Discussing ARIA in HTML5 integration

Al -- Could you provide some clarification?

As best I understood it, HTML is embedding ARIA -- which means HTML is
a generic user, and -- strictly speaking -- no coordination is
actually needed.  (Except that the HTML draft can't become a
recommendation until there is a normative version of aria to which to

In practice, the ARIA authors might profit from feedback from a major
user, and the HTML group should benefit from advice.  But no
coordination is actually *required*.

All that MUST happen is that the HTML group defines a mapping from the
HTML and XHTML serializations and DOM to the ARIA infospace.  The DOM
SHOULD look like what the ARIA spec defines (if it defines DOM
bindings), and the serializations (particularly XHTML) SHOULD be
something compatible with other embeddings -- but even this isn't a

My understanding was that even the SHOULD had been met by establishing
that aria-* attributes would be controlled by the ARIA spec.

You suggest otherwise by writing:
> Neither the published plans of the two
> working groups nor our intuition as...

So what does still need to be done for your first goal, of supporting
@role and tolerating @aria-*?

The second goal (mapping from html itself, instead of requiring author
intervention) also seems like something that the HTML group must do,
except that they may well misunderstand the aria-* attributes.  But
where the HTML group misunderstands, I would assume that individual
authors (and library authors) would have just as much trouble -- so it
would be a high priority for clarification.


Al Gilman wrote:

> ... the following two milestones:

> (first) HTML supporting an @role attribute and
> tolerating @aria-*  attributes defined per the
> PFWG charter in HTML processing so as to
> support an HTML+ARIA  profile of markup.

> (second) taking functionality that is presented
> in a more factored,  attribute-based form in
> the WAI-ARIA layer and enabling the same
> functionality through native features in HTML5.

> We have received some public comments
> that sound as though the second were as
> near-term as the first. Neither the published
> plans of the two working groups nor our
> intuition as software architects suggests
> this to be the case.

> We will continue to be diligent in overtly
> sharing all information necessary for
> the first milestone above with the HTML
> WG, i.e. in public.

> But we don't want to see scope creep in
> that dialog to try to address the second
> at the same time.

Received on Wednesday, 9 April 2008 23:50:21 UTC