A user account is required in order to edit this wiki, but we've had to disable public user registrations due to spam.

To request an account, ask an autoconfirmed user on Chat (such as one of these permanent autoconfirmed members).

MetaExtensions

From WHATWG Wiki
Revision as of 09:26, 23 June 2011 by Hsivonen (talk | contribs) (Informatively fill in a basic DC spec hole about date formats)
Jump to navigation Jump to search

This page lists the allowed extension values for the name="" attribute of the <meta> element in HTML5. You may add your own values to this list, which makes them legal HTML5 metadata names. We ask that you try to avoid redundancy; if someone has already defined a name that does roughly what you want, please reuse it. Also, please be sure to include all the items required by the spec including a link to a specification that specifies the keyword as an HTML meta keyword. Note that URL-valued properties must not be registered as meta names but should be registered as rel keywords instead. Also note that changes to this registry may not be reflected in validators in real time.

Registered Extensions

Keyword Brief description Link to specification Synonyms Status
apple-mobile-web-app-capable Sets whether a web application runs in full-screen mode. Apple Safari HTML Reference Could maybe be assumed when application-name is set? Proposal
apple-mobile-web-app-status-bar-style Sets the style of the status bar for a web application. Apple Safari HTML Reference Proposal
baiduspider Synonym of robots for targeting Baidu only. Baidu documentation Proposal
format-detection Enables or disables automatic detection of possible phone numbers in a webpage in Safari on iOS. Apple Safari HTML Reference Proposal
googlebot Synonym of robots for targeting Googlebot only. Google documentation Proposal
google-site-verification Used to verify ownership for Webmaster Tools. Google documentation Proposal
robots A comma-separated list of operators explaining how search engine crawlers should treat the content. Possible values are "noarchive" to prevent cached versions, "noindex" to prevent indexing, and "nofollow" works as the link rel value with the same name. This meta name is already supported by every popular search engine.
The content value "NOODP" has been offered elsewhere, so I'm proposing it here. It blocks robots from using Open Directory Project descriptions of a website instead of Web pages' own meta descriptions. It may have been introduced by Microsoft.
The content value "NOYDIR" has been offered by Yahoo, so I'm proposing it here. It blocks Yahoo's robot from using the Yahoo directory's descriptions of a website instead of Web pages' own meta descriptions. Whether any other robot supports this is unknown but possibly no other search engine uses Yahoo's directory anyway.
Robots exclusion protocol, NOODP value: Google, Yahoo, NOYDIR value: Yahoo, as accessed 4-28-09 Proposal
slurp Synonym of robots for targeting Yahoo! only. Yahoo! documentation Proposal
teoma Synonym of robots for targeting Teoma and Ask.com only. Ask.com documentation Proposal
viewport Provides a way for documents to specify (using markup rather than CSS) the size, zoom factor, and orientation of the viewport that is used as the base for the document's initial containing block. The following properties can be used in the value of the content attribute:
  • width
  • height
  • initial-scale
  • minimum-scale
  • maximum-scale
  • user-scalable

Examples:

<meta name="viewport" content="initial-scale=1.0">
<meta name="viewport" content="width=480, initial-scale=2.0, user-scalable=1">

For more details, see the Viewport META element section of the CSS Device Adaptation draft specification.

For meta elements that have a name attribute whose value is viewport, the CSS Device Adaptation draft specification defines the recognized properties for the content attribute>, as well as an algorithm for parsing the value of that attribute. Proposal
rating The Restricted to Adults label (RTA) provides a way for adult oriented websites to indicate that their content is off limits to children. RTA was introduced in 2006 and is currently used by a large number of adult web content providers. RTA is recognized by all major parental control filters.

Example:

