See also the WHATWG FAQ.
HTML is the core foundational standard being worked on by the WHATWG community. It is continuously maintained and supersedes HTML4, XHTML1, DOM Level 2 HTML, and all previous HTML specifications — addressing many of the shortcomings of those specifications while at the same time enhancing HTML to more adequately cover the needs of web applications. Along with defining the HTML markup language, it also defines many of the core requirements that form the basis of the web runtime.
Going forward, the WHATWG is just working on "HTML", without worrying about version numbers. When people talk about "HTML5" in the context of the WHATWG, they usually mean just "the latest work on HTML", not necessarily a specific version. For more details, see the section called "Is this HTML5?" in the standard.
Use a validator.
The whole standard is more or less stable. There are some parts of it that describe new technologies that have not yet been implemented everywhere, but at this point those additions are only added after the design itself is pretty stable. Such additions must also have the support of two or more implementers, per our working mode.
In practice, implementations all follow the latest standard anyway, not so-called "finished" snapshots. The problem with following a snapshot is that you end up following something that is known to be wrong. That's obviously not the way to get interoperability!
This has in fact been a real problem at the W3C, where mistakes are found and fixed in the editors' drafts of specifications, but implementers who aren't fully engaged in the process go and implement obsolete snapshots instead, including those bugs. This has resulted in serious differences between browsers.
For more information on this, see the WHATWG FAQ entry What does "Living Standard" mean?.
Browsers do not implement HTML+, HTML2, HTML3.2, HTML4, HTML4.01, etc, as separate versions. They all just have a single implementation that covers all these versions at once. That is what the HTML Standard defines: how to write a browser (or other implementation) that handles all previous versions of HTML, as well as all the latest features.
One of the main goals of the HTML Standard and the WHATWG effort as a whole is to make it possible for archeologists hundreds of years from now to write a browser and view HTML content, regardless of when it was written. Making sure that we handle all documents is one of our most important goals. Not having versions does not preclude this; indeed it makes it significantly easier.
It shouldn't matter if and when old pages become invalid.
Validity (more often referred to as document conformance in the WHATWG) is a quality assurance tool to help authors avoid mistakes. We don't make things non-conforming (invalid) for the sake of it, we use conformance as a guide for developers to help them avoid bad practices or mistakes (like typos). So there's not really any need to worry about whether old pages are conforming or not. It's only helpful when you're writing a new page, and it's always most helpful to have the latest advice. It wouldn't be useful to check for conformance against last week's rules, for instance. After all, we fixed mistakes in those rules this week! For more details, see part of the introduction of the standard.
There are a number of ways to track changes to the standard:
- The Twitter feed: @htmlstandard
- The GitHub commits log
- The standard is available in the Git repository. You may use any Git client to check out the latest version and use your client's diff tools to compare revisions and see what has been changed.
- At a broader level, Anne and Simon once wrote a document that gave a high-level overview of changes to HTML over the last decade or so: https://html-differences.whatwg.org/
The HTML Standard is available in three forms: single-page (very large), multi-page, and the developer's edition.
The W3C publishes some forked versions of the HTML Standard, and of other WHATWG standards. We have requested that they stop publishing these but they have refused. They copy most of our fixes into their forks, but their forks are usually weeks to months behind. They also make intentional changes, and sometimes even unintentional changes, to their versions. We highly recommend not paying any attention to the W3C forks of WHATWG standards.
Here are some sites to help you work out what you can use:
The following sites also have some useful information:
If you know of any more (or if you have some yourself) then send a pull request to add them to the list! (Or, if you think any of the above have lost usefulness over time, send a pull request removing them and outlining your reasoning.)
The WHATWG is now using a Living Standard development model, so this question is no longer really pertinent. See above, under "What is HTML5?". The real question is, when can you use new features? For an answer to that question, see "How do I know if a particular feature in the spec is ready to use?".
Back before the Living Standard development model, we were planning to put the contents of the HTML Standard through the W3C process. This was before we understood the fatal flaws of such a snapshot-based development mode.
At the time, the W3C Recommendation label had high standards, such as 100% test coverage of two complete and fully interoperable implementations. In 2008, the editor estimated it would take another 14 years to reach that point, based on comparing it to the amount of work done for HTML4 and other large specifications like CSS2/2.1.
Since then, we've realized that much like the waterfall model is not a good fit for software development, it is also not a good way of developing standards. These days we keep the HTML Standard continually under development, adding tests as we go and verifying them against implementations, per our working mode. So, the 2022 date is no longer relevant.
