Re: Alternate proposal for ISSUE-30 longdesc

Maciej Stachowiak, Mon, 22 Feb 2010 09:10:09 -0800:
> On Feb 22, 2010, at 8:42 AM, Charles McCathieNevile wrote:
>> On Mon, 22 Feb 2010 15:30:04 +0100, Maciej Stachowiak 
>>> On Feb 22, 2010, at 6:12 AM, Charles McCathieNevile wrote:
>>>> On Mon, 22 Feb 2010 13:44:37 +0100, Maciej Stachowiak

http://www.w3.org/html/wg/wiki/ChangeProposals/LongdescConformingWithWarning

>>>> Unlike Hixie's proposal, I think that this proposal can turn into
>>>> something I can support.
>>> 
>>> Do you think changes are needed for you to feel comfortable supporting
>>> it?
>> 
>> Fundamentally it depends what the warning *says*
> 
> I think that would mostly be for validators to determine, rather than 
> the spec. But the spec could give guidance. My proposal is that 
> validators should encourage authors to consider alternatives to 
> longdesc, most notably aria-describedby.
> 
> Do you have any other suggestions on what guidance the spec should 
> give on the warning?

The validator message should depend on how @longdesc is used:

1)	@longdesc="external-resource": 
No warning because there is no alternative (except rewriting the page - 
or pages!). Advice to simply replace @longdesc with @aria-describedby 
could lead authors to write aria-describedby="external-resource".

2)	@longdesc="#internal-resource": 
there could be encouragement to _in addition_ use aria-describedby. 
Reason to use both: not all user agents support aria and not all user 
agents that support aria support @longdesc either (judging from today's 
GUI web browsers). 

3)	Wrong use: If the <img> contains both an @aria-describedby="" and a 
@longdesc and the #fragment of the latter is not found amongst the 
idrefs of the former, then there could be a warning that the longdesc 
should probably be updated or removed. And ditto if the URL of the 
@longdesc obviously is no URL.

By the way, w.r.t. to the counter proposal:

	- it should define what @longdesc corresponds to in ARIA. Proposal:  
A) longdesc="#internal" corresponds to @aria-describedby with only one 
idref, B) longdesc="external" has no correspondence in ARIA.

	- @longdesc should be presented in section 3.2.6 (Annotations for 
assistive technology products)

PS: One could have said that when aria-describedby="internal" is used, 
then validators should encourage authors to _also_ use @longdesc, for 
compatibility.   If the spec doesn't say that, then that in itself 
represents an encouragement to use aria-describedby="internal" rather 
than longdesc="#internal"!!!
-- 
leif halvard silli

Received on Monday, 22 February 2010 19:21:34 UTC