<meta name="RATING" content="RTA-5042-1996-1400-1577-RTA" />
RTA documentation Proposal
csrf-param Cross-site request forgery protection parameter for Ruby on Rails Rails API Proposal
csrf-token Cross-site request forgery protection token for Ruby on Rails Rails API Proposal
verify-v1 Legacy verification for Google Sitemaps. Inside Google Sitemaps: More about meta tag verification Proposal
msvalidate.01 One of the verification elements used by Bing. Bing Webmaster Tools Proposal
dcterms.description An account of the resource. Description may include but is not limited to: an abstract, a table of contents, a graphical representation, or a free-text account of the resource. DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal
dcterms.creator An entity primarily responsible for making the resource. Examples of a Creator include a person, an organization, or a service. DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal
dcterms.issued Date of formal issuance (e.g., publication) of the resource. (DC doesn't spec a date format but the established practice is YYYY-MM-DD.) DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal
dcterms.language A language of the resource. Recommended best practice is to use a controlled vocabulary such as RFC 4646 [RFC4646]. DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Redundant with the lang attribute on the html element. (Browsers pay attention to the lang attribute but not dcterms.language) Proposal
dcterms.modified Date on which the resource was changed. (DC doesn't spec a date format but the established practice is YYYY-MM-DD.) DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal
dcterms.subject The topic of the resource. DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal
dcterms.title A name given to the resource. DCMI Metadata Terms mapped according to Expressing Dublin Core metadata using HTML/XHTML meta and link elements Proposal

Proposals that don't meet the requirements for a registration

Note that these proposals can be moved back to the registry table if the problems listed in the rightmost column of this table are addressed.

Keyword Brief description Link to specification Synonyms Status Registration requirement failure
audience To aid search engines in classifying and to aid directory compilers, an audience most appropriate for the page may be suggested. Subject matter may not be a good clue; for example, an analysis of children's literature may be directed to teachers.
A value is free-form case-insensitive text without a comma. Multiple values are to be comma-separated. Singular and plural forms have the same meaning.
Recognized values:
-- "all" and "everyone", which have the same meaning
-- "adult" and "mature" have the same meaning and are for content that only adults may access, but no one responsible for preventing a nonadult or the immature from accessing the page or its content should rely on either or both of these values to do so without other means (not the same as "grownup", which see)
-- "child" and "juvenile", which have the same meaning
-- "teen"
-- "grownup" is not identical to "adult" or "mature" in not implying a precise boundary but is approximately any person who may be able to understand and apply the content (e.g., car driving instruction that may be read by a minor not yet old enough to drive a car but who would likely benefit from somewhat early exposure to the instruction)
-- "parent" to include guardian and temporary caregiver
-- "teacher" to include professor and ad hoc instructor
-- "elementary school student" to include any student below high school
-- "high school student"
-- "elhi" to include any student in elementary school through high school
-- "college student" including graduate and professional school
-- "business" including management, finance, and prospective customers (this includes e-commerce and investor sites)
-- "health" including any health care provider including alternative and ad hoc
-- "patient" for any health care recipient
-- "lawyer" including judge, paralegal, and jailhouse lawyer
-- "law client" for any prospective recipient of a lawyer's service (not usually a social work client) with lawyer including paralegal and jailhouse lawyer but not necessarily judge
-- "craft" for any craftworker including laborer and artisan
-- "artist" including musician, actor, dancer, and sculptor and including creator and performer
-- "military" including paramilitary
-- "news" including any consumer of rapidly-developing news
-- "introductory" and "beginner", which have the same meaning
-- "intermediate" and "midlevel", which have the same meaning
-- "advanced" and "advance", which have the same meaning
-- "scholarly" and "scholar", which have the same meaning
-- "popular" generally referring to a writing style
-- "older" including retiree
-- "institution" including from corporation to conspiracy (such as for management advice)
-- "government" including agencies and prospective politicians
-- values using any integer or single-digit decimal in the form of "grade 8" or "grade 6.4" including to refer to a reading comprehension level (this generally will not exceed 12 and might be meaningless above 20 so higher values may be interpreted as the highest meaningful value)
-- "viewers" for when content (such as a movie) is intended almost entirely to be seen rather than read
-- "listeners" for when content (such as music) is intended almost entirely to be heard rather than read but not generally including text-to-speech support
-- "tts", "text-to-speech", or "text to speech", which three have the same meaning and which are for a page that has substantial support for TTS or that will be readily understood through TTS without need for such support (TTS is often aided by, e.g., pre-resolving pronunciation ambiguities in page coding)
-- values using any numbers in the form of "3-6 years old", whether a range or a single-number value
-- values using any decade in the form of "born in 1970s"
Unrecognized values such as "botanists", "Texans", and "writers who use red ink" may be used but at a risk that a search engine or directory editor will either fail to recognize it or will interpret it in unpredictable ways, or will in the future.
Spellings that are erroneous or slightly different from a recognized value may be interpreted by a search engine or directory editor as representing a recognized value.
The absence of the keyword defaults to a value of "all" but without overriding another indication arrived at by other means.
Incomplete proposal Lacks link to a spec
bot-. . . Robot owners, to allow page authors access to robotic capabilities, e.g., to deny them, should prefix "bot-" to the name of their robot, especially for proprietary bots.
Example: If a robot were to be named "dullbucklequiz", the name in the meta element would be "bot-dullbucklequiz".
The value "bot-" alone represents all bots so prefixed, like a wildcard.
Arguably, there's no need for a list here of any specific bots if http://user-agents.org or http://www.botsvsbrowsers.com/ (and perhaps other sites) is reliable.
Incomplete proposal Lacks link to a spec, tries to register a space of names instead of enumerated names
created The datetime at which the document was created. The value is an ISO8601 date. The date MUST follow the W3C Profile of ISO 8601 with a granularity of "Complete date:" or finer. The BBC use this name. Incomplete proposal Lacks link to a spec
creator The creator is an off-Web or pre-Web creator of a work for which an author authored a Web page, so that the creator and the author may be different people.
Searching for one content creator's work requires a standard robot-parsable format for the information. A personal name, institutional name, or other text entry is permissible.
One element represents only one creator. Multiple creators are to be represented with multiple tags.
Search engines may index by any component of a name, so a content creator need only enter a name once in one first-last or family-given order (e.g., Pat Thunderbird or Thunderbird, Pat, but not requiring both).
Talk Incomplete proposal Lacks link to a spec


