HTML5 defines the fifth major revision of the core language of the World Wide Web, HTML. "HTML5 differences from HTML4" describes the differences between HTML4 and HTML5 and provides some of the rationale for the changes. This document may not provide accurate information as the HTML5 specification is still actively in development. When in doubt, always check the HTML5 specification itself. [HTML5]
This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at http://www.w3.org/TR/.
This is the 24 June 2010 W3C Working Draft produced by the HTML Working Group, part of the HTML Activity. The Working Group intends to publish this document as a Working Group Note to accompany the HTML5 specification. The appropriate forum for comments is firstname.lastname@example.org, a mailing list with a public archive.
Publication as a Working Draft does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.
This document was produced by a group operating under the 5 February 2004 W3C Patent Policy. W3C maintains a public list of any patent disclosures made in connection with the deliverables of the group; that page also includes instructions for disclosing a patent. An individual who has actual knowledge of a patent which the individual believes contains Essential Claim(s) must disclose the information in accordance with section 6 of the W3C Patent Policy.
HTML has been in continuous evolution since it was introduced to the Internet in the early 1990s. Some features were introduced in specifications; others were introduced in software releases. In some respects, implementations and author practices have converged with each other and with specifications and standards, but in other ways, they continue to diverge.
The HTML5 draft reflects an effort, started in 2004, to study contemporary HTML implementations and deployed content. The draft:
HTML5 is still a draft. The contents of HTML5, as well as the contents of this document which depend on HTML5, are still being discussed on the HTML Working Group and WHATWG mailing lists. The open issues are linked from the HTML5 draft.
HTML5 is defined in a way that it is backwards compatible with the way user agents handle deployed content. To keep the authoring language relatively simple for authors several elements and attributes are not included as outlined in the other sections of this document, such as presentational elements that are better dealt with using CSS.
User agents, however, will always have to support these older elements
and attributes and this is why the HTML5 specification clearly separates
requirements for authors and user agents. For instance, this means that
authors cannot use the
isindex or the
element, but user agents are required to support them in a way that is
compatible with how these elements need to behave for compatibility with
Since HTML5 has separate conformance requirements for authors and user agents there is no longer a need for marking features "deprecated".
The HTML5 specification will not be considered finished before there are at least two complete implementations of the specification. A test suite will be used to measure completeness of the implementations. This approach differs from previous versions of HTML, where the final specification would typically be approved by a committee before being actually implemented. The goal of this change is to ensure that the specification is implementable, and usable by authors once it is finished.
HTML5 defines an HTML syntax that is compatible with HTML4 and XHTML1
documents published on the Web, but is not compatible with the more
esoteric SGML features of HTML4, such as processing
instructions and shorthand
markup as these are not supported by most user agents. Documents using
the HTML syntax are almost always served with the
HTML5 also defines detailed parsing rules (including "error handling")
for this syntax which are largely compatible with popular implementations.
User agents must use these rules for resources that have the
text/html media type. Here is an example document that
conforms to the HTML syntax:
<!doctype html> <html> <head> <meta charset="UTF-8"> <title>Example document</title> </head> <body> <p>Example paragraph</p> </body> </html>
HTML5 also defines a
text/html-sandboxed media type for
documents using the HTML syntax. This can be used when hosting untrusted
The other syntax that can be used for HTML5 is XML. This syntax is
compatible with XHTML1 documents and implementations. Documents using this
syntax need to be served with an XML media type and elements need to be
put in the
http://www.w3.org/1999/xhtml namespace following
the rules set forth by the XML specifications. [XML]
Below is an example document that conforms to the XML syntax of HTML5.
Note that XML documents must be served with an XML media type such as
<?xml version="1.0" encoding="UTF-8"?> <html xmlns="http://www.w3.org/1999/xhtml"> <head> <title>Example document</title> </head> <body> <p>Example paragraph</p> </body> </html>
For the HTML syntax of HTML5, authors have three means of setting the character encoding:
Content-Typeheader for instance.
metaelement with a
charsetattribute that specifies the encoding within the first 512 bytes of the document. E.g.
<meta charset="UTF-8">could be used to specify the UTF-8 encoding. This replaces the need for
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">although that syntax is still allowed.
