

...then the UA must raise a security exception.

Otherwise, the user agent must verify that the the string representing the script's domain in IDNA format can be obtained without errors. If it cannot, then the user agent must raise a security exception.

The user agent may also raise a security exception at this time if, for some reason, permission to create a direct TCP connection to the relevant host is denied. Reasons could include the UA being instructed by the user to not allow direct connections, or the UA establishing (for instance using UPnP) that the network topology will cause connections on the specified port to be directed at the wrong host.

If no exceptions are raised by the previous steps, then a new Connection object must be created, its peer attribute must be set to a string consisting of the name of the target host, a colon (U+003A COLON), and the port number as decimal digits, and its network attribute must be set to the same value as the peer attribute.

This object must then be returned.

The user agent must then begin trying to establish a connection with the target host and specified port. (This typically would begin in the backgound, while the script continues to execute.)

If the secure boolean argument is set to true, then the user agent must establish a secure connection with the target host and specified port using TLS or another protocol, negotiated with the server. [RFC2246] If this fails the user agent must act as if it had closed the connection.

Once a secure connection is established, or if the secure boolean argument is not set to true, then the user agent must continue to connect to the server using the protocol described in the section entitled clients connecting over TCP. All data on connections made using TLS must be sent as "application data".

Once the connection is established, the UA must act as described in the section entitled sending and receiving data over TCP.

User agents should allow multiple TCP connections to be established per host. In particular, user agents should not apply per-host HTTP connection limits to connections established with the TCPConnection constructor.

6.3.5. Broadcast connections

The LocalBroadcastConnection() constructor on the Window interface returns a new object implementing the Connection interface, set up to broadcast on the local network.

When this constructor is invoked, a new Connection object must be created.

The network attribute of the object must be set to the string representing the script's domain in IDNA format. If this string cannot be obtained, then the user agent must raise a security exception exception when the constructor is called.

The peer attribute must be set to the empty string.

The object must then be returned, unless, for some reason, permission to broadcast on the local network is to be denied. In the latter case, a security exception must be raised instead. User agents may deny such permission for any reason, for example a user preference.

If the object is returned (i.e. if no exception is raised), the user agent must the begin broadcasting and listening on the local network, in the background, as described below. The user agent may define "the local network" in any way it considers appropriate and safe; for instance the user agent may ask the user which network (e.g. Bluetooth, IrDA, Ethernet, etc) the user would like to broadcast on before beginning broadcasting.

UAs may broadcast and listen on multiple networks at once. For example, the UA could broadcast on both Bluetooth and Wifi at the same time.

As soon as the object is returned, the connection has been established, which implies that the open event must be fired. Broadcast connections are never closed.

6.3.5.1. Broadcasting over TCP/IP

Should we drop this altogether? Letting people fill the local network with garbage seems unwise.

We need to register a UDP port for this. For now this spec refers to port 18080/udp.

Since this feature requires that the user agent listen to a particular port, some platforms might prevent more than one user agent per IP address from using this feature at any one time.

On TCP/IP networks, broadcast connections transmit data using UDP over port 18080.

When the send(data) method is invoked on a Connection object that was created by the LocalBroadcastConnection() constructor, the user agent must follow these steps:

Create a string consisting of the value of the network attribute of the Connection object, a U+0020 SPACE character, a U+0002 START OF TEXT character, and the data argument.
Encode the string as UTF-8.
If the resulting byte stream is longer than 65487 bytes, raise an INDEX_SIZE_ERR DOM exception and stop.
Create a UDP packet whose data is the byte stream, with the source and destination ports being 18080, and with appropriate length and checksum fields. Transmit this packet to IPv4 address 255.255.255.255 or IPv6 address ff02::1, as appropriate. IPv6 applications will also have to enable reception from this address.

When a broadcast connection is opened on a TCP/IP network, the user agent should listen for UDP packets on port 18080.

When the user agent receives a packet on port 18080, the user agent must attempt to decode that packet's data as UTF-8. If the data is not fully correct UTF-8 (i.e. if there are decoding errors) then the packet must be ignored. Otherwise, the user agent must check to see if the decoded string contains a U+0020 SPACE character. If it does not, then the packet must again be ignored (it might be a peer discovery packet from a PeerToPeerConnection() constructor). If it does then the user agent must split the string at the first space character. All the characters before the space are then known as d, and all the characters after the space are known as s. If s is not at least one character long, or if the first character of s is not a U+0002 START OF TEXT character, then the packet must be ignored. (This allows for future extension of this protocol.)

Otherwise, for each Connection object that was created by the LocalBroadcastConnection() constructor and whose network attribute exactly matches d, a read event must be fired on the Connection object. The string s, with the first character removed, must be used as the data, and the source IP address of the packet as the source.

Making the source IP available means that if two or more machines in a private network can be made to go to a hostile page simultaneously, the hostile page can determine the IP addresses used locally (i.e. on the other side of any NAT router). Is there some way we can keep link-local IP addresses secret while still allowing for applications to distinguish between multiple participants?

6.3.5.2. Broadcasting over Bluetooth

Does anyone know enough about Bluetooth to write this section?

6.3.5.3. Broadcasting over IrDA

Does anyone know enough about IrDA to write this section?

6.3.6. Peer-to-peer connections

The PeerToPeerConnection() constructor on the Window interface returns a new object implementing the Connection interface, set up for a direct connection to a user-specified host.

When this constructor is invoked, a new Connection object must be created.

The peer attribute must be set to the empty string.

The object must then be returned, unless, for some reason, permission to establish peer-to-peer connections is generally disallowed, for example due to administrator settings. In the latter case, a security exception must be raised instead.

The user agent must then, typically while the script resumes execution, find a remote host to establish a connection to. To do this it must start broadcasting and listening for peer discovery messages and listening for incoming connection requests on all the supported networks. How this is performed depends on the type of network and is described below.

The UA should inform the user of the clients that are detected, and allow the user to select one to connect to. UAs may also allow users to explicit specify hosts that were not detected, e.g. by having the user enter an IP address.

If an incoming connection is detected before the user specifies a target host, the user agent should ask the user to confirm that this is the host they wish to connect to. If it is, the connection should be accepted and the UA will act as the server in this connection. (Which UA acts as the server and which acts as the client is not discernible at the DOM API level.)

If no incoming connection is detected and if the user specifies a particular target host, a connection should be established to that host, with the UA acting as the client in the connection.

No more than one connection must be established per Connection object, so once a connection has been established, the user agent must stop listening for further connections (unless, or until such time as, another Connection object is being created).

If at any point the user cancels the connection process or the remote host refuses the connection, then the user agent must act as if it had closed the connection, and stop trying to connect.

6.3.6.1. Peer-to-peer connections over TCP/IP

Should we replace this section with something that uses Rendez-vous/zeroconf or equivalent?

We need to register ports for this. For now this spec refers to port 18080/udp and 18080/tcp.

Since this feature requires that the user agent listen to a particular port, some platforms might prevent more than one user agent per IP address from using this feature at any one time.

When using TCP/IP, broadcasting peer discovery messages must be done by creating UDP packets every few seconds containing as their data the value of the connection's network attribute, encoded as UTF-8, with the source and destination ports being set to 18080 and appropriate length and checksum fields, and sending these packets to address (in IPv4) 255.255.255.255 or (in IPv6) ff02::1, as appropriate.

Listening for peer discovery messages must be done by examining incoming UDP packets on port 18080. IPv6 applications will also have to enable reception from the ff02::1 address. If their payload is exactly byte-for-byte equal to a UTF-8 encoded version of the value of the connection's network attribute, then the source address of that packet represents the address of a host that is ready to accept a peer-to-peer connection, and it should therefore be offered to the user.

Incoming connection requests must be listened for on TCP port 18080. If an incoming connection is received, the UA must act as a server, as described in the section entitled servers accepting connections over TCP.

If no incoming connection requests are accepted and the user instead specifies a target host to connect to, the UA acts as a client: the user agent must attempt to connect to the user-specified host on port 18080, as described in the section entitled clients connecting over TCP.

Once the connection is established, the UA must act as described in the section entitled sending and receiving data over TCP.

This specification does not include a way to establish secure (encrypted) peer-to-peer connections at this time. If you can see a good way to do this, let me know.

6.3.6.2. Peer-to-peer connections over Bluetooth

Does anyone know enough about Bluetooth to write this section?

6.3.6.3. Peer-to-peer connections over IrDA

Does anyone know enough about IrDA to write this section?

6.3.7. The common protocol for TCP-based connections

The same protocol is used for TCPConnection and PeerToPeerConnection connection types. This section describes how such connections are established from the client and server sides, and then describes how data is sent and received over such connections (which is the same for both clients and servers).

6.3.7.1. Clients connecting over TCP

This section defines the client-side requirements of the protocol used by the TCPConnection and PeerToPeerConnection connection types.

If a TCP connection to the specified target host and port cannot be established, for example because the target host is a domain name that cannot be resolved to an IP address, or because packets cannot be routed to the host, the user agent should retry creating the connection. If the user agent gives up trying to connect, the user agent must act as if it had closed the connection.

No information regarding the state of the connection is passed to the application while the connection is being established in this version of this specification.

Once a TCP/IP connection to the remote host is established, the user agent must transmit the following sequence of bytes, represented here in hexadecimal form:

0x48 0x65 0x6C 0x6C 0x6F 0x0A

This represents the string "Hello" followed by a newline, encoded in UTF-8.

The user agent must then read all the bytes sent from the remote host, up to the first 0x0A byte (inclusive). That string of bytes is then compared byte-for-byte to the following string of bytes:

0x57 0x65 0x6C 0x63 0x6F 0x6E 0x65 0x0A

This says "Welcome".

If the server sent back a string in any way different to this, then the user agent must close the connection and give up trying to connect.

Otherwise, the user agent must then take the string representing the script's domain in IDNA format, encode it as UTF-8, and send that to the remote host, followed by a 0x0A byte (a U+000A LINE FEED in UTF-8).

The user agent must then read all the bytes sent from the remote host, up to the first 0x0A byte (inclusive). That string of bytes must then be compared byte-for-byte to the string that was just sent to the server (the one with the IDNA domain name and ending with a newline character). If the server sent back a string in any way different to this, then the user agent must close the connection and give up trying to connect.

Otherwise, the connection has been established (and events and so forth get fired, as described above).

If at any point during this process the connection is closed prematurely, then the user agent must close the connection and give up trying to connect.

6.3.7.2. Servers accepting connections over TCP

This section defines the server side of the protocol described in the previous section. For authors, it should be used as a guide for how to implement servers that can communicate with Web pages over TCP. For UAs these are the requirements for the server part of PeerToPeerConnections.

Once a TCP/IP connection from a remote host is established, the user agent must transmit the following sequence of bytes, represented here in hexadecimal form:

0x57 0x65 0x6C 0x63 0x6F 0x6E 0x65 0x0A

This says "Welcome" and a newline in UTF-8.

The user agent must then read all the bytes sent from the remote host, up to the first 0x0A byte (inclusive). That string of bytes is then compared byte-for-byte to the following string of bytes:

0x48 0x65 0x6C 0x6C 0x6F 0x0A

"Hello" and a newline.

If the remote host sent back a string in any way different to this, then the user agent must close the connection and give up trying to connect.

The user agent must then read all the bytes sent from the remote host, up to the first 0x0A byte (inclusive). That string of bytes must then be compared byte-for-byte to the string that was just sent to that host (the one with the IDNA domain name and ending with a newline character). If the remote host sent back a string in any way different to this, then the user agent must close the connection and give up trying to connect.

Otherwise, the connection has been established (and events and so forth get fired, as described above).

For author-written servers (as opposed to the server side of a peer-to-peer connection), the script's domain would be replaced by the hostname of the server. Alternatively, such servers might instead wait for the client to send its domain string, and then simply echo it back. This would allow connections from pages on any domain, instead of just pages originating from the same host. The client compares the two strings to ensure they are the same before allowing the connection to be used by author script.

If at any point during this process the connection is closed prematurely, then the user agent must close the connection and give up trying to connect.

6.3.7.3. Sending and receiving data over TCP

When the send(data) method is invoked on the connection's corresponding Connection object, the user agent must take the data argument, replace any U+0000 NULL and U+0017 END OF TRANSMISSION BLOCK characters in it with U+FFFD REPLACEMENT CHARACTER characters, then transmit a U+0002 START OF TEXT character, this new data string and a single U+0017 END OF TRANSMISSION BLOCK character (in that order) to the remote host, all encoded as UTF-8.

When the user agent receives bytes on the connection, the user agent must buffer received bytes until it receives a 0x17 byte (a U+0017 END OF TRANSMISSION BLOCK character). If the first buffered byte is not a 0x02 byte (a U+0002 START OF TEXT character encoded as UTF-8) then all the data up to the 0x17 byte, inclusive, must be dropped. (This allows for future extension of this protocol.) Otherwise, all the data from (but not including) the 0x02 byte and up to (but not including) the 0x17 byte must be taken, interpreted as a UTF-8 string, and a read event must be fired on the Connection object with that string as the data. If that string cannot be decoded as UTF-8 without errors, the packet should be ignored.