msnbot Synonym of robots for targeting Bing only. Incomplete proposal Lacks link to a spec
ia_archive Synonym of robots for targeting Internet Archive and Alexa only. Incomplete proposal Lacks link to a spec
datetime-coverage The author may be the best expert on which time frame is most relevant to the content. Leaving that to search engine analysis may be too chancy without search engine optimization, which analysis is difficult to apply by algorithm to, e.g., historical papers that may focus on the 1800s but mention 1731 and 1912 perhaps unimportantly.
The value for this keyword is a date or time -- not a range and not vague, for which other keywords are proposed -- in a format in accordance with http://www.w3.org/TR/NOTE-datetime (albeit a note that's at W3C only for discussion). Any of the six levels of granularity in that note are acceptable, such as expressing only a year.
Should this keyword appear more than once, all the values so appearing are determinative. Multiple values are to be expressed with separate meta elements lest the note be revised in the future in a way incompatible with comma-separating a list.
Incomplete proposal Lacks link to a spec
datetime-coverage-end This is identical to the keyword datetime-coverage except that it represents only the end. If this keyword is used without datetime-coverage-start (also proposed), its value is interpreted as ending a range without a start.
Should this keyword appear more than once, all the values so appearing are determinative, in which case each represents the end of a different range assumed to be nonnesting. Example: If four elements happen to be in the order of datetime-coverage-end=1865, datetime-coverage-start=1914, datetime-coverage-end=1918, and datetime-coverage-start=1862, assuming proper formatting, the ranges should be interpreted as 1862-1865 and 1914-1918.
Incomplete proposal Lacks link to a spec
datetime-coverage-start This is identical to the keyword datetime-coverage except that it represents only the start. If this keyword is used without datetime-coverage-end (also proposed), its value is interpreted as starting a range without an end.
Should this keyword appear more than once, all the values so appearing are determinative, in which case each represents the start of a different range assumed to be nonnesting. Example: If four elements happen to be in the order of datetime-coverage-start=1862, datetime-coverage-start=1914, datetime-coverage-end=1865, and datetime-coverage-end=1918, assuming proper formatting, the ranges should be interpreted as 1862-1865 and 1914-1918.
Incomplete proposal Lacks link to a spec
datetime-coverage-vague This is identical to the keyword datetime-coverage except that its value is not necessarily crisp. This keyword should be used only when datetime-coverage, datetime-coverage-start, and datetime-coverage-end are inappropriate, but there's no ban on using all four. Any text without a comma can be the value (e.g., Pleistocene, 1820s, Tuesdays, or before we were born); multiple values are comma-separated.
If this keyword is used with datetime-coverage, datetime-coverage-start, or datetime-coverage-end, the vague value should be exploited along with the value/s for the other keyword/s.
Should this keyword appear more than once, all are determinative.
Incomplete proposal Lacks link to a spec
DC. Dublin Core, maintained by Dublin Core MetaData Initiative (DCMI), is an extensive system with some overlap with non-DC names.
This reserves all strings that begin with DC and a dot. Not true; DC-HTML doesn't use hardwired prefixes, but defines the prefixes using link/@rel="scheme.prefix"
DCMI Incomplete proposal Tries to register a space of names instead of enumerated names
dir-content-pointer When several pages in a directory include main content, a table of contents, an index, and the like, a search engine may be able to organize results more usefully by identifying which is which with a standard vocabulary, helpful when different publishers use different conventions when displaying or printing content.
A value is free-form case-insensitive text without a comma and optionally with a trailing number. Multiple values are to be comma-separated (multiple values are appropriate when one document serves multiple purposes). Singular and plural forms have the same meaning.
Recognized values, which are pointer types to which numbers may be suffixed, are limited to "start" meaning 'the first page that should be seen by a user' (this may be anywhere in the directory and anywhere within content), "toc" meaning 'table of contents', "intro" including introductions, forewords, prefaces, and tables of figures, "abstract", "main", "bibliography" and "biblio", which have the same meaning, "index" which may mean 'sitemap' or not, "afterword" and "update" which have the same meaning and need not actually update, "credit" meaning 'credits and acknowledgments', and "author bio" meaning 'author's biography', including any information about the author including credentials and contact information. The number suffix may be spaceless or not.
When numbers are suffixed, a search engine or directory should arrange like items in numerical order in the results, with unnumbered items following like items that are numbered, e.g., intro 1, intro 2, main 1, main 2, main, main, and so on.
Each directory and each subdirectory has its own sequence.
Incomplete proposal Lacks link to a spec
expires meta name='expires' defines the expiration date of the page. This can be used for web pages in preparation for an upcoming event, e.g. a registration form for an exposition or competition, or other cases with a pre-set date when the document will no longer be valid, e.g. a product offer in a special sale or a support page for a product known not to be supported anymore from a given time onward.