For the XML syntax, authors have to use the rules as set forth in the XML specifications to set the character encoding.
The HTML syntax of HTML5 requires a DOCTYPE to be specified to ensure that the browser renders the page in standards mode. The DOCTYPE has no other purpose and is therefore optional for XML. Documents with an XML media type are always handled in standards mode. [DOCTYPE]
The DOCTYPE declaration is
<!DOCTYPE html> and is
case-insensitive in the HTML syntax. DOCTYPEs from earlier versions of
HTML were longer because the HTML language was SGML-based and therefore
required a reference to a DTD. With HTML5 this is no longer the case and
the DOCTYPE is only needed to enable standards mode for documents written
using the HTML syntax. Browsers already do this for
The HTML syntax of HTML5 allows for MathML and SVG elements to be used inside a document. E.g. a very simple document using some of the minimal syntax features could look like:
<!doctype html> <title>SVG in text/html</title> <p> A green circle: <svg> <circle r="50" cx="50" cy="50" fill="green"/> </svg> </p>
More complex combinations are also possible. E.g. with the SVG
foreignObject element you could nest MathML, HTML, or both
inside an SVG fragment that is itself inside HTML.
There are a few other syntax changes worthy of mentioning:
langattribute takes the empty string in addition to a valid language identifier, just like
xml:langdoes in XML.
This section is split up in several subsections to more clearly illustrate the various differences there are between HTML4 and HTML5.
The links in this section may stop working if elements are renamed and/or removed. They should function in the latest version of this draft.
The following elements have been introduced for better structure:
represents an independent piece of content of a document, such as a blog
entry or newspaper article.
represents a piece of content that is only slightly related to the rest
of the page.
represents the header of a section.
represents a group of introductory or navigational aids.
represents a footer for a section and can contain information about the
author, copyright information, et cetera.
represents a section of the document intended for navigation.
represents a piece of self-contained flow content, typically referenced
as a single unit from the main flow of the document.
<figure> <video src="ogg"></video> <figcaption>Example</figcaption> </figure>
can be used as caption (it is optional).
Then there are several other new elements:
for multimedia content. Both provide an API so application authors can
script their own user interface, but there is also a way to trigger a
user interface provided by the user agent.
elements are used together with these elements if there are multiple
streams available of different types.
is used for plugin content.
represents represents a run of text in one document marked or
highlighted for reference purposes, due to its relevance in another
represents a completion of a task, such as downloading or when
performing a series of expensive operations.
represents a measurement, such as disk usage.
represents a date and/or time.
represents a line break opportunity.
is used for rendering dynamic bitmap graphics on the fly, such as graphs
represents a command the user can invoke.
together with the a new
list attribute for
input can be used to make comboboxes:
<input list="browsers"> <datalist id="browsers"> <option value="Safari"> <option value="Internet Explorer"> <option value="Opera"> <option value="Firefox"> </datalist>
represents control for key pair generation.
represents some type of output, such as from a calculation done through
type attribute now has the
following new values:
The idea of these new types is that the user agent can provide the user interface, such as a calendar date picker or integration with the user's address book, and submit a defined format to the server. It gives the user a better experience as his input is checked before sending it to the server meaning there is less time to wait for feedback.
HTML5 has introduced several new attributes to various elements that were already part of HTML4:
area elements now have a
media attribute for consistency with the
area elements have a new attribute
ping that specifies a space-separated list of URLs
which have to be pinged when the hyperlink is followed. Currently user
tracking is mostly done through redirects. This attribute allows the
user agent to inform users which URLs are going to be pinged as well as
giving privacy-conscious users a way to turn it off.
area element, for consistency with the
link elements, now also has the
base element can now have a
attribute as well, mainly for consistency with the
element. (This is already widely supported.) Also, the
target attribute for the
area elements is no longer deprecated, as it is useful in
Web applications, e.g. in conjunction with
value attribute for the
li element is no
longer deprecated as it is not presentational. The same goes for the
start attribute of the
meta element has a
charset attribute now
as this was already widely supported and provides a nice way to specify
the character encoding for the
autofocus attribute can be specified on the
input (except when the
type attribute is
button elements. It provides a declarative way to focus a
form control during page load. Using this feature should enhance the
user experience as the user can turn it off if the user does not like
it, for instance.