This protocol does not yet allow binary data (e.g. an image or media data) to be efficiently transmitted. A future version of this protocol might allow this by using the prefix character U+001F INFORMATION SEPARATOR ONE, followed by binary data which uses a particular byte (e.g. 0xFF) to encode byte 0x17 somehow (since otherwise 0x17 would be treated as transmission end by down-level UAs).

6.3.8. Security

Need to write this section.

If you have an unencrypted page that is (through a man-in-the-middle attack) changed, it can access a secure service that is using IP authentication and then send that data back to the attacker. Ergo we should probably stop unencrypted pages from accessing encrypted services, on the principle that the actual level of security is zero. Then again, if we do that, we prevent insecure sites from using SSL as a tunneling mechanism.

Should consider dropping the subdomain-only restriction. It doesn't seem to add anything, and prevents cross-domain chatter.

6.3.9. Relationship to other standards

Should have a section talking about the fact that we blithely ignoring IANA's port assignments here.

Should explain why we are not reusing HTTP for this. (HTTP is too heavy-weight for such a simple need; requiring authors to implement an HTTP server just to have a party line is too much of a barrier to entry; cannot rely on prebuilt components; having a simple protocol makes it much easier to do RAD; HTTP doesn't fit the needs and doesn't have the security model needed; etc)

6.4. Cross-document messaging

Web browsers, for security and privacy reasons, prevent documents in different domains from affecting each other; that is, cross-site scripting is disallowed.

While this is an important security feature, it prevents pages from different domains from communicating even when those pages are not hostile. This section introduces a messaging system that allows documents to communicate with each other regardless of their source domain, in a way designed to not enable cross-site scripting attacks.

We may want to just put postMessage on Window instead of Document, as that reduces the XSS risk.

6.4.1. Processing model

When a script invokes the postMessage(message) method on a Document object, the user agent must create an event that uses the MessageEvent interface, with the event name message, which bubbles, is cancelable, and has no default action. The data attribute must be set to the value passed as the message argument to the postMessage() method, the domain attribute must be set to the domain of the document that the script that invoked the methods is associated with, the uri attribute must be set to the URI of that document, and the source attribute must be set to the Document object representing that document.

The event must then be dispatched at the Document object itself.

Authors should check the domain attribute to ensure that messages are only accepted from domains that they expect to receive messages from. Otherwise, bugs in the author's message handling code could be exploited by hostile sites.

For example, if document A contains an object element that contains document B, and script in document A calls postMessage() on document B, then a message event will be fired on that element, marked as originating from document A. The script in document A might look like:

var o = document.getElementsByTagName('object')[0];
o.contentDocument.postMessage('Hello world');

To register an event handler for incoming events, the script would use addEventListener() (or similar mechanisms). For example, the script in document B might look like:

document.addEventListener('message', receiver, false);
function receiver(e) {
  if (e.domain == 'example.com') {
    if (e.data == 'Hello world') {
      e.source.postMessage('Hello');
    } else {
      alert(e.data);
    }
  }
}

This script first checks the domain is the expected domain, and then looks at the message, which it either displays to the user, or responds to by sending a message back to the document which sent the message in the first place.

Implementors are urged to take extra care in the implementation of this feature. It allows authors to transmit information from one domain to another domain, which is normally disallowed for security reasons. It also requires that UAs be careful to allow access to certain properties but not others.

7. Repetition templates

See WF2 for now

8. The HTML syntax

8.1. Writing HTML documents

This section only applies to documents, authoring tools, and markup generators. In particular, it does not apply to conformance checkers; conformance checkers must use the requirements given in the next section ("parsing HTML documents").

Documents must consist of the following parts, in the given order:

Optionally, a single U+FEFF BYTE ORDER MARK (BOM) character.
Any number of comments and space characters.
A DOCTYPE.
Any number of comments and space characters.
The root element, in the form of an html element.
Any number of comments and space characters.

The various types of content mentioned above are described in the next few sections.

In addition, there are some restrictions on how character encoding declarations are to be serialised, as discussed in the section on that topic.

8.1.1. The DOCTYPE

A DOCTYPE is a mostly useless, but required, header.

DOCTYPEs are required for legacy reasons. When omitted, browsers tend to use a different rendering mode that is incompatible with some specifications. Including the DOCTYPE in a document ensures that the browser makes a best-effort attempt at following the relevant specifications.

A DOCTYPE must consist of the following characters, in this order:

A U+003C LESS-THAN SIGN (<) character.
A U+0021 EXCLAMATION MARK (!) character.
A U+0044 LATIN CAPITAL LETTER D or U+0064 LATIN SMALL LETTER D character.
A U+004F LATIN CAPITAL LETTER O or U+006F LATIN SMALL LETTER O character.
A U+0043 LATIN CAPITAL LETTER C or U+0063 LATIN SMALL LETTER C character.
A U+0054 LATIN CAPITAL LETTER T or U+0074 LATIN SMALL LETTER T character.
A U+0059 LATIN CAPITAL LETTER Y or U+0079 LATIN SMALL LETTER Y character.
A U+0050 LATIN CAPITAL LETTER P or U+0070 LATIN SMALL LETTER P character.
A U+0045 LATIN CAPITAL LETTER E or U+0065 LATIN SMALL LETTER E character.
One or more space characters.
A U+0048 LATIN CAPITAL LETTER H or U+0068 LATIN SMALL LETTER H character.
A U+0054 LATIN CAPITAL LETTER T or U+0074 LATIN SMALL LETTER T character.
A U+004D LATIN CAPITAL LETTER M or U+006D LATIN SMALL LETTER M character.
A U+004C LATIN CAPITAL LETTER L or U+006C LATIN SMALL LETTER L character.
Zero or more space characters.
A U+003E GREATER-THAN SIGN (>) character.

In other words, <!DOCTYPE HTML>, case-insensitively.

8.1.2. Elements

There are four different kinds of elements: void elements, CDATA elements, RCDATA elements, and normal elements.

Void elements: base, link, meta, hr, br, img, embed, param, area, col, input
CDATA elements: style, script
RCDATA elements: title, textarea
Normal elements: All other allowed HTML elements are normal elements.

Tags are used to delimit the start and end of elements in the markup. CDATA, RCDATA, and normal elements have a start tag to indicate where they begin, and an end tag to indicate where they end. The start and end tags of certain normal elements can be omitted, as described later. Those that cannot be omitted must not be omitted. Void elements only have a start tag; end tags must not be specified for void elements.

The contents of the element must be placed between just after the start tag (which might be implied, in certain cases) and just before the end tag (which again, might be implied in certain cases). The exact allowed contents of each individual element depends on the content model of that element, as described earlier in this specification. Elements must not contain content that their content model disallows. In addition to the restrictions placed on the contents by those content models, however, the four types of elements have additional syntactic requirements.

Void elements can't have any contents (since there's no end tag, no content can be put between the start tag and the end tag.)

CDATA elements can have text, but the text must not contain the two character sequence "</" (U+003C LESS-THAN SIGN, U+002F SOLIDUS).

RCDATA elements can have text and character entity references, but the text must not contain the character U+003C LESS-THAN SIGN (<) or the character U+0026 AMPERSAND (&).

Normal elements can have text, character entity references, other elements, and comments, but the text must not contain the character U+003C LESS-THAN SIGN (<) or the character U+0026 AMPERSAND (&). Some normal elements also have yet more restrictions on what content they are allowed to hold, beyond the restrictions imposed by the content model and those described in this paragraph. Those restrictions are described below.

Tags contain a tag name, giving the element's name. HTML elements all have names that only use characters in the range U+0061 LATIN SMALL LETTER A .. U+007A LATIN SMALL LETTER Z, or, in uppercase, U+0041 LATIN CAPITAL LETTER A .. U+005A LATIN CAPITAL LETTER Z, and U+002D HYPHEN-MINUS (-). In the HTML syntax, tag names may be written with any mix of lower- and uppercase letters that, when converted to all-lowercase, matches the element's tag name; tag names are case-insensitive.

8.1.2.1. Start tags

Start tags must have the following format:

The first character of a start tag must be a U+003C LESS-THAN SIGN (<).
The next few characters of a start tag must be the element's tag name.
If there are to be any attributes in the next step, there must first be one or more space characters.
Then, the start tag may have a number of attributes, the syntax for which is described below. Attributes may be separated from each other by one or more space characters.
After the attributes, there may be one or more space characters. (Some attributes are required to be followed by a space. See the attributes section below.)
Then, if the element is one of the void elements, then there may be a single U+002F SOLIDUS (/) character. This character has no effect except to appease the markup gods. As this character is therefore just a symbol of faith, atheists should omit it.
Finally, start tags must be closed by a U+003E GREATER-THAN SIGN (>) character.

8.1.2.2. End tags

End tags must have the following format:

The first character of an end tag must be a U+003C LESS-THAN SIGN (<).
The second character of an end tag must be a U+002F SOLIDUS (/).
The next few characters of an end tag must be the element's tag name.
After the tag name, there may be one or more space characters.
Finally, end tags must be closed by a U+003E GREATER-THAN SIGN (>) character.

8.1.2.3. Attributes

Attributes for an element are expressed inside the element's start tag.

Attributes have a name and a value. Attribute names use characters in the range U+0061 LATIN SMALL LETTER A .. U+007A LATIN SMALL LETTER Z, or, in uppercase, U+0041 LATIN CAPITAL LETTER A .. U+005A LATIN CAPITAL LETTER Z, and U+002D HYPHEN-MINUS (-). In the HTML syntax, attribute names may be written with any mix of lower- and uppercase letters that, when converted to all-lowercase, matches the attribute's name; attribute names are case-insensitive.

Attribute values are a mixture of text and character entity references, except with the additional restriction that the text cannot contain a U+0026 AMPERSAND (&) character.

Attributes can be specified in four different ways:

Empty attribute syntax

Just the attribute name.

In the following example, the disabled attribute is given with the empty attribute syntax:

<input disabled>

If an attribute using the empty attribute syntax is to be followed by another attribute, then there must be a space character separating the two.

Unquoted attribute value syntax

The attribute name, followed by zero or more space characters, followed by a single U+003D EQUALS SIGN character, followed by zero or more space characters, followed by the attribute value, which, in addition to the requirements given above for attribute values, must not contain any literal space characters, U+003E GREATER-THAN SIGN (>) characters, or U+003C LESS-THAN SIGN (<) characters, and must not, furthermore, start with either a literal U+0022 QUOTATION MARK (") character or a literal U+0027 APOSTROPHE (') character.

In the following example, the value attribute is given with the unquoted attribute value syntax:

<input value=yes>

If an attribute using the unquoted attribute syntax is to be followed by another attribute or by one of the optional U+002F SOLIDUS (/) characters allowed in step 6 of the start tag syntax above, then there must be a space character separating the two.

Single-quoted attribute value syntax

The attribute name, followed by zero or more space characters, followed by a single U+003D EQUALS SIGN character, followed by zero or more space characters, followed by a single U+0027 APOSTROPHE (') character, followed by the attribute value, which, in addition to the requirements given above for attribute values, must not contain any literal U+0027 APOSTROPHE (') characters, and finally followed by a second single U+0027 APOSTROPHE (') character.

In the following example, the type attribute is given with the single-quoted attribute value syntax:

<input type='checkbox'>

Double-quoted attribute value syntax

The attribute name, followed by zero or more space characters, followed by a single U+003D EQUALS SIGN character, followed by zero or more space characters, followed by a single U+0022 QUOTATION MARK (") character, followed by the attribute value, which, in addition to the requirements given above for attribute values, must not contain any literal U+0022 QUOTATION MARK (") characters, and finally followed by a second single U+0022 QUOTATION MARK (") character.

In the following example, the name attribute is given with the double-quoted attribute value syntax:

<input name="be evil">

8.1.2.4. Optional tags

Certain tags can be omitted.

An html element's start tag may be omitted if the first thing inside the html element is not a space character or a comment.

An html element's end tag may be omitted if the html element is not immediately followed by a space character or a comment.

A head element's start tag may be omitted if the first thing inside the head element is an element.

A head element's end tag may be omitted if the head element is not immediately followed by a space character or a comment.

A body element's start tag may be omitted if the first thing inside the body element is not a space character or a comment, except if the first thing inside the body element is a script or style element and the node immediately preceding the body element is a head element whose end tag has been omitted.

A body element's end tag may be omitted if the body element is not immediately followed by a space character or a comment.

A li element's end tag may be omitted if the li element is immediately followed by another li element or if there is no more content in the parent element.

A dt element's end tag may be omitted if the dt element is immediately followed by another dt element or a dd element.

A dd element's end tag may be omitted if the dd element is immediately followed by another dd element or a dt element, or if there is no more content in the parent element.

A p element's end tag may be omitted if the p element is immediately followed by an address, blockquote, dl, fieldset, form, h1, h2, h3, h4, h5, h6, hr, menu, ol, p, pre, table, or ul element, or if there is no more content in the parent element.

An optgroup element's end tag may be omitted if the optgroup element is immediately followed by another optgroup element, or if there is no more content in the parent element.