Search engines should respond to this meta tag in a reasonable way, i.e. by removing the page from their main search results after the expiration date (possibly still returning the result in a special search for expired pages as long as the page exists and is not explicitly excluded via robots.txt or meta name='robots' etc.) or simply by indicating to the user that this result is out-of-date.

The content attribute should define the expiration date in accordance with http://www.w3.org/TR/NOTE-datetime . The meta tag should not be used for pages without expiration date. However, for historical reasons, search engines should also interpret other date formats where possible and should be prepared to find values such as "", "0", "no" and "never". Such non-date values are to be interpreted as no expiration date.

Correctly formatted example:

<meta name='expires' content='2012-12-31T23:59Z'>

This tag is not to be confused with and has a different meaning than meta http-equiv='expires'.

Incomplete proposal Lacks link to a spec
format-print This is to allow a user agent to inform an operating system or a printer driver of the preferred print medium, such as the paper size.
A value is free-form case-insensitive text without a comma. Multiple values are to be comma-separated (multiple values might be appropriate because standard paper sizes vary around the world). Singular and plural forms have the same meaning.
Recognized values are limited to "letter", "A4", "legal", "A5", "B5", "monarch", "envelope 10" meaning size #10, "envelope 6-3-4" meaning size #6 3/4, values with integers and decimals in the form of "8.5 x 11" or "8.5x11" in which spacing of the "x" does not affect meaning, "paper", which means 'paper of the default color (usually white) and weight (usually 20-lb. stock)', "white", "yellow", "pink", "blue", "green", "violet", or "multicolor", which means a medium of the given color or mixed, "letterhead", "p2 letterhead" meaning 'letterhead intended for any page except the first', "watermark" meaning a 'special watermark such as an organization's own', and "plain" meaning 'not preprinted and not letterhead (it may have a paper manufacturer's watermark not related to letterhead)'.
Omitting "paper" when another recognized value is given defaults to an implied meaning of 'paper' with the other value; e.g., "letter" means 'letter paper'; the same principle applies to a medium's color (the default being white for paper and colorless for transparency) and plainness or lack thereof (the default being plain).
Other values should be proposed before being recognized here. Label sizes should be proposed here for labels that are not on backing sheets that fit one of the recognized values, e.g., labels on narrow rolls. Blueprint paper sizes should be proposed here. Media other than standard paper, such as onion skin, heavier paper, card, and clear or color transparency, should be proposed here.
The user agent may, with the user's or user sysadmin's permission (as by a menu-driven default), interpret a value to offer an alternative the user might accept and software and firmware other than the UA may interpret a value to the same end with or without permission, so this keyword is only suggestive; e.g., "letter" may be interpreted as "A4".
The absence of the keyword defaults to a value determined by other than the page, e.g., by the printer driver or the user agent.
Talk Incomplete proposal Lacks link to a spec
geographic-coverage The author may be the best expert on the geographic relevance of the content. Leaving that to search engine analysis may be too chancy without search engine optimization, which analysis is difficult to apply by algorithm to, e.g., historical papers and epidemiological studies which may mention locales only once.
Absence of the keyword defaults to a value of world (not universe), unless the search engine chooses to interpret the page or larger unit for some other value, probably based on other than just contact information given in the website.
The value for this keyword is a semicolon-separated list of one or more place-values, the order of which do not matter. One place-value will use commas to separate, in order, an optional standard natural language symbol applicable to the place-value (when omitted the language applicable to the page will control), a place-class, one or more place-subclasses if any, and one or more place name parts (where, e.g., in "Cape Town, South Africa", "Cape Town" is a place name part but "Town" is not). Spaces after semicolons and commas are optional; spaces within place-values are present when required for each place-value (e.g., "Quezon City", not an invented "QuezonCity").
To distinguish names that might otherwise be too similar, place-classes, all lower-case and hyphenatably spaceless, include outer-space, region (on Earth and crossing or larger than a nation, e.g., southern hemisphere, polar region, temperate zone, or Asia), intntl-water (an 'international water body'), intntl-agcy ('international agency' or 'international collection', e.g., all U.N. member nations), nation, within-nation (limited to only one political level down from nation, e.g., state, province, territory, possession, city not included within other political units of a nation, or any comparable unit), city (including town, village, hamlet, and any comparable political unit below the level of within-nation), addr (including address, full-length street, building, institution, and neighborhood without political boundaries), pol-unit (pol abbreviating 'political') (e.g., a place of disputed nationhood), hist-pol-unit (hist abbreviating 'historical') (e.g., the Roman Empire), feature (e.g., river), num (e.g., latitude and longitude or outer-space equivalent in numbers), and ethereal (including thealogical/theological, fictional including from modern popular entertainment, and ancient secular mythical, but not including that which is asserted to be a state of mind or existence but not a place, such as nirvana). (Example for one hypothetical page: name="geographic-coverage" content="region, sub-Saharan Africa; nation, Panama; city, Panama, Panama; within-nation, Sao Paulo, Brazil; city, Sao Paulo, Sao Paulo, Brazil; within-nation, Mississippi, United States of America; region, Middle East; region, Midwest, United States of America; hist-pol-unit, Northwest Territory, United States of America; feature, river, Indus; outer-space, Indus; ethereal, ultima Thule; ethereal, Heaven; ethereal, Flatland; ethereal, Valhalla; en-US, addr, Hotel Valhalla, Fredrikstad, Norway; es, nation, Espana" (Indus is both a river and a constellation, illustrating the need for place-classes)).
