From HTML WG Wiki
Jump to: navigation, search

Issue 30 longdesc Change Proposal (shadow version)

This is a shadow version of [1] that I expect to delete at some point. I opened it in order to be able to let my creative juices flow, while the original editor is evaluating my contributions.

The following is a Change Proposal for ISSUE 30 when it is reopened.

Original editor: Laura Carlson - [2]
Further editions from: Leif Halvard Silli

Date: January 31, 2011, last updated February 1, 2011.

Please address feedback to the HTML Working Group mailing list (


Instate the longdesc attribute from HTML 4 into HTML5 as an allowed attribute on images.


Use Cases

Uses cases have been identified that specifically require longdesc.

Formal Use Cases

For formal use cases requiring longdesc, please visit Long Description Research: Use Cases.

Primary Use Case Overview

Longdesc affords authors the native capability to provide information that is essential for blind and visually impaired users but would be redundant for sighted users and unacceptable to visual designers' aesthetics.

It is an accommodation mechanism for people who are blind or have a visual impairment and use a screen reader. It is a tool to supply programmatically determinable descriptions of images such as data visualization (i.e. charts and graphs), diagrams, cartoons, logos drawings, illustrations, maps, photographs, etcetera when:

  • An image's content is visually apparent and typically redundant to a sighted person, and/or
  • It is unacceptable to a marketing department or web author to use another technique due to aesthetic considerations. Many artists, designers and marketers do not want their visual designs changed/ruined with visible link text. (Longdesc is naively free from a visual encumbrance.), and/or
  • The image also serves as a link. With longdesc it is programmatically possible to separate the activation of the longdesc for exposure from the UA's universal link activation action (which is usually activated with the ENTER key, the SpaceBar, or by mouse click), so that the linked image retains the expected behavior in response to user interaction while a discrete mechanism is used to retrieve the long description.

The cartoonist Kyle Weems (aka CSS Squirrel) has explained:

@Nick - With the exception of this most recent comic, all the comics are made for sighted users and navigable via the previous/next links below the comic. Those links are targetable by the sort of technology a physically-disabled person would use to navigate most links.

The issue at hand was directed at a piece of technology made for non-sighted users, however, so this comic provided an exception to deal with that specific experience.

@Mattur - I have no qualms with other people using hyperlinks where they desire to provide alternate text. However, as a designer, I object to being told I must use those links myself. As you've pointed out on Twitter, the current design of the comic page would certainly support a hyperlink wrapping around the comic. However, my upcoming design already has functionality mapped to clicking the comic, and won't have space for a large "transcript here" hyperlink sitting around in plain sight (which would be distracting for the 99% of my users that are sighted). In that scenario, longdesc can and does serve my needs.

Also, are we going to pretend that using longdesc is difficult? Yes, people may use it wrong without some correction, but simply saying "Hey, put a URL there," is not complex at all, and most of longdesc's non-use is a lack of awareness or caring (most non-longdesc websites simply don't offer alt text at all.)"

Other Use Cases

The content in a longdesc's target explains what is visually evident. This is similar to an audio description of a video being redundant to people who can see. Audio descriptions describe items that take place visually which are needed for complete understanding to people who can't see. Sighted users don't typically need them. The same is true of longdesc.

However, the following sighted people may be aided by access to a longdesc:

  • Users who have a cognitive impairment that makes it difficult or impossible to read, and use a screen reader.
  • Users who have a visual impairment, but who do not use assistive technology.
  • Users who turn off images to decrease bandwidth use in order to lower their Internet usage fees.
  • Users who have a cognitive impairment, but who do not use assistive technology, who need the long description in order to process the image properly/in the intended order; such users may also benefit from the structured content of a verbose descriptor (for example, the use of an ordered list).
    • However, the majority of people with cognitive impairments would be aided more by a simplified step-by-step visual. Text would be additional clutter and cause confusion as it can create difficulties for users with cognitive impairments. Appropriate graphics can be used to help reduce cognitive load and enhance understanding Cite: Jiwnani, K. "Designing for Users with Cognitive Disabilities", (2001).
  • Authors for ease of authoring and maintenance purposes.

Access to the content of longdesc attribute for the sighted should be similar to television closed captions. Closed captions are encoded or invisible to the sighted by default and must be decoded or made visible. There is a reason that closed captions (as opposed to open captions) are the default on televisions. Sighted people rarely require them. To them, they are visual noise. Clutter. Redundant. But if a sighted person wants to enable closed captions they can do so via a user preference built into the system menu. It is a user choice. Televisions do not have a default on-screen visual indicator. There is no forced visual encumbrance. This is by design.


The 80/20 rule is not applicable for longdesc. Usage is not and will never be widespread because:

  • People with disabilities are a minority.
  • Images that need longdesc are a minority.