Sort of. Often some record of the rationale for a particular design choice can be found within discussions in the GitHub issue tracker, commit logs, or the mailing-list archive or IRC channel archives. Sometimes, there is an explanation in the specification, but doing that everywhere would make the specification huge.
For a few cases that someone did take the time document, the information can be found at the following locations:
- Rationale — a page that documents some reasons behind decisions in the spec, originally written and maintained by Variable. If anyone wants to help him out, try to grab someone on IRC; we're always looking for more contributors and this is a good place to start.
- Why no namespaces
- Why no script implements
- Why not reuse legend or another mini-header element.
Also see HTML feature proposals.
Yes. Unlike HTML4 and XHTML1, the choice of HTML or "XHTML" is solely dependent upon the choice of the media type, rather than the DOCTYPE. See HTML vs. XHTML
In text/html documents:
<!DOCTYPE html>In documents delivered with an XML media type: no DOCTYPE is required and its use is generally unnecessary. However, you may use one if you want (see the following question). Note that the above is well-formed XML.
For compatibility with legacy producers designed for outputting HTML, but which are unable to easily output the above DOCTYPE, this alternative legacy-compat version may be used instead.
<!DOCTYPE html SYSTEM "about:legacy-compat">Note that this is not intended for dealing with any compatibility issues with legacy browsers. It is meant for legacy authoring tools only.
Excluding the string "about:legacy-compat", the DOCTYPE is case insensitive in text/html. In documents delivered with an XML media type, it is case sensitive and must be either of the two variants given above. For this reason, the DOCTYPEs given above are recommended to be used over other case variants, such as <!DOCTYPE HTML> or <!doctype html>.
These alternatives were chosen because they meet the following criteria:
- They trigger standards mode in all current and all relevant legacy browsers.
- They are well-formed in XML.
- It is possible to output at least one of the alternatives, if not both, with extant markup generators.
- They intentionally contain no language version identifier so the DOCTYPE will remain usable for all future revisions of HTML.
- The first is short and memorable to encourage its use.
- The legacy-compat DOCTYPE is intentionally unattractive and self descriptive of purpose to discourage unnecessary use.
Generally, the use of a DOCTYPE in an document delivered with an XML media type is unnecessary. However, there are cases where inclusion of a DOCTYPE is a reasonable thing to do:
- The document is intended to be a polyglot document such that the same text may be treated as either HTML or XML.
- You wish to declare entity references for use within the document. Note that most browsers only read the internal subset and do not retrieve external entities. (This is not compatible with HTML, and thus not suitable for polyglot documents.)
- You wish to use a custom DTD for DTD-based validation. But take note of what's wrong with DTDs.
Fundamentally, this is an XML issue, and is not specific to HTML documents delivered with an XML media type.
All documents with a text/html media type (that is, including those without or with an HTML 2.0, HTML 3.2, HTML4, or XHTML1 DOCTYPE) will be parsed using the same parser algorithm as defined by the HTML spec. This matches what web browsers have done for HTML documents so far and keeps code complexity down. That in turn is good for security, maintainability, and in general keeping the amount of bugs down. The HTML syntax as now defined therefore does not require a per-version parser, and documents with an HTML4 DOCTYPE for example will be parsed as described by the HTML specification.
Validators are allowed to have different code paths for previous levels of HTML.
With an HTML validator that follows the latest specification.
The HTML serialization refers to the syntax of an HTML document defined in the HTML specification. The syntax is inspired by the SGML syntax from earlier versions of HTML, bits of XML (e.g. allowing a trailing slash on void elements, xmlns attributes), and reality of deployed content on the web.
Any document whose media type is determined to be text/html is considered to be an HTML serialization and must be parsed using an HTML parser.
The XML serialization refers to the syntax defined by XML 1.0 and Namespaces in XML 1.0. A resource that has an XML media type, such as application/xhtml+xml or application/xml, is an XML document. XML documents whose root element is <html> in the HTML namespace are sometimes referred to as "XHTML" documents.
The HTML serialization must be served using the text/html media type.
The XML serialization must be served using an XML media type, such as application/xhtml+xml or application/xml. Unlike the situation as of XHTML1, the HTML specification requires that "XHTML" documents not be served with the text/html media type.
Using the incorrect media type (text/html) for a document in the XML serialization will cause the document to be parsed according to parsing requirements for HTML. In other words, it will be treated as what's sometimes called "tag soup". Ensuring the use of an XML media type is the only way to ensure that browsers handle the document as XML.
Void elements in HTML (e.g. the <br>, <img> and <input> elements) do not require a trailing slash. e.g. Instead of writing <br />, you only need to write <br>. This is the same as in HTML4. However, due to the widespread attempts to use XHTML1, there are a significant number of pages using the trailing slash. Because of this, the trailing slash syntax has been permitted on void elements in HTML in order to ease migration from XHTML1 back to HTML.