Ambiguity of place-values should be avoided despite convenience in coding because search engines may each interpret them as they see fit, e.g., it would be hard for an engine to distinguish New York from New York.
For consistency of spelling, several authority lists should be settled upon, with legal, well-known, and disputed names and common abbreviations all being acceptable; but I'm not proposing one here now (relying on IANA's ccTLD list might be too complex to implement and still assure coding consistency, e.g., occasionally ccTLDs can be phased out and off of IANA's list) (a standard vocabulary possibly usable here is the Getty Thesaurus of Geographic Names Online, subject to licensing and charset choice); and promulgating authority lists may best be done publicly by search engine managements, who may disagree with each other.
Allowing Unicode for non-Roman alphabet-using locales is desirable, but at present that may raise technical problems, including computer security issues, that are not yet readily soluble.
Incomplete proposal Lacks link to a spec
keywords-not A comma-separated list of negative keywords that distinguish a closely-related theme from this page's true theme, to support Boolean NOT searches often more realistically than visible text can, especially when both themes share the same lexicon.
If keywords is no longer a supported name for a meta element, keywords-not is superfluous; however, debate has been revived on whether keywords should be supported or not; see the keywords entry in this Wiki.
W3C Bug 6609 Incomplete proposal Lacks link to a spec
page-datetime Better ranking in search engine results for recency or relevance to an event date would be aided by a standard format robots can parse. Users would save search time by not having to load many pages to find which ones are new or date-relevant.
To supply a consistent and known format, the value for this keyword is a date-time expression formed in accordance with http://www.w3.org/TR/NOTE-datetime (albeit a note that's at W3C only for discussion). Any of the six levels of granularity in that note are acceptable, such as expressing only a year.
Should this keyword appear more than once, only the first one so appearing is determinative.
Incomplete proposal Lacks link to a spec
page-version Pages may be revised several times daily. While date-time given to a granularity of a fraction of a second would often suffice, when a page has to be approved more than once before posting, any or no such time may be correct (without this keyword, a comment could be necessary but probably not parsable by an engine). In addition, versions regardless of date may show consecutiveness and can replace a date that must be vague. In that case, a version number may be more useful for searches and so a robot-parsable format is needed.
The keyword's value is stated in ASCII digits, is any nonnegative base-10 rational number expressed as an integer or a decimal, with any number of decimal places allowed, and may be padded with any number of leading zeros to support extraction for ASCII sorting.
Should this keyword appear more than once, only the first one so appearing is determinative.
The versions 0 and 0.n, with n being to any number of places, signify beta versions, i.e., drafts, in the tradition of beta software, while versions 1 and higher ordinarily signify final-release versions. After a final-release version is released, a draft of a later version is not given a version number of 0 or 0.n, but is numbered higher than the last final-release version. It is suggested to page authors that draft status, if applicable, be shown in the visibly displayed text of the page, rather than that this meta tag be relied upon as the sole notice of draft status, as it may be inadequate notice if alone.
To assign a low page-version such as 0.n or 1, the page's URL, if static, may be used as the relevant premise. Thus, if a page is copied or moved to a new URL, the author may choose to restart page-version numbering from 0.n or 1. If a page's URL is dynamic, e.g., if created on the fly from a script, the page author may prefer to use as the relevant premise for assigning a low page-version such as 0.n or 1 the URL of the script or other technology that generates the dynamic-URL page, placing this meta element containing this attribute within the script or other technology, not within the generating page's head element (the generating page's head element may have its own meta element with this attribute describing the generating page). If one page containing the script or other technology that generates another page has more than one means for generating dynamic-URL pages, each means should contain its own meta element with this attribute. Page-version is thus largely independent of the page's date, although both would likely advance roughly in parallel.
Incomplete proposal Lacks link to a spec
publisher Searching for one content publisher's or page publisher's work requires a standard robot-parsable format for the information. This often differs from creator or author when the publisher is an institution. An institutional name, personal name, or other text entry is permissible.
One element represents only one publisher. Multiple publishers are to be represented with multiple tags, although multiple publishers are less common than multiple authors or creators; multiplicity is more likely for a legal name and a well-known name.
Search engines may index by any component of a name, so a publisher need only enter a name once in one order.
Incomplete proposal Lacks link to a spec
referrer Developers wish to control the sharing of referrer information with linked resources and followed links at the meta level. It's sometimes desirable to share referrer information from secure to non-secure resources. The proposal is to add:
<meta name="referrer" content="never|default|always">