An option element's end tag may be omitted if the option element is immediately followed by another option element, or if there is no more content in the parent element.

A colgroup element's start tag may be omitted if the first thing inside the colgroup element is a col element, and if the element is not immediately preceeded by another colgroup element whose end tag has been omitted.

A colgroup element's end tag may be omitted if the colgroup element is not immediately followed by a space character or a comment.

A thead element's end tag may be omitted if the thead element is immediately followed by a tbody or tfoot element.

A tbody element's start tag may be omitted if the first thing inside the tbody element is a tr element, and if the element is not immediately preceeded by a tbody, thead, or tfoot element whose end tag has been omitted.

A tbody element's end tag may be omitted if the tbody element is immediately followed by a tbody or tfoot element, or if there is no more content in the parent element.

A tfoot element's end tag may be omitted if the tfoot element is immediately followed by a tbody element, or if there is no more content in the parent element.

A tr element's end tag may be omitted if the tr element is immediately followed by another tr element, or if there is no more content in the parent element.

A td element's end tag may be omitted if the td element is immediately followed by a td or th element, or if there is no more content in the parent element.

A th element's end tag may be omitted if the th element is immediately followed by a td or th element, or if there is no more content in the parent element.

However, a start tag must never be omitted if it has any attributes.

8.1.2.5. Restrictions on content models

For historical reasons, certain elements have extra restrictions beyond even the restrictions given by their content model.

A p element must not contain blockquote, dl, menu, ol, pre, table, or ul elements, even though these elements are technically allowed inside p elements according to the content models described in this specification. (In fact, if one of those elements is put inside a p element in the markup, it will instead imply a p element end tag before it.)

An optgroup element must not contain optgroup elements, even though these elements are technically allowed to be nested according to the content models described in this specification. (If an optgroup element is put inside another in the markup, it will in fact imply an optgroup end tag before it.)

A table element must not contain tr elements, even though these elements are technically allowed inside table elements according to the content models described in this specification. (If a tr element is put inside a table in the markup, it will in fact imply a tbody start tag before it.)

A single U+000A LINE FEED (LF) character may be placed immediately after a pre element's start tag. This does not affect the processing of the element. The otherwise optional U+000A LINE FEED (LF) character must be included if the element's contents start with that character (because otherwise the leading newline in the contents would be treated like the optional newline, and ignored).

The following two pre blocks are equivalent:

<pre>Hello</pre>

<pre>
Hello</pre>

8.1.3. Text

Text is allowed inside elements, attributes, and comments. Text must consist of valid Unicode characters other than U+0000. Text should not contain control characters other than space characters. Extra constraints are placed on what is and what is not allowed in text based on where the text is to be put, as described in the other sections.

8.1.3.1. Newlines

Newlines in HTML may be represented either as U+000D CARRIAGE RETURN (CR) characters, U+000A LINE FEED (LF) characters, or pairs of U+000D CARRIAGE RETURN (CR), U+000A LINE FEED (LF) characters in that order.

8.1.4. Character entity references

In certain cases described in other sections, text may be mixed with character entity references. These can be used to escape characters that couldn't otherwise legally be included in text.

Character entity references must start with a U+0026 AMPERSAND (&). Following this, there are three possible kinds of character entity references:

Named entities: The ampersand must be followed by one of the names given in the entities section, using the same case. Finally, after the name, the entity must be terminated by a U+003B SEMICOLON character (;).
Decimal numeric entities: The ampersand must be followed by a U+0023 NUMBER SIGN (#) character, followed by one or more digits in the range U+0030 DIGIT ZERO .. U+0039 DIGIT NINE, representing a base-ten integer that itself is a valid Unicode code point that isn't U+0000. The digits must then be followed by a U+003B SEMICOLON character (;).
Hexadecimal numeric entities: The ampersand must be followed by a U+0023 NUMBER SIGN (#) character, which must be followed by either a U+0078 LATIN SMALL LETTER X or a U+0058 LATIN CAPITAL LETTER X character, which must then be followed by one or more digits in the range U+0030 DIGIT ZERO .. U+0039 DIGIT NINE, U+0061 LATIN SMALL LETTER A .. U+0066 LATIN SMALL LETTER F, and U+0041 LATIN CAPITAL LETTER A .. U+0046 LATIN CAPITAL LETTER F, representing a base-sixteen integer that itself is a valid Unicode code point that isn't U+0000. The digits must then be followed by a U+003B SEMICOLON character (;).

8.1.5. Comments

Comments must start with the four character sequence U+003C LESS-THAN SIGN, U+0021 EXCLAMATION MARK, U+002D HYPHEN-MINUS, U+002D HYPHEN-MINUS ().

8.2. Parsing HTML documents

This section only applies to user agents, data mining tools, and conformance checkers.

The rules for parsing XML documents (and thus XHTML documents) into DOM trees are covered by the XML and Namespaces in XML specifications, and are out of scope of this specification. [XML] [XMLNS]

For HTML documents, user agents must use the parsing rules described in this section to generate the DOM trees. Together, these rules define what is referred to as the HTML parser.

While the HTML form of HTML5 bears a close resemblance to SGML and XML, it is a separate language with its own parsing rules.

Some earlier versions of HTML (in particular from HTML2 to HTML4) were based on SGML and used SGML parsing rules. However, few (if any) web browsers ever implemented true SGML parsing for HTML documents; the only user agents to strictly handle HTML as an SGML application have historically been validators. The resulting confusion — with validators claiming documents to have one representation while widely deployed Web browsers interoperably implemented a different representation — has resulted in this version of HTML returning to a non-SGML basis.

Authors interested in using SGML tools in their authoring pipeline are encouraged to use the XML serialisation of HTML5 instead of the HTML serialisation.

This specification defines the parsing rules for HTML documents, whether they are syntactically valid or not. Certain points in the parsing algorithm are said to be parse errors. The error handling for parse errors is well-defined: user agents must either act as described below when encountering such problems, or must abort processing at the first error that they encounter for which they do not wish to apply the rules described below.

Conformance checkers must report at least one parse error condition to the user if one or more parse error conditions exist in the document and must not report parse error conditions if none exist in the document. Conformance checkers may report more than one parse error condition if more than one parse error conditions exist in the document. Conformance checkers are not required to recover from parse errors.

Parse errors are only errors with the syntax of HTML. In addition to checking for parse errors, conformance checkers will also verify that the document obeys all the other conformance requirements described in this specification.

8.2.1. Overview of the parsing model

The input to the HTML parsing process consists of a stream of Unicode characters, which is passed through a tokenisation stage (lexical analysis) followed by a tree construction stage (semantic analysis). The output is a Document object.

Implementations that do not support scripting do not have to actually create a DOM Document object, but the DOM tree in such cases is still used as the model for the rest of the specification.

In the common case, the data handled by the tokenisation stage comes from the network, but it can also come from script, e.g. using the document.write() API.

There is only one set of state for the tokeniser stage and the tree construction stage, but the tree construction stage is reentrant, meaning that while the tree construction stage is handling one token, the tokeniser might be resumed, causing further tokens to be emitted and processed before the first token's processing is complete.

In the following example, the tree construction stage will be called upon to handle a "p" start tag token while handling the "script" start tag token:

...
<script>
 document.write('<p>');
</script>
...

8.2.2. The input stream

The stream of Unicode characters that consists the input to the tokenisation stage will be initially seen by the user agent as a stream of bytes (typically coming over the network or from the local file system). The bytes encode the actual characters according to a particular character encoding, which the user agent must use to decode the bytes into characters.

For HTML, user agents must use the following algorithm in determining the character encoding of a document:

If the transport layer specifies an encoding, use that, and abort these steps.
The user agent may wait for 512 or more bytes of the resource to be available.
Let n be the smaller of either 512 or the number of bytes already available.

For each of the rows in the following table, starting with the first one and going down, if n is equal to or greater than the number of bytes in the first column, and the first bytes of the file match the bytes given in the first column, then use the encoding given in the cell in the second column of that row, and abort these steps:

Bytes in Hexadecimal	Description
00 00 FE FF	UTF-32BE BOM
FF FE 00 00	UTF-32LE BOM
FE FF	UTF-16BE BOM
FF FE	UTF-16LE BOM
EF BB BF	UTF-8 BOM

Otherwise, the user agent will have to search for explicit character encoding information in the file itself. This must proceed as follows:
Let position be a pointer to a byte in the input stream, initially pointing at the first byte. If at any point during these steps the position pointer points beyond the nth byte of the input stream, then skip to the next step of the overall character encoding detection algorithm (the step which mentions frequency analysis below).

Now, repeat the following "two" steps until the algorithm aborts (either because position reaches beyond the nth byte, or because a character encoding is found):
1. If position points to:
  A sequence of bytes starting with: 0x3C 0x21 0x2D 0x2D (ASCII '' sequence) and comes after the second 0x2D byte that was found. (The two 0x2D bytes cannot be the same as the those in the '<!--' sequence.) If no such byte is found before the nth byte, abort this "two step" algorithm.
  
  A sequence of bytes starting with: 0x3C, 0x4D or 0x6D, 0x45 or 0x65, 0x54 or 0x74, 0x41 or 0x61, and finally one of 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x20 (case-insensitive ASCII '<meta' followed by a space)
  1. Advance the position pointer so that it points at the next 0x09, 0x0A, 0x0B, 0x0C, 0x0D, or 0x20 byte (the one in sequence of characters matched above), if there is one before the nth byte. If there isn't, abort the "two step" algorithm.
  2. Get an attribute and its value. If no attribute was sniffed, then skip this inner set of steps, and jump to the second step in the overall "two step" algorithm.
    
    As required above, if the position pointer points beyond the nth byte after the "get an attribute" step, the "two step" algorithm will abort.
  3. Examine the attribute's name:
    
    If it is 'charset'
    
    If the attribute's value is a supported character encoding, then use the given encoding, and abort all these steps. Otherwise, do nothing with this attribute.
    If it is 'content'
    
    The attribute's value is now parsed.
    
    Skip characters in the attribute's value up to and including the first U+003B SEMICOLON (;) character.
    
    Skip any U+0009, U+000A, U+000B, U+000C, U+000D, or U+0020 characters (i.e. spaces) that immediately follow the semicolon.
    
    If the next six characters are not 'charset', abort this very inner set of steps (parsing the attribute's value), and continue looking for other attributes.
    
    Skip any U+0009, U+000A, U+000B, U+000C, U+000D, or U+0020 characters that immediately follow the word 'charset' (there might not be any).
    
    If the next character is not a U+003D EQUALS SIGN ('='), abort this very inner set of steps (parsing the attribute's value), and continue looking for other attributes.
    
    Skip any U+0009, U+000A, U+000B, U+000C, U+000D, or U+0020 characters that immediately follow the word equals sign (there might not be any).
    
    Process the next character as follows:
    
    If it is a U+0022 QUOTATION MARK ('"') and there is a later U+0022 QUOTATION MARK ('"') in the attribute's value
    
    Let tentative encoding be the string between the two quotation marks.
    If it is a U+0027 APOSTROPHE ("'") and there is a later U+0027 APOSTROPHE ("'") in the attribute's value
    
    Let tentative encoding be the string between the two apostrophes.
    If it is an unmatched U+0022 QUOTATION MARK ('"')
    If it is an unmatched U+0027 APOSTROPHE ("'")
    
    There is no tentative encoding.
    Otherwise
    
    Let tentative encoding be the string from this character to the first U+0009, U+000A, U+000B, U+000C, U+000D, or U+0020 character or the end of the attribute's value, whichever comes first.
    
    If there is a tentative encoding and it is the name of a supported character encoding, then use that encoding; abort all these steps.
    Otherwise, skip this 'content' attribute and continue on with any other attributes.
    
    Any other name
    
    Do nothing with that attribute.
  4. Return to step 1 in these inner steps.
  A sequence of bytes starting with a 0x3C byte (ASCII '<'), optionally a 0x2F byte (ASCII '/'), and finally a byte in the range 0x41-0x5A or 0x61-0x7A (an ASCII letter)
  1. Advance the position pointer so that it points at the next 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), 0x20 (ASCII space), 0x3E (ASCII '>'), 0x3C (ASCII '<') byte, if there is one before the nth byte. If there isn't, abort the "two step" algorithm.
  2. If the pointer points to a 0x3C (ASCII '<') byte, then return to the first step in the overall "two step" algorithm.
  3. Repeatedly get an attribute until no further attributes can be found, then jump to the second step in the overall "two step" algorithm.
  A sequence of bytes starting with: 0x3C 0x2D (ASCII '<!')
  A sequence of bytes starting with: 0x3C 0x2F (ASCII '</')
  A sequence of bytes starting with: 0x3C 0x3F (ASCII '<?')
  
  Advance the position pointer so that it points at the first 0x3E byte (ASCII '>') that comes after the 0x3C byte that was found. If no such byte is found before the nth byte, abort this "two step" algorithm.
  
