Re: Drop longdesc, get aria-describedat?

Janina Sajka, Wed, 7 Mar 2012 22:28:43 +0000:

> in a separate response to Silvia Pfeiffer I noted that we're
> considering use cases and requirements  [ snip ]

> So, writing an ARIA-DescribedAT should be considered an option to submit
> to PF. That's certainly acceptable.
> 
> However, this wouldn't produce a solution today, or even next month, and
> a11y has been waiting a long time.

The only thing that can be solved 'today', is formal conformance. You 
say you are collecting use cases. Fine. But actually we need those use 
cases *now*, in Laura's change proposal ...

> My druthers would be to accept longdesc right away and call it obsolete
> but conforming. That clearly signals that a replacement is expected
> while providing needed functionality right away--the same it has been
> available since html 4. As I said, this is my
> preference. Others may have other views.

'Obsolete but conforming' sounds good, to me too. But is not what 
Laura's change proposal says. It seems we eventually need yet another 
change proposal if we want that point made. Also, such a change 
proposal, that focused on @longdesc as the *current* solution, would 
probably look different from the CP we have.

Even in case the issue would be taken to the Director, after a Formal 
Objection, it would probably seem less provocative if he decided 
@longdesc's fate to be 'obsolete but conforming' - compared with 
'conforming'. And also, in case of a FO, the grounds on which one 
objects, ought to have been tried within the group before complaining 
to the Director that we did not get it?

Of course, last vote, 'obsolete but conforming' went straight out. But 
perhaps one could argue better for it now.

> Lastly, I would agree that HTML could also craft a mechanism to serve
> the need in a superior fashion. There's nothing wrong in doing that, but
> that would take some time, certainly. Too bad so much time has been
> wasted trying to make describedby fit where it just wasn't going to fit.

The focus on aria-describedby is actually a focus on <a href>: One 
wants describedby to point to a link. Which, in turn requires AT to 
treat the link as an interactive element.

> And, every day that passes with longdesc shoved aside is another day
> when this core requirement is unmet in the spec.

An important detail, for the credibility of that argument, is that even 
if we do get @longdesc, the requirement is *only met* for the <img> 
element. For other elements in need of a long descriptionlink, it is 
not met. The chairs have, as much as I remember, asked why only <img> 
needs @longdesc - so it is important to have an answer for that.

> As previously noted, it
> hasn't been met for half a decade. That's a long time especially for a
> "living" specification that supposedly so responsive to needs.

Hm, yeah. HTML5 is starting to get old. <smile>. But HTML4 and XHTML1 
still what the MIME registration describe as HTML ... And one could 
also say that half a decade is a long time where aria-describedAT did 
not happen.

> I know you're trying to find a solution, Leif. I'm doing my best to be
> as helpful as I can.

We need to say what we really think to make progress. So thank you for 
that.
-- 
Leif H Silli

Received on Wednesday, 7 March 2012 23:21:36 UTC