Where the value of the content attribute is:

  • never: No referrer information is leaked when following a links or requesting resources.
  • default: Include referrer information in non-secure context or for HTTPS resources with the same origin.
  • always: Include referrer information when following all links and requesting any resources.

Since <meta name="referrer" content="never"> is redundant with a proposed <link rel="nofollow">, the <link rel="nofollow"> proposal should be supplanted with this one, which provides more robust control.

Incomplete proposal Lacks link to a spec
resolutions Authoring web sites to use resolution independent images that display beautifully on high-resolution displays should be made as easy as possible for developers and should not require JavaScript to accomplish.

To accomplish this, I propose a new HTML Meta Tag, resolutions, that can be used to specify that high-resolution versions of images linked to from the page are available and that the browser should use them in place of the lower-resolution default images if it detects that a user is using a high-resolution screen. The resolutions meta tag lists the device-pixel ratios supported by images in the page.

So, for example…

<meta name="resolutions" content="2x">

… means that the developer is telling the browser that she has created 2x resolution images for the images linked to from the current page and named them with a @2x suffix.

To illustrate, if her image tag is as follows…

<img src="/images/flower.jpg" alt="A flower">

… then she has two image files under /images: the low-resolution default (flower.jpg), and a higher-resolution (200%) version named [email protected].

(This is the same naming convention already used by Apple in its Cocoa Touch framework for automatically loading in higher-resolution versions of images.)

Based on the meta tag, if the browser detects that the user is running at a min-device-pixel-ratio of 2.0, it will automatically ask for the 2x version of the image ([email protected]) instead of the default image as specified in the image tag.