placeholder attribute can be specified on the
textarea elements. It represents a
hint intended to aid the user with data entry.
form attribute for
fieldset elements allows for
controls to be associated with a form. I.e. these elements can now be
placed anywhere on a page, not just as descendants of the
<label>Email: <input type=email form=x name=email> </label> <form id=x></form>
required attribute applies to
(except when the
type attribute is
image or some button type such as
textarea. It indicates that the user has to fill in a value
in order to submit the form.
fieldset element now allows the
attribute. It disables all descendant controls when specified.
input element has several new attributes to specify
step. As mentioned before it also has a new
list attribute which can be used together with the
form element has a
that can be used to disable form validation submission (i.e. the form
can always be submitted).
button elements have
formtarget as new attributes. If present, they override the
target attributes on the
menu element has two new attributes:
label. They allow the element to
transform into a menu as found in typical user interfaces as well as
providing for context menus in conjunction with the global
style element has a new
which can be used to enable scoped style sheets. Style rules within such
style element only apply to the local tree.
script element has a new attribute called
async that influences script loading and execution.
html element has a new attribute called
manifest that points to an application cache manifest used
in conjunction with the API for offline Web applications.
link element has a new attribute called
sizes. It can be used in conjunction with the
icon relationship (set through the
attribute) to indicate the size of the referenced icon. Thus allowing
for icons of distinct dimensions.
ol element has a new attribute called
reversed. When present, it indicates that the list order is
iframe element has three new attributes called
which allow for sandboxing content, e.g. blog comments.
Several attributes from HTML4 now apply to all elements. These are
called global attributes:
There are also several new global attributes:
contenteditableattribute indicates that the element is an editable area. The user can change the contents of the element and manipulate the markup.
contextmenuattribute can be used to point to a context menu provided by the author.
data-*collection of author-defined attributes. Authors can define any attribute they want as long as they prefix it with
data-to avoid clashes with future versions of HTML. The only requirement on these attributes is that they are not used for user agent extensions.
draggableattribute can be used together with the new drag & drop API.
hiddenattribute indicates that an element is not yet, or is no longer, relevant.
aria-*collection attributes which can be used to instruct assistive technology.
spellcheckattribute allows for hinting whether content can be checked for spelling or not.
HTML5 also makes all event handler attributes from HTML4, which take the
onevent-name, global attributes and adds
several new event handler attributes for new events it defines. E.g. the
play event which is used by the API for the media elements
These elements have slightly modified meanings in HTML5 to better reflect how they are used on the Web or to make them more useful:
a element without an
href attribute now
represents a placeholder for where a link otherwise might have been
placed. It can also contain flow content rather than being restricted to
address element is now scoped by the new concept of
b element now represents a span of text to be
stylistically offset from the normal prose without conveying any extra
importance, such as keywords in a document abstract, product names in a
review, or other spans of text whose typical typographic presentation is
cite element now solely represents the title of a
work (e.g. a book, a paper, an essay, a poem, a score, a song, a script,
a film, a TV show, a game, a sculpture, a painting, a theatre
production, a play, an opera, a musical, an exhibition, a legal case
report, etc). Specifically the example in HTML4 where it is used to mark
up the name of a person is no longer considered conforming.
hr element now represents a paragraph-level thematic
i element now 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. Usage varies widely by
label element the browser should no longer move
focus from the label to the control unless such behavior is standard for
the underlying platform user interface.
menu element is redefined to be useful for toolbars
and context menus.
small element now represents small print (for side
comments and legal print).
strong element now represents importance rather than
head element no longer allows the
element as child.
The following attributes are allowed but authors are discouraged from using them and instead strongly encouraged to use an alternative solution:
border attribute on
img. It is required
to have the value "
0" when present. Authors can use CSS
language attribute on
script. It is
required to have the value "
when present and cannot conflict with the
Authors can simply omit it as it has no useful function.
name attribute on
a. Authors can use the
id attribute instead.
summary attribute on
table. The HTML5
draft defines several alternative solutions.