  Any other byte
  
  Do nothing with that byte.
2. Move position so it points at the next byte in the input stream, and return to the first step of this "two step" algorithm.
When the above "two step" algorithm says to get an attribute, it means doing this:
1. If the byte at position is one of 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), 0x20 (ASCII space), or 0x2F (ASCII '/') then advance position to the next byte and start over.
2. If the byte at position is 0x3C (ASCII '<'), then move position back to the previous byte, and stop looking for an attribute. There isn't one.
3. If the byte at position is 0x3E (ASCII '>'), then stop looking for an attribute. There isn't one.
4. Otherwise, the byte at position is the start of the attribute name. Let attribute name and attribute value be the empty string.
5. Attribute name: Process the byte at position as follows:
  
  If it is 0x3D (ASCII '='), and the attribute name is longer than the empty string
  Advance position to the next byte and jump to the step below labelled value.
  If it is 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), or 0x20 (ASCII space)
  Jump to the step below labelled spaces.
  If it is 0x2F (ASCII '/'), 0x3C (ASCII '<'), or 0x3E (ASCII '>')
  Stop looking for an attribute. The attribute's name is the value of attribute name, its value is the empty string.
  If it is in the range 0x41 (ASCII 'A') to 0x5A (ASCII 'Z')
  Append the Unicode character with codepoint b+0x20 to attribute name (where b is the value of the byte at position).
  Anything else
  Append the Unicode character with the same codepoint as the value of the byte at position) to attribute name. (It doesn't actually matter how bytes outside the ASCII range are handled here, since only ASCII characters can contribute to the detection of a character encoding.)
6. Advance position to the next byte and return to the previous step.
7. Spaces. If the byte at position is one of 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), or 0x20 (ASCII space) then advance position to the next byte, then, repeat this step.
8. If the byte at position is not 0x3D (ASCII '='), stop looking for an attribute. Move position back to the previous byte. The attribute's name is the value of attribute name, its value is the empty string.
9. Advance position past the 0x3D (ASCII '=') byte.
10. Value. If the byte at position is one of 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), or 0x20 (ASCII space) then advance position to the next byte, then, repeat this step.
11. Process the byte at position as follows:
  If it is 0x22 (ASCII '"') or 0x27 ("'")
  1. Let b be the value of the byte at position.
  2. Advance position to the next byte.
  3. If the value of the byte at position is the value of b, then stop looking for an attribute. The attribute's name is the value of attribute name, and its value is the value of attribute value.
  4. Otherwise, if the value of the byte at position is in the range 0x41 (ASCII 'A') to 0x5A (ASCII 'Z'), then append a Unicode character to attribute value whose codepoint is 0x20 more than the value of the byte at position.
  5. Otherwise, append a Unicode character to attribute value whose codepoint is the same as the value of the byte at position.
  6. Return to the second step in these substeps.
  If it is 0x3C (ASCII '<'), or 0x3E (ASCII '>')
  Stop looking for an attribute. The attribute's name is the value of attribute name, its value is the empty string.
  If it is in the range 0x41 (ASCII 'A') to 0x5A (ASCII 'Z')
  Append the Unicode character with codepoint b+0x20 to attribute value (where b is the value of the byte at position).
  Anything else
  Append the Unicode character with the same codepoint as the value of the byte at position) to attribute value.
12. Process the byte at position as follows:
  
  If it is 0x09 (ASCII TAB), 0x0A (ASCII LF), 0x0B (ASCII VT), 0x0C (ASCII FF), 0x0D (ASCII CR), 0x20 (ASCII space), 0x3C (ASCII '<'), or 0x3E (ASCII '>')
  Stop looking for an attribute. The attribute's name is the value of attribute name and its value is the value of attribute value.
  If it is in the range 0x41 (ASCII 'A') to 0x5A (ASCII 'Z')
  Append the Unicode character with codepoint b+0x20 to attribute value (where b is the value of the byte at position).
  Anything else
  Append the Unicode character with the same codepoint as the value of the byte at position) to attribute value.
13. Advance position to the next byte and return to the previous step.
The user agent may attempt to autodetect the character encoding from applying frequency analysis or other algorithms to the data stream. If autodetection succeeds in determining a character encoding, then use that; abort these steps. [UNIVCHARDET]
Otherwise, use an implementation-defined or user-specified default character encoding. Due to its use in legacy content, windows-1252 is recommended as a default in predominantly Western demographics. In non-legacy environments, the more comprehensive UTF-8 encoding is recommended instead. Since these encodings can in many cases be distinguished by inspection, a user agent may heuristically decide which to use as a default.

For XML documents, the algorithm user agents must use to determine the character encoding is given by the XML specification. This section does not apply to XML documents. [XML]

When a user agent would otherwise use the ISO-8859-1 encoding, it must instead use the Windows-1252 encoding. User agents must not support the CESU-8, UTF-7, BOCU-1 and SCSU encodings. [CESU8] [UTF7] [BOCU1] [SCSU]

Bytes or sequences of bytes in the original byte stream that could not be converted to Unicode characters must be converted to U+FFFD REPLACEMENT CHARACTER code points.

A leading U+FEFF BYTE ORDER MARK (BOM) must be dropped if present.

All U+0000 NULL characters in the input must be replaced by U+FFFD REPLACEMENT CHARACTERs.

U+000D CARRIAGE RETURN (CR) characters, and U+000A LINE FEED (LF) characters, are treated specially. Any CR characters that are followed by LF characters must be removed, and any CR characters not followed by LF characters must be converted to LF characters. Thus, newlines in HTML DOMs are represented by LF characters, and there are never any CR characters in the input to the tokenisation stage.

The next input character is the first character in the input stream that has not yet been consumed. Initially, the next input character is the first character in the input.

The insertion point is the position (just before a character or just before the end of the input stream) where content inserted using document.write() is actually inserted. The insertion point is relative to the position of the character immediately after it, it is not an absolute offset into the input stream. Initially, the insertion point is uninitialised.

The "EOF" character in the tables below is a conceptual character representing the end of the input stream. If the parser is a script-created parser, then the end of the input stream is reached when an explicit "EOF" character (inserted by the document.close() method) is consumed. Otherwise, the "EOF" character is not a real character in the stream, but rather the lack of any further characters.

8.2.3. Tokenisation

Implementations must act as if they used the following state machine to tokenise HTML. The state machine must start in the data state. Most states consume a single character, which may have various side-effects, and either switches the state machine to a new state to reconsume the same character, or switches it to a new state (to consume the next character), or repeats the same state (to consume the next character). Some states have more complicated behaviour and can consume several characters before switching to another state.

The exact behaviour of certain states depends on a content model flag that is set after certain tokens are emitted. The flag has several states: PCDATA, RCDATA, CDATA, and PLAINTEXT. Initially it is in the PCDATA state.

The output of the tokenisation step is a series of zero or more of the following tokens: DOCTYPE, start tag, end tag, comment, character, end-of-file. DOCTYPE tokens have names and can be either correct or in error. Start and end tag tokens have a tagname and a list of attributes, each of which has a name and a value. Comment and character tokens have data.

When a token is emitted, it must immediately be handled by the tree construction stage. The tree construction stage can affect the state of the content model flag, and can insert additional characters into the stream. (For example, the script element can result in scripts executing and using the dynamic markup insertion APIs to insert characters into the stream being tokenised.)

Before each step of the tokeniser, the user agent may check to see if either one of the scripts in the list of scripts that will execute as soon as possible or the first script in the list of scripts that will execute asynchronously, has completed loading. If one has, then it must be executed and removed from its list.

The tokeniser state machine is as follows:

Data state

Consume the next input character:

U+0026 AMPERSAND (&): When the content model flag is set to one of the PCDATA or RCDATA states: switch to the entity data state.; Otherwise: treat it as per the "anything else" entry below.
U+003C LESS-THAN SIGN (<): When the content model flag is set to a state other than the PLAINTEXT state: switch to the tag open state.; Otherwise: treat it as per the "anything else" entry below.
EOF: Emit an end-of-file token.
Anything else: Emit the input character as a character token. Stay in the data state.

Entity data state

(This cannot happen if the content model flag is set to the CDATA state.)

Attempt to consume an entity.

If nothing is returned, emit a U+0026 AMPERSAND character token.

Otherwise, emit the character token that was returned.

Finally, switch to the data state.

Tag open state

The behaviour of this state depends on the content model flag.

If the content model flag is set to the RCDATA or CDATA states

If the next input character is a U+002F SOLIDUS (/) character, consume it and switch to the close tag open state. If the next input character is not a U+002F SOLIDUS (/) character, emit a U+003C LESS-THAN SIGN character token and reconsume the current input character in the data state.

If the content model flag is set to the PCDATA state

Consume the next input character:

U+0021 EXCLAMATION MARK (!): Switch to the markup declaration open state.
U+002F SOLIDUS (/): Switch to the close tag open state.
U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Create a new start tag token, set its tag name to the lowercase version of the input character (add 0x0020 to the character's code point), then switch to the tag name state. (Don't emit the token yet; further details will be filled in before it is emitted.)
U+0061 LATIN SMALL LETTER A through to U+007A LATIN SMALL LETTER Z: Create a new start tag token, set its tag name to the input character, then switch to the tag name state. (Don't emit the token yet; further details will be filled in before it is emitted.)
U+003E GREATER-THAN SIGN (>): Parse error. Emit a U+003C LESS-THAN SIGN character token and a U+003E GREATER-THAN SIGN character token. Switch to the data state.
U+003F QUESTION MARK (?): Parse error. Switch to the bogus comment state.
Anything else: Parse error. Emit a U+003C LESS-THAN SIGN character token and reconsume the current input character in the data state.

Close tag open state

If the content model flag is set to the RCDATA or CDATA states then examine the next few characters. If they do not match the tag name of the last start tag token emitted (case insensitively), or if they do but they are not immediately followed by one of the following characters:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE
U+003E GREATER-THAN SIGN (>)
U+002F SOLIDUS (/)
U+003C LESS-THAN SIGN (<)
EOF

...then there is a parse error. Emit a U+003C LESS-THAN SIGN character token, a U+002F SOLIDUS character token, and reconsume the current input character in the data state.

Otherwise, if the content model flag is set to the PCDATA state, or if the next few characters do match that tag name, consume the next input character:

U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Create a new end tag token, set its tag name to the lowercase version of the input character (add 0x0020 to the character's code point), then switch to the tag name state. (Don't emit the token yet; further details will be filled in before it is emitted.)
U+0061 LATIN SMALL LETTER A through to U+007A LATIN SMALL LETTER Z: Create a new end tag token, set its tag name to the input character, then switch to the tag name state. (Don't emit the token yet; further details will be filled in before it is emitted.)
U+003E GREATER-THAN SIGN (>): Parse error. Switch to the data state.
EOF: Parse error. Emit a U+003C LESS-THAN SIGN character token and a U+002F SOLIDUS character token. Reconsume the EOF character in the data state.
Anything else: Parse error. Switch to the bogus comment state.

Tag name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Switch to the before attribute name state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Append the lowercase version of the current input character (add 0x0020 to the character's code point) to the current tag token's tag name. Stay in the tag name state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
U+002F SOLIDUS (/): Parse error unless this is a permitted slash. Switch to the before attribute name state.
Anything else: Append the current input character to the current tag token's tag name. Stay in the tag name state.

Before attribute name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Stay in the before attribute name state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Start a new attribute in the current tag token. Set that attribute's name to the lowercase version of the current input character (add 0x0020 to the character's code point), and its value to the empty string. Switch to the attribute name state.
U+002F SOLIDUS (/): Parse error unless this is a permitted slash. Stay in the before attribute name state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Start a new attribute in the current tag token. Set that attribute's name to the current input character, and its value to the empty string. Switch to the attribute name state.

Attribute name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Switch to the after attribute name state.
U+003D EQUALS SIGN (=): Switch to the before attribute value state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Append the lowercase version of the current input character (add 0x0020 to the character's code point) to the current attribute's name. Stay in the attribute name state.
U+002F SOLIDUS (/): Parse error unless this is a permitted slash. Switch to the before attribute name state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Append the current input character to the current attribute's name. Stay in the attribute name state.

When the user agent leaves the attribute name state (and before emitting the tag token, if appropriate), the complete attribute's name must be compared to the other attributes on the same token; if there is already an attribute on the token with the exact same name, then this is a parse error and the new attribute must be dropped, along with the value that gets associated with it (if any).

After attribute name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Stay in the after attribute name state.
U+003D EQUALS SIGN (=): Switch to the before attribute value state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+0041 LATIN CAPITAL LETTER A through to U+005A LATIN CAPITAL LETTER Z: Start a new attribute in the current tag token. Set that attribute's name to the lowercase version of the current input character (add 0x0020 to the character's code point), and its value to the empty string. Switch to the attribute name state.
U+002F SOLIDUS (/): Parse error unless this is a permitted slash. Switch to the before attribute name state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Start a new attribute in the current tag token. Set that attribute's name to the current input character, and its value to the empty string. Switch to the attribute name state.

Before attribute value state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Stay in the before attribute value state.
U+0022 QUOTATION MARK ("): Switch to the attribute value (double-quoted) state.
U+0026 AMPERSAND (&): Switch to the attribute value (unquoted) state and reconsume this input character.
U+0027 APOSTROPHE ('): Switch to the attribute value (single-quoted) state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Append the current input character to the current attribute's value. Switch to the attribute value (unquoted) state.