Finally, so as not to flood external sites with high-resolution image requests, this functionality would only work for local images specified via relative links.

Multiple resolutions

The resolutions tag can also contain a list of supported device-pixel ratios so as to support even higher-resolution displays when and if they become available in the future.

For example:

<meta name="resolutions" content="2x, 4x, 8x">

In this case, the developer would provide 2x, 4x, and 8x versions of all images. So, in the running example, she would make flower.jpg, [email protected], [email protected], and [email protected].

Advantages

The advantages of this approach are several:

  1. Makes it very simple for developers to support high-resolution displays like the iPhone 4's Retina screen
  2. Does not require JavaScript
  3. Does not change the default way that things work (if the meta tag is not specified, the browser simply behaves as it always has).
Proposal for native browser support of high-resolution image substitution

How to make your web content look stunning on the iPhone 4’s new Retina display

Incomplete proposal Lacks link to a spec
rights As a page effectively appears in at least two forms, usually one as interpreted and displayed on a device and the other as source code, arguably intellectual property rights that must be asserted must be asserted in ways understandable in both contexts. For example, &copy; is a raw representation that may legally fail as part of copyright notice to someone seeing source code and not the display, important when someone wants to copy source code for use elsewhere and may rely on a defense of innocent infringement (at least in U.S.). While such assertions can be made in a comment element, it may be helpful to have a tag that search engines can parse and index verbatim.
The value may include standard and nonstandard notices, invocations of licenses such as GFDL and ASCAP, and any other information. Content is defined as free-form, leaving the page author discretion for the entry.
Statements in one tag may discuss several portions of the page differently, e.g., with different licenses.
More than one license may be offered, along with the page's relationship to all.
Not all statements need be license grants. A statement may state whom to ask for reprint permission or may reserve all rights, for example.
Only one meta tag with this keyword may be present. Page authors must not use more than one. A UA finding multiple such tags on one page must ignore all of them.
The copyright symbol that would be generated by its character entity is not recommended for legal notice in source code when the word 'Copyright' may be used instead, because the entity may be read in raw form, but use is up to the page author. The same concept applies to any intellectual property rights symbol for which a suitable alternative is available, such as for trademark or service mark.
ASCII text would not suffice when a name or notice legally may have to be in a non-Roman alphabet, but no alternative may yet exist in HTML5.
Search engine storage may impose a length limit, but, because of legal consequences, if the value's length exceeds a given limit the search index should retain or interpret none of it but only refer to it.
The content string may only be copied verbatim in its full length, referred to, or ignored. It may not be, for example, paraphrased, truncated, interpreted, or classified except in addition to being copied verbatim in its full length.
Ignoring shall not void, nullify, or alter any rights stated in such tag.
For the synonymy, IP, IP-rights, and IP-right are not reserved; while the abbreviation IP 'intellectual property' is common among attorneys in the U.S., page authors will more likely be computerate, and the abbreviation may be wanted for 'Internet Protocol'.
Talk Incomplete proposal Lacks link to a spec
rights-standard The purpose is to enable search engines and other cataloging services to compile the types of rights allocated to the work.

Format:

<meta name="rights-standard" content="type;rights" />

The following are case-insensitive, except object ID.

  • type - item these rights apply to
    • (null) - everything; trailing semicolon is not included in this case
    • code - the HTML code (and embedded other?) of the page.
    • content - the entire content of the page
    • object - a specific object, (for example, an image).
      • format is "object;" followed by the object ID; e.g., object;left_middle-column
  • rights - what rights are assigned to the item
    • "coprYYYY" -- work is copyrighted as of YYYY year
    • "CC:" - Creative Commons; followed by any number (in any order) of additional attributes:
      • "BY" - By Attribution
      • "SA" - Share Alike
      • "ND" - No Derivatives
      • "NC" - No Commercial
    • "PD" - Public domain

Examples:

  • Page content is CC By Attribution
 <meta name='rights-standard' content='content;cc:by' />
  • Page code is CC By Attribution, Share alike
 <meta name='rights-standard' content='code;cc:bysa' />
  • Object with ID "item_1" is CC By Attribution, no derivatives, no commercial; object with ID "item_2" is public domain.
 <meta name='rights-standard' content='object;item_1;cc:byndnc' />
 <meta name='rights-standard' content='object;item_2;pd' />
  • Everything on the page is copyrighted
 <meta name='rights-standard' content='copr2009' />
Talk:MetaExtensions#rights:_why_reversion

