[html4all] Fwd: longdesc questions

[Joshue O Connor]

Hi Gez,

> The labelledby and describedby properties are part of the states and
> properties module [1]. They are included with a group of properties
> that can be used to describe live regions, but aren't restricted to
> them, as they can be used to describe/label anything. 

Thanks for clearing that up.

>  I like the
> versatility, so think they would make a good generic replacement for
> specific attributes, such as 'for', 'longdesc', 'headers', and so on.

I also think its a good idea. For several reasons, ease of use by
authors and as we spoke of in another mail the fact that AT can already
handle this kind of content as is already doing so using the labeled for
attribute etc.

> I'm not sure what you mean about WAI-ARIA being adopted by HTML5.

These attributes are as a result of ARIA states and roles [1] so in
order for these attributes to be included in the HTML 5 spec there is
therefore a necessary adoption of some aspects of these states and roles
modules. Your next sentence clears this up. I would however rather see
them included in the spec rather than being bolted on. This may make it
easier for authors also. Isn't it a case that by using a specific prefix
for the ARIA attributes they are not natively a part of the specification?

>  If
> they don't include the properties (either natively or using a vendor
> specific prefix for ARIA attributes), then it's less likely to work
> (although people could still kludge it in the same way they are doing
> so now to get it to work with HTLM 4).

Yes, so it is rather inelegant.

> plenty of others that could be included natively, so
> that people can write semantic markup with strong structural
> relationships that can be exposed to AT, rather than doing this with
> semantically neutral elements.

Sounds good to me. However, are the labelledby and describedby
properties not semantically neutral themselves, hence their ability to
be context dependent?

Cheers

Josh

[1] http://www.w3.org/TR/aria-state/