The current HTML specification also introduces the ability to embed MathML elements. On elements inside a math element, the trailing slash works just like it does in XML; that is, it closes the element. This is only inside that context however; it does not work for normal HTML elements.
You have to be really careful for this to work, and it's almost certainly not worth it. You'd be better off just using an HTML-to-XML parser. That way you can just use HTML normally while still using XML pipeline tools.
HTML vs. XHTML has some related guidance.
In the XML syntax, you are required to specify the namespace:
<html xmlns="http://www.w3.org/1999/xhtml">In text/html documents, the xmlns attribute is currently allowed on any HTML element, but only if it has the value http://www.w3.org/1999/xhtml. It doesn't do anything at all; it is merely allowed for the purpose of easing migration from XHTML1. It is not actually a namespace declaration in HTML, because HTML doesn't support namespaces. See the question "What about namespaces in HTML?".
HTML is defined in terms of the DOM and during parsing of a text/html document, all HTML elements are automatically put in the HTML namespace, http://www.w3.org/1999/xhtml. However, unlike the XML serialization, there is no real namespace syntax available in the HTML serialization (see previous question). In other words, you do not need to declare the namespace in your HTML markup, as you do in XHTML.
In addition, the HTML syntax provides for a way to embed elements from MathML and SVG. Elements placed inside the container element math or svg will automatically be put in the MathML namespace or the SVG namespace, respectively, by the parser. Namespace syntax is not required, but again an xmlns attribute is allowed if its value is the right namespace.
In conclusion, while HTML does not allow the XML namespace syntax, there is a way to embed MathML and SVG and the xmlns attribute can be used on any element under the given constraints, in a way that is reasonably compatible on the DOM level.
Regardless of whether documents are delivered as text/html or with an XML media type, UTF-8 is the only conformant character encoding.
For HTML, it is strongly recommended that you specify the encoding using the HTTP Content-Type header. If you are unable to configure your server to send the correct headers, then you may use the <meta> element:
<meta charset="UTF-8">In addition, the following restrictions apply:
- The character encoding name given must be the name of the character encoding used to serialize the file.
- The character encoding declaration must be serialized without the use of character references or character escapes of any kind.
- The
<meta>element used for this purpose must occur within the first 512 bytes of the file. It is considered good practice for this to be the first child of the<head>element so that it is as close to the beginning of the file as possible.
Note that this <meta> element is different from HTML4, though it is compatible with many browsers because of the way encoding detection has been implemented.
To ease transition from HTML4 to the current HTML specification, although the former is the recommended syntax, you may also use the following. (This does not apply to documents in the XML syntax):
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">In documents delivered with an XML media type, XML rules for determining the character encoding declaration apply. The <meta> element is never used for determining the encoding of such documents; for those, you should use either the HTTP Content-Type header or the XML declaration to specify the encoding.
<?xml version="1.0" encoding="UTF-8"?>However, because documents which do not have an XML declaration that explicitly specifies an encoding are processed as UTF-8, it's not necessary for conforming documents to include an XML declaration to specify the encoding.
See the list of differences between HTML and XHTML in the wiki.
Though the intent is that documents delivered with a text/html media type and that documents delivered with an XML media type can both produce identical DOMs, there still are some differences between working with an HTML DOM and an XML DOM.
Case sensitivity:
- Whenever possible, avoid testing
element.tagNameandnode.nodeName(or dotoLowerCase()before testing).
Namespaces:
- Use the namespace-aware version for creating elements:
document.createElementNS(ns, elementName).
Actually it doesn't. This is a misconception that comes from the confusion between conformance requirements for documents, and the requirements for user agents.
Due to the fundamental design principle of supporting existing content, the spec must define how to handle all HTML, regardless of whether documents are conforming or not. Therefore, the spec defines precisely how to handle and recover from erroneous markup, much of which would be considered "tag soup".
For example, the spec defines algorithms for dealing with syntax errors such as incorrectly-nested tags, which will ensure that a well-structured DOM tree can be produced. Defining that is essential for achieving interoperability between browsers and reducing the dependence upon browsers needing to reverse engineer each other's parsing behavior.
However, the conformance requirements for authors are defined separately from the processing requirements. Just because browsers are required to handle erroneous content, it does not make such markup conforming.
For example, user agents are required to support processing of the <marquee> element, but authors must not use the <marquee> element in conforming documents.
It is important to make the distinction between the rules that apply to user agents and the rules that apply to authors for producing conforming documents. They are completely orthogonal.
