Re: @longdesc implemented as ARIA construct

James Craig, Wed, 15 Sep 2010 09:54:38 -0700:
> On Sep 15, 2010, at 6:06 AM, Leif Halvard Silli wrote:
> 
>> Dear WAI-XTECH listees! Is this the right place to propose changes to 
>> ARIA?
> 
> Here is fine, but if you'd like to make a formal suggestion, email 
> public-pfwg-comments@w3.org

OK. Is there  not a bug tracker?
 
>> I happen to be of the belief it would strengthen the case for @longdesc 
>> in HTML5 if it was demonstrated how one can do the same thing via ARIA. 
>> Perhaps, out of this debate, we could even agree to change ARIA to make 
>> it better suited for such a thing? 
> 
> I can't speak for all members of the PFWG, but I personally agree 
> that @longdesc should be removed from HTML5, and I know that several 
> other PF member representatives feel the same way. I don't feel 
> strongly enough about it to start another holy war on this list, but 
> suffice it to say that I also don't feel that we should help 
> strengthen the case for @longdesc. Save your strength and fight a 
> more important battle.

Strictly speaking, the most important thing is that it is possible to 
link from an role="img" to a long description. And I personally see 
many advantages to a anchor element solution.

   Some of the @longdesc supporters (Laura, I'm looking at you) seems 
to see ARIA solutions as a hack. And I think that some of them don't 
want to look at ARIA for fear that it could be shown possible to use 
ARIA.

  But I have to warn you, ARIA people, as well: if you add the 
possibilithy of a link inside role="img", then you _could_ in fact add 
a argument in favor of @longdesc. I hope that that will not prevent you 
from looking at the idea I presented above?

I described this in more detail here: 
http://lists.w3.org/Archives/Public/public-html-a11y/2010Sep/0429

Simply put: Apart from the ability to providing AT a link to a long 
description, I also see the proposal above as a principal issue. 
Namely: role="img" appears to be modeled after how <img> works. Thus it 
is a failur to implement <img> as role="img" that you haven't sofar 
allowed a long description link inside role="img" elements. 

I am sure that *if* you add ability to link from within role="img" to 
long description, then you would please many and certainly add an 
argumetn in favor of deprecating @longdesc. ARIA could even describe 
that authors MAY use @longdesc, but are advice to use this rel=longdesc 
link feature that I suggest. If a role="img" element has both 
featueres, then ARIA should tell AT to use only one of them, for 
instance.
-- 
leif halvard silli

Received on Wednesday, 15 September 2010 17:15:55 UTC