Re: Revert request

On 3/13/12 3:05 PM, Sam Ruby wrote:
> As to aria-describedAt, we have statements from Judy and Janina, but a 
> consensus call from the a11y TF would be helpful here.  This is a case 
> where the long period of time since the issue was originally opened 
> will be relevant.  People can argue that someday somebody might spec 
> something that is better/more likely to be supported/etc., and even 
> believe that there are plausible paths by which such could be 
> accomplished -- if nobody actually does that work, that result speaks 
> for itself. 

I'd wager that we can gain consensus that by at least ARIA2, there will 
be sufficient vocabulary to implement longdesc features. There will 
likely be so much vocabulary as to supersede the features of longdesc. 
But, that's ARIA, not HTML.

Many ARIA semantics have counterparts in HTML. That ARIA will have 
superior ways to describe longdesc semantics is not on its face, reason 
to deprecate the longdesc attribute from HTML.

The question is, does HTML have any features in store to replace 
longdesc? While the "rel" semantic is an option, it has not become a 
viable, it shows no signs of becoming viable. That's where we ought to 
slow down this marry-go-round and get off.

Conflating DOM ARIA and Semantic HTML is hazardous to the health of AT 
support. ARIA has healthy boundaries, they should be guarded.

If it would help to catalog cases where there is redundancy between ARIA 
and HTML, I'm happy to do so. I consider those cases a good thing, not a 
bad one. ARIA and HTML allow us to work across a much broader range of 
UAs (primary and secondary) than either individually.

Something like the following semantic would be perfectly in line with 
the rest of the Semantic HTML+ARIA cross-types:
<img alt="My image description" longdesc="imagedescription.html" 
aria-describedat="imagedescription.html" />

In doing so, we'd have a mapping that @longdesc would map to 
@aria-describedat provided aria-describedat is not specified.

Thus the above could be shortened to:
<img alt="My image description" longdesc="imagedescription.html" />

That's very typical.

If aria-describedat had any sort of distinction or additional semantic, 
it might even point to a different location than the longdesc would. 
Additionally, redundancy might help readers as they cycle through 
aria-describedat labeled items.


-Charles

Received on Tuesday, 13 March 2012 22:22:06 UTC