Re: Getting beyond the ping pong match (was RE: Cleaning House)

Dão Gottwald wrote:

> Since the role attribute has never been part of HTML, I don't see how
>  leaving it out could be a retrograde step.

Well, rather as with GRIDDL, the standards that allow the use of role in 
HTML are not yet finalized:

http://www.w3.org/TR/grddl/

http://www.w3.org/WAI/PF/adaptable/HTML4/

> I have no knowledge of those applications, but they could obviously
> continue to work as they do today. 

It would be better to standardize how WAI roles work, given their 
widespread support from the AT community (Freedom Scientific, GW-Micro, 
NVDA, Sun/GNOME/Orca, Fire Vox, OatSoft).

> Furthermore they could start using the class attribute, if appropriate.

The draft I refer to above actually uses the class attribute like this:

<div class="axs checkbox">

Where axs is acting as a namespace (in the generic, non-XML sense).

For backwards compatibility, a script converts that into a role in the 
XHTML2 namespace. But there are major limitation to this technique:

http://www.w3.org/WAI/PF/adaptable/HTML4/embedding-20061212.html#limitations

The most cogent one is: "Each HTML element can only be associated with a 
single role." I haven't yet worked out why that limitation exists, however.

--
Benjamin Hawkes-Lewis