The elements in this section are not to be used by authors. User agents
will still have to support them and various sections in HTML5 define how.
E.g. the obsolete
isindex element is handled by the parser
The following elements are not in HTML5 because their effect is purely presentational and their function is better handled by CSS:
The following elements are not in HTML5 because their usage affected usability and accessibility for the end user in a negative way:
The following elements are not included because they have not been used often, created confusion, or their function can be handled by other elements:
acronymis not included because it has created a lot of confusion. Authors are to use
applethas been obsoleted in favor of
isindexusage can be replaced by usage of form controls.
dirhas been obsoleted in favor of
noscript element is only conforming in the HTML
syntax. It is not included in the XML syntax as its usage relies on an
Some attributes from HTML4 are no longer allowed in HTML5. If they need to have any impact on user agents for compatibility reasons it is defined how they should work in those scenarios.
In addition, HTML5 has none of the presentational attributes that were in HTML4 as their functions are better handled by CSS:
HTML5 introduces a number of APIs that help in creating Web applications. These can be used together with the new elements introduced for applications:
HTML5 has extended the
HTMLDocument interface from DOM
Level 2 HTML in a number of ways. The interface is now implemented on
all objects implementing the
Document interface so
it stays meaningful in a compound document context. It also has several
noteworthy new members:
getElementsByClassName() to select elements by their
class name. The way this method is defined will allow it to work for any
class attributes and a
object such as SVG and MathML.
innerHTML as an easy way to parse and serialize an HTML
or XML document. This attribute was previously only available on
HTMLElement in Web browsers and not part of any standard.
hasFocus to determine
which element is currently focused and whether the
has focus respectively.
getSelection() which returns an object that represents
the current selection(s).
HTMLElement interface has also gained several
extensions in HTML5:
getElementsByClassName() which is basically a scoped
version of the one found on
innerHTML as found in Web browsers today. It is also
defined to work in XML context (when it is used in an XML document).
classList is a convenient accessor for
className. The object it returns, exposes methods
toggle()) for manipulating the element's classes. The
link elements have a
similar attribute called
relList that provides the same
functionality for the
The changelogs in this section indicate what has been changed between
publications of the HTML5 drafts. Rationale for changes can be found in
mailing list archives, and the This Week in
HTML5 series of blog posts. More fundamental rationale is being
collected on the WHATWG Rationale wiki page. Many
editorial and minor technical changes are not included in these
changelogs. I.e. implementors are strongly encouraged to follow the
development of the main specification on a frequent basis so they become
aware of all changes that affect them early on.
The changes in the changelogs are in rough chronological order to ease editing this document.
titleelement is optional for
srcdocdocuments and other scenarios were a title is already available. As is the case with email.
keywordsis now a standard metadata name for the
allow-top-navigationvalue has been added for the
sandboxattribute on the
iframeelement. It allows the embedded content to navigate its parent when specified.
wbrelement has been added.
alternatekeyword for the
relattribute of the
linkelement can now be used to point to feeds again, even if the feed is not an alternative for the document.
In addition lots of minor changes, clarifications, and fixes have been made to the document.
dialogelement has been removed. A section with advice on how to mark up conversations has effectively replaced it.
document.headhas been introduced to provide convenient access to the
headelement from script.
feedhas been removed.
alternatewith specific media types is to be used instead.
createHTMLDocument()has been introduced as API to allow easy creation of HTML documents.
progresselements no longer have "magic" processing of their contents because it could not be made to work internationally.
progresselements, as well as the
outputelement, can now be labeled using the
text/html-sandboxed, was introduced to allow hosting of potentially hostile content without it causing harm.
srcdocattribute for the
iframeelement was introduced to allow embedding of potentially hostile content inline. It is expected to be used together with the
figureelement now uses a new element
legendbecause people want to use HTML5 long before it reaches W3C Recommendation.
detailselement now uses a new element
summaryfor exactly the same reason.
autobufferattribute on media elements was renamed to
A whole lot of other smaller issues have also been resolved. The above list summarizes what is thought to be of primary interest to authors.