However, longdesc has been used in that minority. Obsoleting longdesc breaks the web for over 150 sites in the wild. The longdesc attribute is in current use on company, organizational, governmental, educational, and personal sites throughout the world, including but not limited to:

  • Argentina
  • Australia
  • Austria
  • Belgium
  • Canada
  • China
  • England
  • France
  • Germany
  • Ireland
  • Italy
  • Japan
  • Myanmar
  • New Zealand
  • Portugal
  • Spain
  • South Korea
  • United Kingdom
  • United States

Numerous sites do not have longdesc visual indicators. Some go to great lengths to hide a longdesc visual indicator via invisible D-link or invisible a href.

In an effort to mitigate damages of user agents that do not yet have a long description feature built directly into them, some sites do provide a separate redundant link or a D-link to the long description. This link does not programmatically tie the image to the description. With proper implementation in browsers all such links could be solely longdesc.


The typical implemention of the longdesc attribute in visual user agents and screenreaderes, can be illustrated by the following pseudo-code: <a href="link" target="_blank" rel="longdesc" />. Meaning: the URI of a longdesc is typically opened in a new window or tab, and the user is made fully aware — through an announcement of the link or because the link is accessed in a context menu — that it is a long description kind of link. The, often loathed, trick of opening the link in a new window or tab, means that the usage state is preserved: a simple closing of the tab or window brings the user right back to the same situation and the exact same location he or she — or the screenreader voice — was reading before the opening of the link. Thus it causes minimal disruption — or minimal fear of disruption (browsers of course have back buttons too) – for the user. This is also, by the way, how the cite attribute often is implemented.

Assistive Technology

Modern screen readers have good support of the longdesc attribute. They typically announce the presence of a long description when available, and provide users with the option of reading it by executing a specified keystroke.


Failure by some browser vendors to fully implement a feature is not in itself sufficient justification for obsoleting it. However, longdesc has been implemented as follows:

Authoring Tools

Many authoring tools support longdesc. Among them are:

Other Tools Supporting Longdesc

longdesc Advantages

aria-describedby (is not a Functional Replacement for longdesc

  • aria-describedby kills of links: ARIA 1.0 specifies that anything that aria-describedby points to is presented to the user as if it it occurred inside an attribute. Hence, if aria-describedby points to an element which is — or contains — a link, the link will be completely dead — the AT won't even inform the user about the link presence.
  • aria-describedby
  • aria-describedby naively forces a visual encumbrance on sighted users.
  • aria-describedby will annotate text in the target id referenced by the idref. This means assistive technology users would not be able to control how they interact with the long description (as they can with longdesc). It is read aloud without any user intervention, forcing the longer description on the user whether they want it or not.
  • aria-describedby is limited to text that appears in the same document as the image being described.
  • As, by definition, a long description is in fact long, aria-describedby is not good solution for a longdesc.
  • The content associated using aria-describedby as currently implemented, is limited to unstructured text. AT treats aria-describedby target content as though it does not have any mark-up. It is treated as a string of text.
  • aria-describedby target content is a forced visual encumbrance on sighted users by default. Many artists, designers and marketers do not want their visual designs changed/ruined with redundant text.
  • aria-describedby is not native HTML.
  • PF "likes the idea of having built in semantics in HTML and in particular would prefer to have common document elements."
  • It is unlikely that many content creators or developers will learn ARIA (something not native HTML). They already feel like they've learned far more than they should have to know under their job description. And in many cases, their supervisors agree. (reference Cliff Tyllick)



Spec Changes:

  • Some differenes from the Change Proposal that the HTML Working Group has handled earlier:

    In harmony with HTML5's general resctrictions on nesting of interactive content as well as in harmony with the requirement in HTML4 that longdesc access and link access have separate user interfaces (which is also how user agents and AT have implemented it), this Change Proposal does not suggest that longdesc turns an img element into interactive content, in HTML5's meaning of that phrase. Rather, an img element with the longdesc attribute is comparable to a blockquote, del, ins or q element with the cite attribute present. Parts of the suggested spec text changes are therefore taken from HTML5's specifications of the cite attribute, including the phrase ‘should allow users to follow such description links’.

  • Specification changes in HTML5's img section:
    1. Add longdesc to the list of content attributes;
    2. Add attribute DOMString longDesc; to the attributes listed in the DOM Interface for the img element;
    3. Add the following text to the body of the same section:

      The longdesc attribute can be used to specify the address of a resource – typically a document – that describes the image. The description resource might be one that was crafted especially to supplement the alternative text (typically kept inside the alt attribute) of the image as it appears in the document — thus it might be a contextual description. Or it could be a description that is more independent of the particular page's use of that image that is thought to be useful because it provides information and background that can help the user to understand or identify (including remember) the image and why it appears in this particular context — thus, it might be a more general description of the image. In either case, the description might essentially offer an accessible version of the image information – for example, if the image contains a diagram or a comic, the description could include the same info as narrative text, as a list, inside a table element or in a combination — thus a functional description or equivalent of the image. Description resources that takes long time to open, might break the user expectation about such links.