Creative Commons types and attributes

Incomplete proposal Lacks link to a spec
subj-. . . To classify by subject a page's content, a standard subject taxonomy that will be recognized by a search engine or directory will help. Because many such high-quality taxonomies exist, only a prefix is proposed. Over time, particular taxonomies, in print or online, may be recognized here and keywords assigned for each.
The keyword will be constructed case-insensitively with subkeywords in the form subj-[nationAbbrev]-[taxonomy]-[edition][-optionalSubedition], e.g., subj-US-MeSH-2009online (perhaps). After "subj-", the second subkeyword will identify the nation where the taxonomy is published or offered as an aid in identifying the taxonomy and does not limit the subject coverage; e.g., a taxonomy published in Japan may be ideal for classifying Canadian botany or Peruvian economy.
As subject values may vary between editions of one taxonomy, an edition and optionally a subedition is to be identified in the third and optionally the fourth subkeywords. The subedition, if any, is any update or revision occurring between editions, such that a value drawn from that edition and subedition is stable. The means of identifying edition and subedition should be included in the registration of a keyword.
Examples of taxonomies from the U.S. include MeSH (medical) and the Library of Congress Subject Headings.
The value identifying a subject for a Web page will be drawn from the cited taxonomy's edition and subedition.
If the value should have a style to prevent ambiguity in interpretation, that style is to be registered here for that keyword. Multiple values are expressed with multiple meta elements, one value for each, since comma-separation is probably not compatible with all taxonomies.
If a value requires case-sensitivity to prevent confusion, the entry here registering the keyword must accommodate that need with the needs of HTML 5 with an appropriate rule. To that end, a proposal to allow case-sensitivity in meta tags under some circumstances has been offered in the W3C bug reporting system.
W3C Bug 6854 subject-. . . Incomplete proposal Lacks link to a spec, tries to register a space of names instead of enumerated names
MSSmartTagsPreventParsing Microsoft introduced into Internet Explorer 6 Beta a feature that some website designers wished to preclude from applying in order to prevent public misunderstanding of their websites. The feature allowed a browser to add information but at a risk that users wouldn't know that it wasn't supplied by the website. This keyword was provided by Microsoft for those of us who wanted it.
Its value was "TRUE". Microsoft spelled the keyword with some capitals and the value in all capitals but whether capitalization was required for either is unknown; some opinions vary. Since it need be understood by only one browser, and that one a beta version, full standards compliance should not be assumed, and original case may be required. (This tag is used by Google: "<meta content='true' name='MSSmartTagsPreventParsing'/>" appeared (with internal quote marks as singles) in the source code for <http://googleblog.blogspot.com/2009/04/listening-to-google-health-users.html>, as accessed 4-27-09.)
Microsoft has apparently removed this instruction from its website on the ground that the beta version is no longer available and is not supported, but that doesn't assure that some users aren't still using the beta browser, perhaps inadvertently. Therefore, designers may wish to continue using the keyword and value and they are preserved here.
e.g., The Register (U.K.), Univ. Oregon (U.S.) (PDF p. 18), & John Chambers (U.S.) (job résumé near root), all as accessed 4-19-09 Incomplete proposal Lacks spec, potentially never minted by MS as a meta name (as opposed to a http-equiv value), even if minted by Microsoft, abandoned before shipping in any final release of IE
yandex-verification Used to verify ownership for Yandex Webmaster. Yandex Webmaster ownership verification Incomplete proposal The alleged spec URL doesn't actually mention the meta keyword.

Failed Proposals

Keyword Brief description Link to more details Synonyms Status
cache This doesn't actually work; use HTTP headers instead.
Value must be "public", "private", or "no-cache". Intended as a simple way to tell user agents whether to store a copy of the document or not. An alternate for HTTP/1.1's cache-control; for publishers without access to modifying cache-control.
none Unendorsed
no-email-collection HTML5 prohibits URL-valued meta names. They should be rel keywords instead.
Intended to reference legal policy of web site indicating that harvesting of e-mail addresses on the site is not permitted and in violation of applicable laws such as the CAN-SPAM Act of 2003.
Project Honey Pot Unendorsed

Process

For the "Status" section to be changed to "Ratified", the proposed keyword must be defined by a W3C specification in the Candidate Recommendation or Recommendation state. If it fails to go through this process, it is "Unendorsed".

For more details, see the HTML5 specification.