The spec allows <a> to contain blocks. It doesn't support putting href="" on any element, though.
Supporting href on any element has several problems associated with it that make it difficult to support in HTML. The main reason this isn't in HTML is that browser vendors have reported that implementing it would be extremely complex. Browser vendors get to decide what they implement, and there's no point to us telling them to do something they aren't going to do. In addition:
- It isn't backwards compatible with existing browsers.
- It adds no new functionality that can't already be achieved using the
aelement and a little script. - It doesn't make sense for all elements, such as interactive elements like
inputandbutton, where the use of href would interfere with their normal function.
The only advantage it seems to add is that it reduces typing for authors in some cases, but that is not a strong enough reason to support it in light of the other reasons.
Wrapping <a> elements around blocks solves most use cases. It doesn't handle making rows in tables into links, though; for those just do something like this instead:
<tr onclick="location = this.getElementsByTagName('a')[0]"> ... </tr>You can give a header to a list using the <figure> and <figcaption> elements:
<figure>
<figcaption>Apples</figcaption>
<ul>
<li>Granny Smith</li>
<li>Evil Apple of Knowledge</li>
<li>Apple, Inc</li>
</ul>
</figure>You can also label a group of lists using a definition list:
<dl>
<dt>Dry:</dt>
<dd>
<ul>
<li>1c flour</li>
<li>1/4c sugar</li>
<li>1tsp baking soda</li>
</ul>
</dd>
<dt>Wet:</dt>
<dd>
<ul>
<li>1 egg </li>
<li>1/2c milk</li>
<li>1tsp vanilla extract</li>
</ul>
</dd>
</dl>These techniques are preferred over adding an <lh> element as proposed in the old HTML3 draft, mostly because of thorny issues with parsing near <li> elements.
It does. You can use custom elements to build your own fully-featured DOM elements.
Short of that, there are actually quite a number of ways for people to invent their own extensions to HTML:
- Authors can use the
classattribute to extend elements, effectively creating their own elements, while using the most applicable existing "real" HTML element, so that browsers and other tools that don't know of the extension can still support it somewhat well. This is the tack used by Microformats, for example. - Authors can include data for scripts to process using the
data-*=""attributes. These are guaranteed to never be touched by browsers, and allow scripts to include data on HTML elements that scripts can then look for and process. - Authors can use the
<meta name="" content="">mechanism to include page-wide metadata. Names should be registered on the wiki's MetaExtensions page. - Authors can use the
rel=""mechanism to annotate links with specific meanings. This is also used by Microformats. Names should be registered on the wiki's RelExtensions page. - Authors can embed raw data using the
<script type="">mechanism with a custom type, for further handling by a script. - Authors can create plugins and invoke them using the
<embed>element. This is how Flash works. - Authors can extend APIs using the JS prototyping mechanism. This is widely used by script libraries, for instance.
- Authors can use the microdata feature (the
item=""anditemprop=""attributes) to embed nested name-value pairs of data to be shared with other applications and sites. - Authors can propose new elements and attributes and, if the wider community and user-agent vendors agree that they are worth the effort, they can be added to the language.
HTML allows <div> as a grouping element in <dl>. See the <dl> specification and issue #1937 wherein this was added.
Every feature we add to the web platform has a cost:
- Implementation: someone has to write code for it in each browser
- Testing: someone has to write the tests to check the features is working
- QA: someone has to regularly run the tests to make sure the feature doesn't regress
- Code maintenance: when browser vendors refactor code, they have to refactor more code if there's more features
- Tutorials: people who write tutorials have to include the feature, or handle feedback asking for them to do so
- Cognitive load: authors learning the platform have more documentation to wade through even if they don't care about the feature
- Extra features discourage exploration: Having more features means less overall feature usage.
- Page maintenance: authors have to know how to maintain the feature if other people have used it in pages they now maintain
- Spec writing: someone has to write the spec for the feature and ensure it's maintained
- Bug fixing: when bugs are found in the spec or implementations, someone has to figure out a fix, implement it, test it, ship it, tests have to be fixed, documentation has to be updated, etc
- Code size: each feature increases the size of browsers (both on-disk binaries and in-memory resident size)
No. The two groups have different goals. The WHATWG spec is intended to describe what browsers should aim for, introducing new features and describing reality in as much, and as accurate, detail as possible. The W3C spec is intended to follow the W3C process to REC.
On the WHATWG side, the editors read the feedback sent to both groups and take all input into account — and indeed there are far more places where input on HTML is sent than just these two mailing lists (e.g. blogs, [email protected], forums, direct mail, meetings, etc). (In particular, the editors do not look at the source of technical arguments when attempting to determine what path to take on an issue or other.)