      URI variations: If the description document is long, for instance because it contains descriptions for several images, authors are encouraged to include a fragment identifier pointing to the specific part of that document that describes this image. The fragment identifier of the longdesc attribute might even lead to a specific part on the very same page where the img element occurs — this either in order to be explicit about which content is associated with the image or in order to link to a section that is hidden until the URL of the longdesc is activated. In the latter case it is even possible to keep the description in a data URI inside the longdesc attribute itself.

      If the longdesc attribute is present, it must be a valid URL potentially surrounded by spaces. Conformance checkers may «sniff» the URL an issue warnings if it suspects that the description resource is unlikely to contain a description of the image.

      To obtain the corresponding description link, the value of the attribute must be resolved relative to the element. User agents, and in particular AT, should allow users to follow such description links.

      The description link of the longdesc attribute must be accessible even when the img has the usemap attribute and thus is interactive as well as when the parent element, for example an a element, is interactive. A user agent therefore has to offer a user interface for accessing the 'longdesc' resource that is different from the mechanism for accessing the interactive feature.

  • Specification changes in HTML5's ARIA section:
    • specify Strong native semantics and implied ARIA semantics against role=presentation whenever the img contains the longdesc attribute.
      In other words, neither role=presentation nor an empty alt attribute (which is equivalent to role=presentation) is permitted if the img has the longdesc attribute set.

Changes for other sections:


  • should mention it as the preferred way to point to a description of the image if this is desired, rather than mis-using the alt attribute for this purpose.


  • In section A key part of the content: without making the question of permission to completely drop the textual substitute — the alt attribute — part of this very Change Proposal, state that where an image is a key part of the content, longdesc can be used to point to a supplementing description resource for the image.


This has no impact on existing HTML-4 browsers, many of which fail to make longdesc accessible other than via the DOM. Failure to make this change will have an impact on assistive technologies such as screen readers, which use the longdesc attribute to find descriptions of images.

This would require conformance checking to accept the attribute as valid, and would imply maintaining the existing requirement on Authoring Tools to allow the author to use this functionality. It would maintain conformance of HTML-4 tools and content, rather than the current expected change leaving them non-conforming.



HTML 4 on longdesc

  • HTML4 16.4.2 Long descriptions of frames: The longdesc attribute allows authors to make frame documents more accessible to people using non-visual user agents.
  • HTML4 Index of Attributes: longdesc "link to long description (complements alt)".
  • HTML4 longdesc = uri: "This attribute specifies a link to a long description of the image. This description should supplement the short description provided using the alt attribute. When the image has an associated image map, this attribute should provide information about the image map's contents. This is particularly important for server-side image maps. Since an IMG element may be within the content of an A element, the user agent's mechanism in the user interface for accessing the "longdesc" resource of the former must be different than the mechanism for accessing the href resource of the latter...The alt attribute provides a short description of the image. This should be sufficient to allow users to decide whether they want to follow the link given by the longdesc attribute to the longer description, here "sitemap.html".
  • HTML4 A.3 Changes for accessibility "Authors may provide long descriptions of tables, images, and frames (see the longdesc attribute)."

HTML5 on longdesc

longdesc is currently listed as an obsolete feature in the HTML5 editor's draft. It advises to "Use a regular a element to link to the description, or (in the case of images) use an image map to provide a link from the image to the image's description."

Issue 30

Previous Proposals

Formal Objections to ISSUE 30 Decision


Related Bugs

  1. HTML Bug 10015: longdesc URL checking - Status VERIFIED INVALID.
  2. HTML Bug 10016: longdesc and @role (ARIA) - Status RESOLVED WONTFIX.
  3. HTML Bug 10017: longdesc and @aria-describedby (ARIA) - Status VERIFIED INVALID.
  4. HTML Bug 10019: Native user agent support for exposing longdesc to all users - Status RESOLVED WONTFIX.
  5. HTML Bug 10434: Mint a link type for pointing to long descriptions (rel="longdesc") - Status RESOLVED WORKSFORME.
  6. HTML Bug 10455: Mint a describedby attribute for the img element - Status RESOLVED WONTFIX.
  7. HTML Bug 10853: HTML5 lacks a verbose description mechanism - Status RESOLVED WONTFIX.
  8. HTML Bug 10967: Add @desclink, a description link attr. for any embedded element + figure - Status RESOLVED WONTFIX.
  9. HTML Bug 11012: Also say that <area>/image maps is an alternative to @longdesc - Status: RESOLVED FIXED.