In addition to all of the above, Microdata, the 2D context API for
canvas, and Web Messaging (
have been split into their own drafts at the W3C (the WHATWG still
publishes a version of HTML5 that includes them):
Specific microdata vocabularies are gone altogether in the W3C draft of HTML5 and are not published as a separate draft. The WHATWG draft of HTML5 still includes them.
timeelement is empty user agents have to render the time in a locale-specific manner.
loadevent is dispatched at
Window, but now has
Documentas its target.
pushState()now affects the
onredoare now on
startTimemember that indicates where the current resource starts.
headerhas been renamed to
hgroupand a new
headerelement has been introduced.
createImageData()now also takes
createPattern()can now take a
videoelement as argument too.
footerelement is no longer allowed in
headeris not allowed in
accesskeyis now properly defined.
articlenow take a
textLengthhas been added as member of the
rpelement now takes phrasing content rather than a single character.
location.reload()is now defined.
hashchangeevent now fires asynchronously.
spellcheckIDL attribute now maps to a
hasFeature()support has been reduced to a minimum.
Audio()constructor sets the
tdelement is no longer allowed in
DataTransferobject now have a
bbhave been removed due to their design not being agreed upon.
On top of this list quite a few minor clarifications, typos, issues specific to implementors, and other small problems have been resolved.
In addition, the following parts of HTML5 have been taken out and will likely be further developed at the IETF:
spellcheckhas been added.
thisin the global object returns a
WindowProxyobject rather than the
valueIDL attribute for
inputelements in the File Upload state is now defined.
designModewas changed to be more in line with legacy implementations.
drawImage()method of the 2D drawing API can now take a
videoelement as well.
document.domainis now IPv6-compatible.
videoelement gained an
autobufferboolean attribute that serves as a hint.
metaelement with a
charsetattribute in XML documents if the value of that attribute matches the encoding of the document. (Note that it does not specify the value, it is just a talisman.)
bufferingThrottledmembers of media elements have been removed.
postMessage()API now takes an array of
MessagePortobjects rather than just one.
add()method on the
selectelement and the
optionsmember of the
selectelement is now optional.
buttonelements have been renamed to
localStorage) at the same time. The
getStorageUpdates()method to allow it to be explicitly released.
placeholderattribute has been added to the
keygenelement for key pair generation.
datagridelement was revised to make the API more asynchronous and allow for unloaded parts of the grid.
In addition, several parts of HTML5 have been taken out and will be further developed by the Web Applications Working Group as standalone specifications:
ImageDataobjects has been changed from an array to a
canvaselement and its API.
canvashave been made in response to implementation and author feedback. E.g. clarifying what happens when NaN and Infinity are passed and fixing the definitions of
innerHTMLin XML was slightly changed to improve round-tripping.
toDataURL()method on the
canvaselement now supports setting a quality level when the media type argument is
posterattribute of the
videoelement now affects its intrinsic dimensions.
typeattribute of the
linkelement has been clarified.
linkwhen the expected type is an image.
hrefattribute of the
baseelement does not depend on
xmlnsattribute with the value
http://www.w3.org/1999/xhtmlis now allowed on all HTML elements.
data-*attributes and custom attributes on the
embedelement now have to match the XML
Nameproduction and cannot contain a colon.
volumeon media elements is now 1.0 rather than 0.5.
event-sourcewas renamed to
eventsourcebecause no other HTML element uses a hyphen.
bbhas been added. It represents a user agent command that the user can invoke.
addCueRange()method on media elements has been modified to take an identifier which is exposed in the callbacks.
parentattribute of the
Windowobject is now defined.
embedelement is defined to do extension sniffing for compatibility with servers that deliver Flash as
text/plain. (This is marked as an issue in the specification to figure out if there is a better way to make this work.)
embedcan now be used without its
getElementsByClassName()is defined to be ASCII case-insensitive in quirks mode for consistency with CSS.
localNameno longer returns the node name in uppercase.
data-*attributes are defined to be always lowercase.
openerattribute of the
Windowobject is not to be present when the page was opened from a link with
topattribute of the
Windowobject is now defined.
aelement now allows nested flow content, but not nested interactive content.
headerelement means to document summaries and table of contents.
autosubmitattribute has been removed from the
insertAdjacentHTML()has been added.
