Re: Is longdesc a good solution? (was: Acessibility of <audio> and <video>)

James Craig 2008-09-06 04.56:

> Justin James wrote:


>> [...] If ARIA supports identical information as longdesc
>> does, I am in favor of it over using an attribute, simply because I think
>> that "people who really care" will be using ARIA, and overall I find the
>> system practical.
> 
> I would call it "equivalent," rather than "identical" functionality, 
> since longdesc is intended to be an external resource, and 
> aria-describedby is intended to be internal resource. For example, a DIV 
> representing an image could be marked to indicate that the caption text 
> is its long description:


Intended to be external resource, you said. Intended to be an 
initially invisible resource, I say. HTML 4 doesn't say the long 
description must point to another page, it just says it is a URI.

@Aria-describedby, however, is - as I understand it - meant to 
draw a formal link between e.g. an <img> and a relevant 
description. (And that description might not even be a real 
textual equivalent for the image.)

   [...]

> That said, there is nothing to stop you using it in a similar way to 
> longdesc, where aria-describedby would pointed to a link that referenced 
> an external file. 


So aria-describedby contains a URI and not - as I thought - an IDREF?

> It'd be interesting to see if browsers and AT 
> supported using it in this way though.


That indeed sounds like eating into the field of @longdesc and 
stretching the intent of @aria-describedby.

In the opposite direction, using @longdesc to point to a resource 
inside the page, allready works in AT which support @longdesc.

Proof of concept: http://www.malform.no/acidlongdesctest/qablog/

The page is from Karl's blog, in a post were a @longdesc was 
requested. Here I did what Karl was asked to do. Just doubleclick 
the illustration. (I would love to know where a text link would 
fit in on this page.)

(Caveat: It is a hastily done demonstration reusing code from 
another page. I think it works well in WebKit and Firefox, less 
well in Opera [needs quadruple click right now], currently not in 
IE - though AT which works with IE should work well. )
-- 
leif halvard silli

Received on Saturday, 6 September 2008 04:05:30 UTC