Attribute value (double-quoted) state

Consume the next input character:

U+0022 QUOTATION MARK ("): Switch to the before attribute name state.
U+0026 AMPERSAND (&): Switch to the entity in attribute value state.
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Append the current input character to the current attribute's value. Stay in the attribute value (double-quoted) state.

Attribute value (single-quoted) state

Consume the next input character:

U+0027 APOSTROPHE ('): Switch to the before attribute name state.
U+0026 AMPERSAND (&): Switch to the entity in attribute value state.
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Append the current input character to the current attribute's value. Stay in the attribute value (single-quoted) state.

Attribute value (unquoted) state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Switch to the before attribute name state.
U+0026 AMPERSAND (&): Switch to the entity in attribute value state.
U+003E GREATER-THAN SIGN (>): Emit the current tag token. Switch to the data state.
U+003C LESS-THAN SIGN (<)
EOF: Parse error. Emit the current tag token. Reconsume the character in the data state.
Anything else: Append the current input character to the current attribute's value. Stay in the attribute value (unquoted) state.

Entity in attribute value state

Attempt to consume an entity.

If nothing is returned, append a U+0026 AMPERSAND character to the current attribute's value.

Otherwise, append the returned character token to the current attribute's value.

Finally, switch back to the attribute value state that you were in when were switched into this state.

Bogus comment state

(This can only happen if the content model flag is set to the PCDATA state.)

Consume every character up to the first U+003E GREATER-THAN SIGN character (>) or the end of the file (EOF), whichever comes first. Emit a comment token whose data is the concatenation of all the characters starting from and including the character that caused the state machine to switch into the bogus comment state, up to and including the last consumed character before the U+003E character, if any, or up to the end of the file otherwise. (If the comment was started by the end of the file (EOF), the token is empty.)

Switch to the data state.

If the end of the file was reached, reconsume the EOF character.

Markup declaration open state

(This can only happen if the content model flag is set to the PCDATA state.)

If the next two characters are both U+002D HYPHEN-MINUS (-) characters, consume those two characters, create a comment token whose data is the empty string, and switch to the comment state.

Otherwise if the next seven chacacters are a case-insensitive match for the word "DOCTYPE", then consume those characters and switch to the DOCTYPE state.

Otherwise, is is a parse error. Switch to the bogus comment state. The next character that is consumed, if any, is the first character that will be in the comment.

Comment state

Consume the next input character:

U+002D HYPHEN-MINUS (-): Switch to the comment dash state
EOF: Parse error. Emit the comment token. Reconsume the EOF character in the data state.
Anything else: Append the input character to the comment token's data. Stay in the comment state.

Comment dash state

Consume the next input character:

U+002D HYPHEN-MINUS (-): Switch to the comment end state
EOF: Parse error. Emit the comment token. Reconsume the EOF character in the data state.
Anything else: Append a U+002D HYPHEN-MINUS (-) character and the input character to the comment token's data. Switch to the comment state.

Comment end state

Consume the next input character:

U+003E GREATER-THAN SIGN (>): Emit the comment token. Switch to the data state.
U+002D HYPHEN-MINUS (-): Parse error. Append a U+002D HYPHEN-MINUS (-) character to the comment token's data. Stay in the comment end state.
EOF: Parse error. Emit the comment token. Reconsume the EOF character in the data state.
Anything else: Parse error. Append two U+002D HYPHEN-MINUS (-) characters and the input character to the comment token's data. Switch to the comment state.

DOCTYPE state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Switch to the before DOCTYPE name state.
Anything else: Parse error. Reconsume the current character in the before DOCTYPE name state.

Before DOCTYPE name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Stay in the before DOCTYPE name state.
U+0061 LATIN SMALL LETTER A through to U+007A LATIN SMALL LETTER Z: Create a new DOCTYPE token. Set the token's name name to the uppercase version of the current input character (subtract 0x0020 from the character's code point), and mark it as being in error. Switch to the DOCTYPE name state.
U+003E GREATER-THAN SIGN (>): Parse error. Emit a DOCTYPE token whose name is the empty string and that is marked as being in error. Switch to the data state.
EOF: Parse error. Emit a DOCTYPE token whose name is the empty string and that is marked as being in error. Reconsume the EOF character in the data state.
Anything else: Create a new DOCTYPE token. Set the token's name name to the current input character, and mark it as being in error. Switch to the DOCTYPE name state.

DOCTYPE name state

First, consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Switch to the after DOCTYPE name state.
U+003E GREATER-THAN SIGN (>): Emit the current DOCTYPE token. Switch to the data state.
U+0061 LATIN SMALL LETTER A through to U+007A LATIN SMALL LETTER Z: Append the uppercase version of the current input character (subtract 0x0020 from the character's code point) to the current DOCTYPE token's name. Stay in the DOCTYPE name state.
EOF: Parse error. Emit the current DOCTYPE token. Reconsume the EOF character in the data state.
Anything else: Append the current input character to the current DOCTYPE token's name. Stay in the DOCTYPE name state.

Then, if the name of the DOCTYPE token is exactly the four letters "HTML", then mark the token as being correct. Otherwise, mark it as being in error.

Because lowercase letters in the name are uppercased by the algorithm above, the "HTML" letters are actually case-insensitive relative to the markup.

After DOCTYPE name state

Consume the next input character:

U+0009 CHARACTER TABULATION
U+000A LINE FEED (LF)
U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0020 SPACE: Stay in the after DOCTYPE name state.
U+003E GREATER-THAN SIGN (>): Emit the current DOCTYPE token. Switch to the data state.
EOF: Parse error. Emit the current DOCTYPE token. Reconsume the EOF character in the data state.
Anything else: Parse error. Mark the DOCTYPE token as being in error, if it is not already. Switch to the bogus DOCTYPE state.

Bogus DOCTYPE state

Consume the next input character:

U+003E GREATER-THAN SIGN (>): Emit the current DOCTYPE token. Switch to the data state.
EOF: Parse error. Emit the current DOCTYPE token. Reconsume the EOF character in the data state.
Anything else: Stay in the bogus DOCTYPE state.

When an end tag token is emitted, the content model flag must be switched to the PCDATA state.

When an end tag token is emitted with attributes, that is a parse error.

A permitted slash is a U+002F SOLIDUS character that is immediately followed by a U+003E GREATER-THAN SIGN, if, and only if, the current token being processed is a start tag token whose tag name is one of the following: base, link, meta, hr, br, img, embed, param, area, col, input

8.2.3.1. Tokenising entities

This section defines how to consume an entity. This definition is used when parsing entities in text and in attributes.

The behaviour depends on the identity of the next character (the one immediately after the U+0026 AMPERSAND character):

U+0023 NUMBER SIGN (#)

Consume the U+0023 NUMBER SIGN.

The behaviour further depends on the character after the U+0023 NUMBER SIGN:

U+0078 LATIN SMALL LETTER X

U+0058 LATIN CAPITAL LETTER X

Consume the X.

Follow the steps below, but using the range of characters U+0030 DIGIT ZERO through to U+0039 DIGIT NINE, U+0061 LATIN SMALL LETTER A through to U+0078 LATIN SMALL LETTER F, and U+0041 LATIN CAPITAL LETTER A, through to U+0058 LATIN CAPITAL LETTER F (in other words, 0-9, A-F, a-f).

When it comes to interpreting the number, interpret it as a hexadecimal number.

Anything else

Follow the steps below, but using the range of characters U+0030 DIGIT ZERO through to U+0039 DIGIT NINE (i.e. just 0-9).

When it comes to interpreting the number, interpret it as a decimal number.

Consume as many characters as match the range of characters given above.

If no characters match the range, then don't consume any characters (and unconsume the U+0023 NUMBER SIGN character and, if appropriate, the X character). This is a parse error; nothing is returned.

Otherwise, if the next character is a U+003B SEMICOLON, consume that too. If it isn't, there is a parse error.

If one or more characters match the range, then take them all and interpret the string of characters as a number (either hexadecimal or decimal as appropriate), and return a character token for the Unicode character whose code point is that number. If the number is not a valid Unicode character (e.g. if the number is higher than 1114111), or if the number is zero, then return a character token for the U+FFFD REPLACEMENT CHARACTER character instead.

Anything else

Consume the maximum number of characters possible, with the consumed characters case-sensitively matching one of the identifiers in the first column of the entities table.

If no match can be made, then this is a parse error. No characters are consumed, and nothing is returned.

Otherwise, if the next character is a U+003B SEMICOLON, consume that too. If it isn't, there is a parse error.

Return a character token for the character corresponding to the entity name (as given by the second column of the entities table).

If the markup contains I'm &notit without you, the entity is parsed as "not", as in, I'm ¬it without you. But if the markup was I'm &notin without you, the entity would be parsed as "notin", resulting in I'm ∉ without you.

This isn't quite right. For some entities, UAs require a semicolon, for others they don't. We probably need to do the same for backwards compatibility. If we do that we might be able to add more entities, e.g. for mathematics. Probably the way to mark whether or not an entity requires a semicolon is with an additional column in the entity table lower down.

8.2.4. Tree construction

The input to the tree construction stage is a sequence of tokens from the tokenisation stage. The tree construction stage is associated with a DOM Document object when a parser is created. The "output" of this stage consists of dynamically modifying or extending that document's DOM tree.

Tree construction passes through several phases. Initially, UAs must act according to the steps described as being those of the initial phase.

This specification does not define when an interactive user agent has to render the Document available to the user, or when it has to begin accepting user input.

When the steps below require the UA to append a character to a node, the UA must collect it and all subsequent consecutive characters that would be appended to that node, and insert one Text node whose data is the concatenation of all those characters.

DOM mutation events must not fire for changes caused by the UA parsing the document. (Conceptually, the parser is not mutating the DOM, it is constructing it.) This includes the parsing of any content inserted using document.write() and document.writeln() calls. [DOM3EVENTS]

Not all of the tag names mentioned below are conformant tag names in this specification; many are included to handle legacy content. They still form part of the algorithm that implementations are required to implement to claim conformance.

The algorithm described below places no limit on the depth of the DOM tree generated, or on the length of tag names, attribute names, attribute values, text nodes, etc. While implementators are encouraged to avoid arbitrary limits, it is recognised that practical concerns will likely force user agents to impose nesting depths.

8.2.4.1. The initial phase

Initially, the tree construction stage must handle each token emitted from the tokenisation stage as follows:

A DOCTYPE token that is marked as being in error

A comment token

A start tag token

An end tag token

A character token that is not one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

An end-of-file token

This specification does not define how to handle this case. In particular, user agents may ignore the entirety of this specification altogether for such documents, and instead invoke special parse modes with a greater emphasis on backwards compatibility.

Browsers in particular have generally used DOCTYPE-based sniffing to invoke an "alternative conformance mode" known as quirks mode on certain documents. In this mode, emphasis is put on legacy compatibility rather than on standards compliance. This specification takes no position on this behaviour; documents without DOCTYPEs or with DOCTYPEs that do not conform to the syntax allowed by this specification are considered to be out of scope of this specification.

As far as parsing goes, the quirks I know of are:

Comment parsing is different.

The following is considered one script block (!):

<script><!-- document.write('</script>'); --></script>

</br> and </p> do magical things.
p can contain table
Safari and IE have special parsing rules for <% ... %> (even in standards mode, though clearly this should be quirks-only).

Maybe we should just adopt all those and be done with it. One parsing mode to rule them all. Or legitimise/codify the quirks mode parsing in some way.

Would be interesting to do a search to see how many pages hit each of the above.

A DOCTYPE token marked as being correct

Append a DocumentType node to the Document node, with the name attribute set to the name given in the DOCTYPE token (which will be "HTML"), and the other attributes specific to DocumentType objects set to null, empty lists, or the empty string as appropriate.

Then, switch to the root element phase of the tree construction stage.

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Append that character to the Document node.

8.2.4.2. The root element phase

After the initial phase, as each token is emitted from the tokenisation stage, it must be processed as described in this section.

A DOCTYPE token

Parse error. Ignore the token.

A comment token

Append a Comment node to the Document object with the data attribute set to the data given in the comment token.

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Append that character to the Document node.

A character token that is not one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

A start tag token

An end tag token

An end-of-file token

Create an HTMLElement node with the tag name html, in the HTML namespace. Append it to the Document object. Switch to the main phase and reprocess the current token.

Should probably make end tags be ignored, so that "</head><html>" puts the comment befor the root node (or should we?)

The root element can end up being removed from the Document object, e.g. by scripts; nothing in particular happens in such cases, content continues being appended to the nodes as described in the next section.

8.2.4.3. The main phase

After the root element phase, each token emitted from the tokenisation stage must be processed as described in this section. This is by far the most involved part of parsing an HTML document.

The tree construction stage in this phase has several pieces of state: a stack of open elements, a list of active formatting elements, a head element pointer, a form element pointer, and an insertion mode.

We could just fold insertion modes and phases into one concept (and duplicate the two rules common to all insertion modes into all of them).