The two groups have different specs, so each has authority over its spec. The specs can and have diverged on some topics; unfortunately, these differences are not documented anywhere.
Yes. The WHATWG originally committed to remaining consistent with the W3C spec unless the W3C working group showed a lapse in judgement. When that (in Hixie's opinion) occurred, there was little choice left but to let the specs diverge.
The plan to get the specs to converge again, such as it is, is to just do a better job with the WHATWG spec, such that it becomes the logical and obvious choice for anyone wanting to figure out which spec they should use.
Here are some documents that detail the history of HTML:
- A feature history of the modern web platform (2003 onward) (on GitHub)
- HTML's timeline on the ESW wiki (1997 to 2008)
- The history section in the HTML standard itself
The inclusion of those elements is a largely pragmatic decision based upon their widespread usage, and their utility for cases which are not covered by more-specific elements.
While there are a number of common use cases for italics which are covered by more-specific elements, such as emphasis (<em>), citations (<cite>), definitions (<dfn>) and variables (<var>), there are many other use cases which are not covered well by these elements. For example: a taxonomic designation, a technical term, an idiomatic phrase from another language, a thought, or a ship name.
Similarly, although a number of common use cases for bold text are also covered by more-specific elements, such as strong emphasis (<strong>), headings (<h1>-<h6>) or table headers (<th>), there are others which are not, such as keywords in a document abstract or product names in a review.
Some people argue that in such cases, the <span> element should be used with an appropriate class name and associated stylesheet. However, the <b> and <i> elements provide for a reasonable fallback styling in environments that don't support stylesheets or which do not render visually, such as screen readers, and they also provide some indication that the text is somehow distinct from its surrounding content.
In essence, the <i> and <b> elements convey distinct, though non-specific, semantics, which are to be determined by the reader in the context of their use. In other words, although they don't convey specific semantics by themselves, but instead they indicate that the content is somehow semantically distinct from its surroundings — leaving the interpretation of the semantics up to the reader.
This is further explained in the article The <b> and <i> Elements.
Similarly, the <small> element is defined for content that is commonly typographically rendered in small print, and which is often referred to as "fine print"; that could include copyright statements, disclaimers and other legal text commonly found at the end of a document.
The problem with elements like <font> isn't that they are presentational per se, it's that they are media-dependent (they apply to visual browsers but not to speech browsers). While <b>, <i> and <small> historically have been presentational, they are defined in a media-independent manner in HTML5. For example, <small> corresponds to the really quickly spoken part at the end of radio advertisements.
From what some have seen, <cite> is almost always used to mean "italics". More careful authors have used the element to mark up names and titles, and some people have gone out of their way to only mark up citations.
So, we can't really decide what the element should be based on past practice, like we usually do.
This leaves the question of what is the most useful use we can put the element to, if we keep it. The conclusion so far has been that the most useful use for <cite> is as an element to allow typographic control over titles, since those are often made italics, and that semantic is roughly close to what it meant in previous versions, and happens to match at least one of the common uses for the element. Generally, however, names and titles aren't typeset the same way, so making the element apply to both would lead to confusing typography.
There are already many ways of marking up names already (e.g. the hCard microformat, the microdata vCard vocabulary, <span> and class names, etc), if you really need it.
Some hopefully helpful hints:
- One way to look at it is how would you draw the page outline/table-of-contents? Each entry in the table of contents should be a
<section>/<article>/<aside>/<nav>, and if it's not in the table of contents and doesn't have an<h1>, it should probably not be a<section>/<article>/<aside>/<nav>. - You can still use
<div>. It's the right element if you need a styling hook because CSS can't give you enough to do what you want. - Generally,
<section>s should start with an<h1>and the section title. It's not a hard-and-fast rule, but if you find yourself in a situation where an<h1>would be inappropriate, you probably want<div>rather than<section>. - Sections can contain Articles, and vice versa. e.g. you can have a section that is news, a section that is editorials, a section that is sports, each with many articles, and each of those can have subsections, and each section can have comments, which are marked up using
<article>, and each comment could be big enough that it has separate<section>s, and so on.
The Web Forms 2.0 specification was folded into what is now the HTML Standard.
The Web Controls 1.0 specification was overtaken by events and has been abandoned. Its problem space is mostly handled by ARIA and Web Components now.
The DOM Parsing specification was abandoned by the WHATWG because the W3C was doing a better job of maintaining that specification. We do not want to cause confusion in the market place, so when another organization writes a specification that covers the same technology as one of ours, we only continue to publish it if our version is technically superior.