xml:langis now allowed in HTML when
langis also specified and they have the same value. In XML
langis allowed if
xml:langis also specified and they have the same value.
frameElementattribute of the
Windowobject is now defined.
altattribute is omitted a
titleattribute, an enclosing
figureelement with a
legendelement descendant, or an enclosing section with an associated heading must be present.
irrelevantattribute has been renamed to
definitionURLattribute of MathML is now properly supported. Previously it would have ended up being all lowercase during parsing.
load()method on media elements has been redefined as asynchronous. It also tries out files in turn now rather than just looking at the
typeattribute of the
canPlayType()has been added to the media elements.
bufferedBytesattributes have been removed from the media elements.
Locationobject gained a
qelement has changed again. Punctuation is to be provided by the user agent again.
beforeunloadevents are now defined.
headersattribute pointing to a
thelement, but authors are required to only let them point to
metaelement has a
charsetattribute it must occur within the first 512 bytes.
StorageEventobject now has a
Windowobjects is now defined.
Windowobject gained the
toolbarattributes giving information about the user interface.
document.domainnow relies on the Public Suffix List. [PSL]
Web Forms 2.0, previously a standalone specification, has been fully integrated into HTML5 since last publication. The following changes were made to the forms chapter:
datalistelements through the
dataattribute has been removed.
dispatchFormChange()methods have been removed from the
inputmodeattribute has been removed.
inputelement in the File Upload state no longer supports the
inputelements in the File Upload state is no longer authoritative.
textareahave been removed.
submit()method now just submits, it no longer ensures the form controls are valid.
inputelement in the Range state now defaults to the middle, rather than the minimum value.
sizeattribute on the
inputelement is now conforming (rather than deprecated).
objectelements now partake in form submission.
typeattribute of the
inputelement gained the values
inputelement gained a
multipleattribute which allows for either multiple e-mails or multiple files to be uploaded depending on the value of the
formelements now have a
novalidateattribute to indicate that the form fields should not be required to have valid values upon submission.
labelelement contains an
inputit may still have a
forattribute as long as it points to the
inputelement it contains.
inputelement now has an
inputelement gained a
pingattribute have changed.
<meta http-equiv=content-type>is now a conforming way to set the character encoding.
canvaselement has been cleaned up. Text support has been added.
globalStorageis now restricted to the same-origin policy and renamed to
localStorage. Related event dispatching has been clarified.
postMessage()API changed. Only the origin of the message is exposed, no longer the URL. It also requires a second argument that indicates the origin of the target document.
dataTransferobject now has a
typesattribute indicating the type of data being transferred.
melement is now called
figureelement no longer requires a caption.
olelement has a new
queryCommandEnabled()and related methods.
headersattribute has been added for
tableelement has a new
data-nameand can access these through the DOM using
dataset[name]on the element in question.
qelement has changed to require punctuation inside rather than having the browser render it.
targetattribute can now have the value
showModalDialogAPI has been added.
document.domainAPI has been defined.
sourceelement now has a new
pixelratioattribute useful for videos that have some kind encoding error.
bufferingThrottledIDL attributes have been added to the
beginevent has been renamed to
loadstartfor consistency with the Progress Events specification.
charsetattribute has been added to
iframeelement has gained the
seamlessattributes which provide sandboxing functionality.
rpelements have been added to support ruby annotation.
showNotification()method has been added to show notification messages to the user.
afterprintevents has been added.
The editor would like to thank Ben Millard, Cameron McCormack, Charles McCathieNevile, Dan Connolly, David Håsäther, Dennis German, Frank Ellermann, Frank Palinkas, Gordon P. Hemsley, Henri Sivonen, James Graham, Jens Meiert, Jeremy Keith, Jürgen Jeka, Krijn Hoetmer, Leif Halvard Silli, Maciej Stachowiak, Marcos Caceres, Mark Pilgrim, Martijn Wargers, Martyn Haigh, Masataka Yakura, Michael Smith, Olivier Gendrin, Øistein E. Andersen, Philip Taylor, Simon Pieters, Toby Inkster, and Yngve Spjeld Landro for their contributions to this document as well as to all the people who have contributed to HTML5 over the years for improving the Web!