8.2.4.3.1. The stack of open elements

Initially the stack of open elements contains just the html root element node created in the last phase before switching to this phase (or, in the innerHTML case, the html element created to represent the element whose innerHTML attribute is being set). That's the topmost node of the stack. It never gets popped off the stack. (This stack grows downwards.)

The current node is the bottommost node in this stack.

Elements in the stack fall into the following categories:

Special: The following HTML elements have varying levels of special parsing rules: address, area, base, basefont, bgsound, blockquote, body, br, center, col, colgroup, dd, dir, div, dl, dt, embed, fieldset, form, frame, frameset, h1, h2, h3, h4, h5, h6, head, hr, iframe, image, img, input, isindex, li, link, listing, menu, meta, noembed, noframes, noscript, ol, optgroup, option, p, param, plaintext, pre, script, select, spacer, style, tbody, textarea, tfoot, thead, title, tr, ul, and wbr.
Scoping: The following HTML elements introduce new scopes for various parts of the parsing: button, caption, html, marquee, object, table, td and th.
Formatting: The following HTML elements are those that end up in the list of active formatting elements: a, b, big, em, font, i, nobr, s, small, strike, strong, tt, and u.
Phrasing: All other elements found while parsing an HTML document.

Still need to add these new elements to the lists: event-source, section, nav, article, aside, header, footer, datagrid, command

The stack of open elements is said to have an element in scope or have an element in table scope when the following algorithm terminates in a match state:

Initialise node to be the current node (the bottommost node of the stack).
If node is the target node, terminate in a match state.
Otherwise, if node is a table element, terminate in a failure state.
Otherwise, if the algorithm is the "has an element in scope" variant (rather than the "has an element in table scope" variant), and node is one of the following, terminate in a failure state:
- caption
- td
- th
- button
- marquee
- object
Otherwise, if node is an html element, terminate in a failure state. (This can only happen if the node is the topmost node of the stack of open elements, and prevents the next step from being invoked if there are no more elements in the stack.)
Otherwise, set node to the previous entry in the stack of open elements and return to step 2. (This will never fail, since the loop will always terminate in the previous step if the top of the stack is reached.)

Nothing happens if at any time any of the elements in the stack of open elements are moved to a new location in, or removed from, the Document tree. In particular, the stack is not changed in this situation. This can cause, amongst other strange effects, content to be appended to nodes that are no longer in the DOM.

In some cases (namely, when closing misnested formatting elements), the stack is manipulated in a random-access fashion.

8.2.4.3.2. The list of active formatting elements

Initially the list of active formatting elements is empty. It is used to handle mis-nested formatting element tags.

The list contains elements in the formatting category, and scope markers. The scope markers are inserted when entering buttons, object elements, marquees, table cells, and table captions, and are used to prevent formatting from "leaking" into tables, buttons, object elements, and marquees.

When the steps below require the UA to reconstruct the active formatting elements, the UA must perform the following steps:

If there are no entries in the list of active formatting elements, then there is nothing to reconstruct; stop this algorithm.
If the last (most recently added) entry in the list of active formatting elements is a marker, or if it is an element that is in the stack of open elements, then there is nothing to reconstruct; stop this algorithm.
Let entry be the last (most recently added) element in the list of active formatting elements.
If there are no entries before entry in the list of active formatting elements, then jump to step 8.
Let entry be the entry one earlier than entry in the list of active formatting elements.
If entry is neither a marker nor an element that is also in the stack of open elements, go to step 4.
Let entry be the element one later than entry in the list of active formatting elements.
Perform a shallow clone of the element entry to obtain clone. [DOM3CORE]
Append clone to the current node and push it onto the stack of open elements so that it is the new current node.
Replace the entry for entry in the list with an entry for clone.
If the entry for clone in the list of active formatting elements is not the last entry in the list, return to step 7.

This has the effect of reopening all the formatting elements that were opened in the current body, cell, or caption (whichever is youngest) that haven't been explicitly closed.

The way this specification is written, the list of active formatting elements always consists of elements in chronological order with the least recently added element first and the most recently added element last (except for while steps 8 to 11 of the above algorithm are being executed, of course).

When the steps below require the UA to clear the list of active formatting elements up to the last marker, the UA must perform the following steps:

Let entry be the last (most recently added) entry in the list of active formatting elements.
Remove entry from the list of active formatting elements.
If entry was a marker, then stop the algorithm at this point. The list has been cleared up to the last marker.
Go to step 1.

8.2.4.3.3. Creating and inserting HTML elements

When the steps below require the UA to create an element for a token, the UA must create a node implementing the interface appropriate for the element type corresponding to the tag name of the token (as given in the section of this specification that defines that element, e.g. for an a element it would be the HTMLAnchorElement interface), with the tag name being the name of that element, with the node being in the HTML namespace, and with the attributes on the node being those given in the given token.

When the steps below require the UA to insert an HTML element for a token, the UA must first create an element for the token, and then append this node to the current node, and push it onto the stack of open elements so that it is the new current node.

The steps below may also require that the UA insert an HTML element in a particular place, in which case the UA must create an element for the token and then insert or append the new node in the location specified. (This happens in particular during the parsing of tables with invalid content.)

The interface appropriate for an element that is not defined in this specification is HTMLElement.

8.2.4.3.4. Closing elements that have implied end tags

When the steps below require the UA to generate implied end tags, then, if the current node is a dd element, a dt element, an li element, a p element, a td element, a th element, or a tr element, the UA must act as if an end tag with the respective tag name had been seen and then generate implied end tags again.

The step that requires the UA to generate implied end tags but lists an element to exclude from the process, then the UA must perform the above steps as if that element was not in the above list.

8.2.4.3.5. The element pointers

Initially the head element pointer and the form element pointer are both null.

Once a head element has been parsed (whether implicitly or explicitly) the head element pointer gets set to point to this node.

The form element pointer points to the last form element that was opened and whose end tag has not yet been seen. It is used to make form controls associate with forms in the face of dramatically bad markup, for historical reasons.

8.2.4.3.6. The insertion mode

Initially the insertion mode is "before head". It can change to "in head", "after head", "in body", "in table", "in caption", "in column group", "in table body", "in row", "in cell", "in select", "after body", "in frameset", and "after frameset" during the course of the parsing, as described below. It affects how certain tokens are processed.

If the tree construction stage is switched from the main phase to the trailing end phase and back again, the various pieces of state are not reset; the UA must act as if the state was maintained.

When the steps below require the UA to reset the insertion mode appropriately, it means the UA must follow these steps:

Let last be false.
Let node be the last node in the stack of open elements.
If node is the first node in the stack of open elements, then set last to true. If the element whose innerHTML attribute is being set is neither a td element nor a th element, then set node to the element whose innerHTML attribute is being set. (innerHTML case)
If node is a select element, then switch the insertion mode to "in select" and abort these steps. (innerHTML case)
If node is a td or th element, then switch the insertion mode to "in cell" and abort these steps.
If node is a tr element, then switch the insertion mode to "in row" and abort these steps.
If node is a tbody, thead, or tfoot element, then switch the insertion mode to "in table body" and abort these steps.
If node is a caption element, then switch the insertion mode to "in caption" and abort these steps.
If node is a colgroup element, then switch the insertion mode to "in column group" and abort these steps. (innerHTML case)
If node is a table element, then switch the insertion mode to "in table" and abort these steps.
If node is a head element, then switch the insertion mode to "in body" ("in body"! not "in head"!) and abort these steps. (innerHTML case)
If node is a body element, then switch the insertion mode to "in body" and abort these steps.
If node is a frameset element, then switch the insertion mode to "in frameset" and abort these steps. (innerHTML case)
If node is an html element, then: if the head element pointer is null, switch the insertion mode to "before head", otherwise, switch the insertion mode to "after head". In either case, abort these steps. (innerHTML case)
If last is true, then set the insertion mode to "in body" and abort these steps. (innerHTML case)
Let node now be the node before node in the stack of open elements.
Return to step 3.

8.2.4.3.7. How to handle tokens in the main phase

Tokens in the main phase must be handled as follows:

A DOCTYPE token

Parse error. Ignore the token.

A start tag token with the tag name "html"

If this start tag token was not the first start tag token, then it is a parse error.

For each attribute on the token, check to see if the attribute is already present on the top element of the stack of open elements. If it is not, add the attribute and its corresponding value to that element.

An end-of-file token

If there are more than two nodes on the stack of open elements, or if there are two nodes but the second node is not a body node, this is a parse error.

Otherwise, if the parser was originally created in order to handle the setting of an element's innerHTML attribute, and there's more than one element in the stack of open elements, and the second node on the stack of open elements is not a body node, then this is a parse error. (innerHTML case)

Stop parsing.

This fails because it doesn't imply HEAD and BODY tags. We should probably expand out the insertion modes and merge them with phases and then put the three things here into each insertion mode instead of trying to factor them out so carefully.

Anything else

Depends on the insertion mode:

If the insertion mode is "before head"

Handle the token as follows:

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag token with the tag name "head"

Set the head element pointer to this new element node.

Append the new element to the current node and push it onto the stack of open elements.

Change the insertion mode to "in head".

A start tag token whose tag name is one of: "base", "link", "meta", "script", "style", "title"

Act as if a start tag token with the tag name "head" and no attributes had been seen, then reprocess the current token.

This will result in a head element being generated, and with the current token being reprocessed in the "in head" insertion mode.

An end tag with the tag name "html"

Act as if a start tag token with the tag name "head" and no attributes had been seen, then reprocess the current token.

Any other end tag

Parse error. Ignore the token.

A character token that is not one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Any other start tag token

Act as if a start tag token with the tag name "head" and no attributes had been seen, then reprocess the current token.

This will result in an empty head element being generated, with the current token being reprocessed in the "after head" insertion mode.

If the insertion mode is "in head"

Handle the token as follows.

The rules for handling "title", "style", and "script" start tags are similar, but not identical.

It is possible for the tree construction stage's main phase to be in the "in head" insertion mode without the current node being a head element, e.g. if a head end tag is immediately followed by a meta start tag.

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag with the tag name "title"

Append the new element to the node pointed to by the head element pointer, or, if that is null (innerHTML case), to the current node.

Switch the tokeniser's content model flag to the RCDATA state.

Then, collect all the character tokens that the tokeniser returns until it returns a token that is not a character token.

If this process resulted in a collection of character tokens, append a single Text node to the title element node whose contents is the concatenation of all those tokens' characters.

The tokeniser's content model flag will have switched back to the PCDATA state.

If the next token is an end tag token with the tag name "title", ignore it. Otherwise, this is a parse error.

A start tag with the tag name "style"

Switch the tokeniser's content model flag to the CDATA state.

Then, collect all the character tokens that the tokeniser returns until it returns a token that is not a character token, or until it stops tokenising.

If this process resulted in a collection of character tokens, append a single Text node to the style element node whose contents is the concatenation of all those tokens' characters.

The tokeniser's content model flag will have switched back to the PCDATA state.

If the next token is an end tag token with the tag name "style", ignore it. Otherwise, this is a parse error.

A start tag with the tag name "script"

Mark the element as being "parser-inserted". This ensures that, if the script is external, any document.write() calls in the script will execute in-line, instead of blowing the document away, as would happen in most other cases.

Switch the tokeniser's content model flag to the CDATA state.

Then, collect all the character tokens that the tokeniser returns until it returns a token that is not a character token, or until it stops tokenising.

If this process resulted in a collection of character tokens, append a single Text node to the script element node whose contents is the concatenation of all those tokens' characters.

The tokeniser's content model flag will have switched back to the PCDATA state.

If the next token is not an end tag token with the tag name "script", then this is a parse error; mark the script element as "already executed". Otherwise, the token is the script element's end tag, so ignore it.

If the parser was originally created in order to handle the setting of a node's innerHTML attribute, then mark the script element as "already executed", and skip the rest of the processing described for this token (including the part below where "scripts that will execute as soon as the parser resumes" are executed). (innerHTML case)

Marking the script element as "already executed" prevents it from executing when it is inserted into the document a few paragraphs below. Scripts missing their end tags and scripts that were inserted using innerHTML aren't executed.

Let the old insertion point have the same value as the current insertion point. Let the insertion point be just before the next input character.

Append the new element to the current node, unless the insertion mode is "in head" and the head element pointer is not null, in which case append it to the node pointed to by the head element pointer. Special processing occurs when a script element is inserted into a document that might cause some script to execute, which might cause new characters to be inserted into the tokeniser.

Let the insertion point have the value of the old insertion point. (In other words, restore the insertion point to the value it had before the previous paragraph. This value might be the "undefined" value.)

At this stage, if there is a script that will execute as soon as the parser resumes, then:

If the tree construction stage is being called reentrantly, say from a call to document.write():

Abort the processing of any nested invokations of the tokeniser, yielding control back to the caller. (Tokenisation will resume when the caller returns to the "outer" tree construction stage.)

Otherwise:

Follow these steps:

Let the script be the script that will execute as soon as the parser resumes. There is no longer a script that will execute as soon as the parser resumes.
Pause until the script has completed loading.
Let the insertion point be just before the next input character.
Execute the script.
Let the insertion point be undefined again.
If there is once again a script that will execute as soon as the parser resumes, then repeat these steps from step 1.

A start tag with the tag name "base", "link", or "meta"

Append the new element to the node pointed to by the head element pointer, or, if that is null (innerHTML case), to the current node.

Need to cope with second and subsequent base elements affecting subsequent elements magically.

An end tag with the tag name "head"

If the current node is a head element, pop the current node off the stack of open elements. Otherwise, this is a parse error.

Change the insertion mode to "after head".

An end tag with the tag name "html"

Act as described in the "anything else" entry below.

A start tag with the tag name "head"

Any other end tag

Parse error. Ignore the token.

Anything else

If the current node is a head element, act as if an end tag token with the tag name "head" had been seen.

Otherwise, change the insertion mode to "after head".

Then, reprocess the current token.

In certain UAs, some elements don't trigger the "in body" mode straight away, but instead get put into the head. Do we want to copy that?

If the insertion mode is "after head"

Handle the token as follows:

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag token with the tag name "body"

Insert a body element for the token.

Change the insertion mode to "in body".

A start tag token with the tag name "frameset"

Insert a frameset element for the token.

Change the insertion mode to "in frameset".

A start tag token whose tag name is one of: "base", "link", "meta", "script", "style", "title"

Parse error. Switch the insertion mode back to "in head" and reprocess the token.

Anything else

Act as if a start tag token with the tag name "body" and no attributes had been seen, and then reprocess the current token.

If the insertion mode is "in body"

Handle the token as follows:

A character token

Reconstruct the active formatting elements, if any.

Append the token's character to the current node.

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag token whose tag name is one of: "script", "style"

Process the token as if the insertion mode had been "in head".

A start tag token whose tag name is one of: "base", "link", "meta", "title"

Parse error. Process the token as if the insertion mode had been "in head".

A start tag token with the tag name "body"

Parse error.

If the second element on the stack of open elements is not a body element, or, if the stack of open elements has only one node on it, then ignore the token. (innerHTML case)

Otherwise, for each attribute on the token, check to see if the attribute is already present on the body element (the second element) on the stack of open elements. If it is not, add the attribute and its corresponding value to that element.

An end tag with the tag name "body"

If the second element in the stack of open elements is not a body element, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

this needs to handle closing of implied elements, but without closing them

If the current node is not the body element, then this is a parse error.

Change the insertion mode to "after body".

An end tag with the tag name "html"

Act as if an end tag with tag name "body" had been seen, then, if that token wasn't ignored, reprocess the current token.

The fake end tag token here can only be ignored in the innerHTML case.

A start tag whose tag name is one of: "address", "blockquote", "center", "dir", "div", "dl", "fieldset", "listing", "menu", "ol", "p", "ul"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token.

A start tag whose tag name is "pre"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token.

If the next token is a U+000A LINE FEED (LF) character token, then ignore that token and move on to the next one. (Newlines at the start of pre blocks are ignored as an authoring convenience.)

A start tag whose tag name is "form"

If the form element pointer is not null, ignore the token with a parse error.

Otherwise:

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token, and set the form element pointer to point to the element created.

A start tag whose tag name is "li"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Run the following algorithm:

Initialise node to be the current node (the bottommost node of the stack).
If node is an li element, then pop all the nodes from the current node up to node, including node, then stop this algorithm.
If node is not in the formatting category, and is not in the phrasing category, and is not an address or div element, then stop this algorithm.
Otherwise, set node to the previous entry in the stack of open elements and return to step 2.

Finally, insert an li element.

A start tag whose tag name is "dd" or "dt"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Run the following algorithm:

Initialise node to be the current node (the bottommost node of the stack).
If node is a dd or dt element, then pop all the nodes from the current node up to node, including node, then stop this algorithm.
If node is not in the formatting category, and is not in the phrasing category, and is not an address or div element, then stop this algorithm.
Otherwise, set node to the previous entry in the stack of open elements and return to step 2.

Finally, insert an HTML element with the same tag name as the token's.

A start tag token whose tag name is "plaintext"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token.

Switch the content model flag to the PLAINTEXT state.

Once a start tag with the tag name "plaintext" has been seen, that will be the last token ever seen other than character tokens (and the end-of-file token), because there is no way to switch the content model flag out of the PLAINTEXT state.

An end tag whose tag name is one of: "address", "blockquote", "center", "dir", "div", "dl", "fieldset", "listing", "menu", "ol", "pre", "ul"

If the stack of open elements has an element in scope with the same tag name as that of the token, then generate implied end tags.

Now, if the current node is not an element with the same tag name as that of the token, then this is a parse error.

If the stack of open elements has an element in scope with the same tag name as that of the token, then pop elements from this stack until an element with that tag name has been popped from the stack.

An end tag whose tag name is "form"

If the stack of open elements has an element in scope with the same tag name as that of the token, then generate implied end tags.

Now, if the current node is not an element with the same tag name as that of the token, then this is a parse error.

In any case, set the form element pointer to null.

An end tag whose tag name is "p"

If the stack of open elements has a p element in scope, then generate implied end tags, except for p elements.

If the current node is not a p element, then this is a parse error.

If the stack of open elements has a p element in scope, then pop elements from this stack until the stack no longer has a p element in scope.

An end tag whose tag name is "dd", "dt", or "li"

If the stack of open elements has an element in scope whose tag name matches the tag name of the token, then generate implied end tags, except for elements with the same tag name as the token.

If the current node is not an element with the same tag name as the token, then this is a parse error.

If the stack of open elements has an element in scope whose tag name matches the tag name of the token, then pop elements from this stack until an element with that tag name has been popped from the stack.

A start tag whose tag name is one of: "h1", "h2", "h3", "h4", "h5", "h6"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

If the stack of open elements has in scope an element whose tag name is one of "h1", "h2", "h3", "h4", "h5", or "h6", then this is a parse error; pop elements from the stack until an element with one of those tag names has been popped from the stack.

Insert an HTML element for the token.

An end tag whose tag name is one of: "h1", "h2", "h3", "h4", "h5", "h6"

If the stack of open elements has in scope an element whose tag name is one of "h1", "h2", "h3", "h4", "h5", or "h6", then generate implied end tags.

Now, if the current node is not an element with the same tag name as that of the token, then this is a parse error.

If the stack of open elements has in scope an element whose tag name is one of "h1", "h2", "h3", "h4", "h5", or "h6", then pop elements from the stack until an element with one of those tag names has been popped from the stack.

A start tag whose tag name is "a"

If the list of active formatting elements contains an element whose tag name is "a" between the end of the list and the last marker on the list (or the start of the list if there is no marker on the list), then this is a parse error; act as if an end tag with the tag name "a" had been seen, then remove that element from the list of active formatting elements and the stack of open elements if the end tag didn't already remove it (it might not have if the element is not in table scope).

In the non-conforming stream <a href="a">a<table><a href="b">b</table>x, the first a element would be closed upon seeing the second one, and the "x" character would be inside a link to "b", not to "a". This is despite the fact that the outer a element is not in table scope (meaning that a regular </a> end tag at the start of the table wouldn't close the outer a element).

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token. Add that element to the list of active formatting elements.

A start tag whose tag name is one of: "b", "big", "em", "font", "i", "nobr", "s", "small", "strike", "strong", "tt", "u"

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token. Add that element to the list of active formatting elements.

An end tag whose tag name is one of: "a", "b", "big", "em", "font", "i", "nobr", "s", "small", "strike", "strong", "tt", "u"

Follow these steps:

Let the formatting element be the last element in the list of active formatting elements that:
- is between the end of the list and the last scope marker in the list, if any, or the start of the list otherwise, and
- has the same tag name as the token.
If there is no such node, or, if that node is also in the stack of open elements but the element is not in scope, then this is a parse error. Abort these steps. The token is ignored.

Otherwise, if there is such a node, but that node is not in the stack of open elements, then this is a parse error; remove the element from the list, and abort these steps.

Otherwise, there is a formatting element and that element is in the stack and is in scope. If the element is not the current node, this is a parse error. In any case, proceed with the algorithm as written in the following steps.
Let the furthest block be the topmost node in the stack of open elements that is lower in the stack than the formatting element, and is not an element in the phrasing or formatting categories. There might not be one.
If there is no furthest block, then the UA must skip the subsequent steps and instead just pop all the nodes from the bottom of the stack of open elements, from the current node up to the formatting element, and remove the formatting element from the list of active formatting elements.
Let the common ancestor be the element immediately above the formatting element in the stack of open elements.
If the furthest block has a parent node, then remove the furthest block from its parent node.
Let a bookmark note the position of the formatting element in the list of active formatting elements relative to the elements on either side of it in the list.
Let node and last node be the furthest block. Follow these steps:
1. Let node be the element immediately prior to node in the stack of open elements.
2. If node is not in the list of active formatting elements, then remove node from the stack of open elements and then go back to step 1.
3. Otherwise, if node is the formatting element, then go to the next step in the overall algorithm.
4. Otherwise, if last node is the furthest block, then move the aforementioned bookmark to be immediately after the node in the list of active formatting elements.
5. If node has any children, perform a shallow clone of node, replace the entry for node in the list of active formatting elements with an entry for the clone, replace the entry for node in the stack of open elements with an entry for the clone, and let node be the clone.
6. Insert last node into node, first removing it from its previous parent node if any.
7. Let last node be node.
8. Return to step 1 of this inner set of steps.
Insert whatever last node ended up being in the previous step into the common ancestor node, first removing it from its previous parent node if any.
Perform a shallow clone of the formatting element.
Take all of the child nodes of the furthest block and append them to the clone created in the last step.
Append that clone to the furthest block.
Remove the formatting element from the list of active formatting elements, and insert the clone into the list of active formatting elements at the position of the aforementioned bookmark.
Remove the formatting element from the stack of open elements, and insert the clone into the stack of open elements immediately after (i.e. in a more deeply nested position than) the position of the furthest block in that stack.
Jump back to step 1 in this series of steps.

The way these steps are defined, only elements in the formatting category ever get cloned by this algorithm.

Because of the way this algorithm causes elements to change parents, it has been dubbed the "adoption agency algorithm" (in contrast with other possibly algorithms for dealing with misnested content, which included the "incest algorithm", the "secret affair algorithm", and the "Heisenberg algorithm").

A start tag token whose tag name is "button"

If the stack of open elements has a button element in scope, then this is a parse error; act as if an end tag with the tag name "button" had been seen, then reprocess the token.

Otherwise:

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token.

Insert a marker at the end of the list of active formatting elements.

A start tag token whose tag name is one of: "marquee", "object"

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token.

Insert a marker at the end of the list of active formatting elements.

An end tag token whose tag name is one of: "button", "marquee", "object"

If the stack of open elements has in scope an element whose tag name is the same as the tag name of the token, then generate implied end tags.

Now, if the current node is not an element with the same tag name as the token, then this is a parse error.

Now, if the stack of open elements has an element in scope whose tag name matches the tag name of the token, then pop elements from the stack until that element has been popped from the stack, and clear the list of active formatting elements up to the last marker.

A start tag token whose tag name is "xmp"

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token.

Switch the content model flag to the CDATA state.

A start tag whose tag name is "table"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token.

Change the insertion mode to "in table".

A start tag whose tag name is one of: "area", "basefont", "bgsound", "br", "embed", "img", "param", "spacer", "wbr"

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token. Immediately pop the current node off the stack of open elements.

A start tag whose tag name is "hr"

If the stack of open elements has a p element in scope, then act as if an end tag with the tag name p had been seen.

Insert an HTML element for the token. Immediately pop the current node off the stack of open elements.

A start tag whose tag name is "image"

Parse error. Change the token's tag name to "img" and reprocess it. (Don't ask.)

A start tag whose tag name is "input"

Reconstruct the active formatting elements, if any.

Insert an input element for the token.

If the form element pointer is not null, then associate the input element with the form element pointed to by the form element pointer.

Pop that input element off the stack of open elements.

A start tag whose tag name is "isindex"

Parse error.

If the form element pointer is not null, then ignore the token.

Otherwise:

Act as if a start tag token with the tag name "form" had been seen.

Act as if a start tag token with the tag name "hr" had been seen.

Act as if a start tag token with the tag name "p" had been seen.

Act as if a start tag token with the tag name "label" had been seen.

Act as if a stream of character tokens had been seen (see below for what they should say).

Act as if a start tag token with the tag name "input" had been seen, with all the attributes from the "isindex" token, except with the "name" attribute set to the value "isindex" (ignoring any explicit "name" attribute).

Act as if a stream of character tokens had been seen (see below for what they should say).

Act as if an end tag token with the tag name "label" had been seen.

Act as if an end tag token with the tag name "p" had been seen.

Act as if a start tag token with the tag name "hr" had been seen.

Act as if an end tag token with the tag name "form" had been seen.

The two streams of character tokens together should, together with the input element, express the equivalent of "This is a searchable index. Insert your search keywords here: (input field)" in the user's preferred language.

Then need to specify that if the form submission causes just a single form control, whose name is "isindex", to be submitted, then we submit just the value part, not the "isindex=" part.

A start tag whose tag name is "textarea"

If the form element pointer is not null, then associate the textarea element with the form element pointed to by the form element pointer.

Append the new element to the current node.

Switch the tokeniser's content model flag to the RCDATA state.

Then, collect all the character tokens that the tokeniser returns until it returns a token that is not a character token, or until it stops tokenising.

If this process resulted in a collection of character tokens, append a single Text node, whose contents is the concatenation of all those tokens' characters, to the new element node.

The tokeniser's content model flag will have switched back to the PCDATA state.

If the next token is an end tag token with the tag name "textarea", ignore it. Otherwise, this is a parse error.

A start tag whose tag name is one of: "iframe", "noembed", "noframes"

A start tag whose tag name is "noscript", if scripting is enabled:

For "iframe" tags, the node must be an HTMLIFrameElement object, for the other tags it must be an HTMLElement object.

Append the new element to the current node.

Switch the tokeniser's content model flag to the CDATA state.

Then, collect all the character tokens that the tokeniser returns until it returns a token that is not a character token, or until it stops tokenising.

If this process resulted in a collection of character tokens, append a single Text node, whose contents is the concatenation of all those tokens' characters, to the new element node.

The tokeniser's content model flag will have switched back to the PCDATA state.

If the next token is an end tag token with the same tag name as the start tag token, ignore it. Otherwise, this is a parse error.

A start tag whose tag name is "select"

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token.

Change the insertion mode to "in select".

A start or end tag whose tag name is one of: "caption", "col", "colgroup", "frame", "frameset", "head", "option", "optgroup", "tbody", "td", "tfoot", "th", "thead", "tr"

An end tag whose tag name is one of: "area", "basefont", "bgsound", "br", "embed", "hr", "iframe", "image", "img", "input", "isindex", "noembed", "noframes", "param", "select", "spacer", "table", "textarea", "wbr"

An end tag whose tag name is "noscript", if scripting is enabled:

Parse error. Ignore the token.

A start or end tag whose tag name is one of: "event-source", "section", "nav", "article", "aside", "header", "footer", "datagrid", "command"

Work in progress!

A start tag token not covered by the previous entries

Reconstruct the active formatting elements, if any.

Insert an HTML element for the token.

This element will be a phrasing element.

An end tag token not covered by the previous entries

Run the following algorithm:

Initialise node to be the current node (the bottommost node of the stack).
If node has the same tag name as the end tag token, then:
1. Generate implied end tags.
2. If the tag name of the end tag token does not match the tag name of the current node, this is a parse error.
3. Pop all the nodes from the current node up to node, including node, then stop this algorithm.
Otherwise, if node is in neither the formatting category nor the phrasing category, then this is a parse error. Stop this algorithm. The end tag token is ignored.
Set node to the previous entry in the stack of open elements.
Return to step 2.

If the insertion mode is "in table"

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Generate implied end tags.

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag whose tag name is "caption"

Clear the stack back to a table context. (See below.)

Insert a marker at the end of the list of active formatting elements.

Insert an HTML element for the token, then switch the insertion mode to "in caption".

A start tag whose tag name is "colgroup"

Clear the stack back to a table context. (See below.)

Insert an HTML element for the token, then switch the insertion mode to "in column group".

A start tag whose tag name is "col"

Act as if a start tag token with the tag name "colgroup" had been seen, then reprocess the current token.

A start tag whose tag name is one of: "tbody", "tfoot", "thead"

Clear the stack back to a table context. (See below.)

Insert an HTML element for the token, then switch the insertion mode to "in table body".

A start tag whose tag name is one of: "td", "th", "tr"

Act as if a start tag token with the tag name "tbody" had been seen, then reprocess the current token.

A start tag whose tag name is "table"

Parse error. Act as if an end tag token with the tag name "table" had been seen, then, if that token wasn't ignored, reprocess the current token.

The fake end tag token here can only be ignored in the innerHTML case.

An end tag whose tag name is "table"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

Now, if the current node is not a table element, then this is a parse error.

Pop elements from this stack until a table element has been popped from the stack.

Reset the insertion mode appropriately.

An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "tbody", "td", "tfoot", "th", "thead", "tr"

Parse error. Ignore the token.

Anything else

Parse error. Process the token as if the insertion mode was "in body", with the following exception:

If the current node is a table, tbody, tfoot, thead, or tr element, then, whenever a node would be inserted into the current node, it must instead be inserted into the foster parent element.

The foster parent element is the parent element of the last table element in the stack of open elements, if there is a table element and it has such a parent element. If there is no table element in the stack of open elements (innerHTML case), then the foster parent element is the first element in the stack of open elements (the html element). Otherwise, if there is a table element in the stack of open elements, but the last table element in the stack of open elements has no parent, or its parent node is not an element, then the foster parent element is the element before the last table element in the stack of open elements.

If the foster parent element is the parent element of the last table element in the stack of open elements, then the new node must be inserted immediately before the last table element in the stack of open elements in the foster parent element; otherwise, the new node must be appended to the foster parent element.

When the steps above require the UA to clear the stack back to a table context, it means that the UA must, while the current node is not a table element or an html element, pop elements from the stack of open elements. If this causes any elements to be popped from the stack, then this is a parse error.

The current node being an html element after this process is an innerHTML case.

If the insertion mode is "in caption"

An end tag whose tag name is "caption"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

Generate implied end tags.

Now, if the current node is not a caption element, then this is a parse error.

Pop elements from this stack until a caption element has been popped from the stack.

Clear the list of active formatting elements up to the last marker.

Switch the insertion mode to "in table".

A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "td", "tfoot", "th", "thead", "tr"

An end tag whose tag name is "table"

Parse error. Act as if an end tag with the tag name "caption" had been seen, then, if that token wasn't ignored, reprocess the current token.

The fake end tag token here can only be ignored in the innerHTML case.

An end tag whose tag name is one of: "body", "col", "colgroup", "html", "tbody", "td", "tfoot", "th", "thead", "tr"

Parse error. Ignore the token.

Anything else

Process the token as if the insertion mode was "in body".

If the insertion mode is "in column group"

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Clear the list of active formatting elements up to the last marker.

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag whose tag name is "col"

Insert a col element for the token. Immediately pop the current node off the stack of open elements.

An end tag whose tag name is "colgroup"

If the current node is the root html element, then this is a parse error, ignore the token. (innerHTML case)

Otherwise, pop the current node (which will be a colgroup element) from the stack of open elements. Switch the insertion mode to "in table".

An end tag whose tag name is "col"

Parse error. Ignore the token.

Anything else

Act as if an end tag with the tag name "colgroup" had been seen, and then, if that token wasn't ignored, reprocess the current token.

The fake end tag token here can only be ignored in the innerHTML case.

If the insertion mode is "in table body"

A start tag whose tag name is "tr"

Clear the stack back to a table body context. (See below.)

Insert a tr element for the token, then switch the insertion mode to "in row".

A start tag whose tag name is one of: "th", "td"

Parse error. Act as if a start tag with the tag name "tr" had been seen, then reprocess the current token.

An end tag whose tag name is one of: "tbody", "tfoot", "thead"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token.

Otherwise:

Clear the stack back to a table body context. (See below.)

Pop the current node from the stack of open elements. Switch the insertion mode to "in table".

A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "tfoot", "thead"

An end tag whose tag name is "table"

If the stack of open elements does not have a tbody, thead, or tfoot element in table scope, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

Clear the stack back to a table body context. (See below.)

Act as if an end tag with the same tag name as the current node ("tbody", "tfoot", or "thead") had been seen, then reprocess the current token.

An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "td", "th", "tr"

Parse error. Ignore the token.

Anything else

Process the token as if the insertion mode was "in table".

When the steps above require the UA to clear the stack back to a table body context, it means that the UA must, while the current node is not a tbody, tfoot, thead, or html element, pop elements from the stack of open elements. If this causes any elements to be popped from the stack, then this is a parse error.

The current node being an html element after this process is an innerHTML case.

If the insertion mode is "in row"

A start tag whose tag name is one of: "th", "td"

Clear the stack back to a table row context. (See below.)

Insert an HTML element for the token, then switch the insertion mode to "in cell".

Insert a marker at the end of the list of active formatting elements.

An end tag whose tag name is "tr"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

Clear the stack back to a table row context. (See below.)

Pop the current node (which will be a tr element) from the stack of open elements. Switch the insertion mode to "in table body".

A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "tfoot", "thead", "tr"

An end tag whose tag name is "table"

Act as if an end tag with the tag name "tr" had been seen, then, if that token wasn't ignored, reprocess the current token.

The fake end tag token here can only be ignored in the innerHTML case.

An end tag whose tag name is one of: "tbody", "tfoot", "thead"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token.

Otherwise, act as if an end tag with the tag name "tr" had been seen, then reprocess the current token.

An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "td", "th"

Parse error. Ignore the token.

Anything else

Process the token as if the insertion mode was "in table".

When the steps above require the UA to clear the stack back to a table row context, it means that the UA must, while the current node is not a tr element or an html element, pop elements from the stack of open elements. If this causes any elements to be popped from the stack, then this is a parse error.

The current node being an html element after this process is an innerHTML case.

If the insertion mode is "in cell"

An end tag whose tag name is one of: "td", "th"

If the stack of open elements does not have an element in table scope with the same tag name as that of the token, then this is a parse error and the token must be ignored.

Otherwise:

Generate implied end tags, except for elements with the same tag name as the token.

Now, if the current node is not an element with the same tag name as the token, then this is a parse error.

Pop elements from this stack until an element with the same tag name as the token has been popped from the stack.

Switch the insertion mode to "in row". (The current node will be a tr element at this point.)

A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "td", "tfoot", "th", "thead", "tr"

If the stack of open elements does not have a td or th element in table scope, then this is a parse error; ignore the token. (innerHTML case)

Otherwise, close the cell (see below) and reprocess the current token.

An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html"

Parse error. Ignore the token.

An end tag whose tag name is one of: "table", "tbody", "tfoot", "thead", "tr"

If the stack of open elements does not have an element in table scope with the same tag name as that of the token (which can only happen for "tbody", "tfoot" and "thead", or, in the innerHTML case), then this is a parse error and the token must be ignored.

Otherwise, close the cell (see below) and reprocess the current token.

Anything else

Process the token as if the insertion mode was "in body".

Where the steps above say to close the cell, they mean to follow the following algorithm:

If the stack of open elements has a td element in table scope, then act as if an end tag token with the tag name "td" had been seen.
Otherwise, the stack of open elements will have a th element in table scope; act as if an end tag token with the tag name "th" had been seen.

The stack of open elements cannot have both a td and a th element in table scope at the same time, nor can it have neither when the insertion mode is "in cell".

If the insertion mode is "in select"

Handle the token as follows:

A character token

Append the token's character to the current node.

A comment token

Append a Comment node to the current node with the data attribute set to the data given in the comment token.

A start tag token whose tag name is "option"

If the current node is an option element, act as if an end tag with the tag name "option" had been seen.

Insert an HTML element for the token.

A start tag token whose tag name is "optgroup"

If the current node is an option element, act as if an end tag with the tag name "option" had been seen.

If the current node is an optgroup element, act as if an end tag with the tag name "optgroup" had been seen.

Insert an HTML element for the token.

An end tag token whose tag name is "optgroup"

First, if the current node is an option element, and the node immediately before it in the stack of open elements is an optgroup element, then act as if an end tag with the tag name "option" had been seen.

If the current node is an optgroup element, then pop that node from the stack of open elements. Otherwise, this is a parse error, ignore the token.

An end tag token whose tag name is "option"

If the current node is an option element, then pop that node from the stack of open elements. Otherwise, this is a parse error, ignore the token.

An end tag whose tag name is "select"

If the stack of open elements does not have an element in table scope with the same tag name as the token, this is a parse error. Ignore the token. (innerHTML case)

Otherwise:

Pop elements from the stack of open elements until a select element has been popped from the stack.

Reset the insertion mode appropriately.

A start tag whose tag name is "select"

Parse error. Act as if the token had been an end tag with the tag name "select" instead.

An end tag whose tag name is one of: "caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"

Parse error.

If the stack of open elements has an element in table scope with the same tag name as that of the token, then act as if an end tag with the tag name "select" had been seen, and reprocess the token. Otherwise, ignore the token.

Anything else

Parse error. Ignore the token.

If the insertion mode is "after body"

Handle the token as follows:

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE

Process the token as it would be processed if the insertion mode was "in body".

A comment token

Append a Comment node to the first element in the stack of open elements (the html element), with the data attribute set to the data given in the comment token.

An end tag with the tag name "html"

If the parser was originally created in order to handle the setting of an element's innerHTML attribute, this is a parse error; ignore the token. (The element will be an html element in this case.) (innerHTML case)

Otherwise, switch to the trailing end phase.

Anything else

Parse error. Set the insertion mode to "in body" and reprocess the token.

If the insertion mode is "in frameset"

Handle the token as follows:

A character token that is one of one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF), U+000B LINE TABULATION, U+000C FORM FEED (FF), or U+0020 SPACE