dual [sic] approach to ARIA embedding [was: Re: upcoming ARIA publicatoins, test coordination (ISSUE-14)]

On 23 Jan 2008, at 12:10 PM, Dan Connolly wrote:
> In the 10 Jan teleconference, I heard that the aria designs
> have been narrowed down to one, but nobody had a pointer
> to details.
> http://www.w3.org/2008/01/10-html-wg-minutes.html#item04

[Al, here]
My bad.  I was supposed to share the following summary with HTML WG:


First of all, thanks to all the groups that supported the
coordination session on 6 November.


At that session we agreed to report back in the context of the
Hypertext CG.

Thanks to your help, we have made significant progress on keeping the
WAI-ARIA work moving forward including dealing with
host-langauge-embedding issues.

This message summarizes the direction we are taking after reviewing
what we learned in the face-to-face sessions and through various
email threads before and after.

Clearly, if you find show-stoppers in this, we will want to reconsider.
But we think this is an approach that all parties can live with.

** basic plan

Note that the WAI EOWG reviewed our documents and made some
recommendations regarding how we package and present our work.
One of their recommendations that we expect to implement is to
combine the Roles spec and the States and Properties spec into a
single document.

We will discuss host language embedding in our new, combined
specification.  We will present two approaches, and give the two
approaches parallel treatment.

The two approaches will be:

* no namespaces, no prefixes

Pretty much as developed in the document at

XHTML2 Working Group has agreed to change the Role Attribute Module
so that the role attribute can be incorporated in other host
languages by mechanisms other than Namespaces in XML.


Note that under this approach, host languages reserve all attribute  
matching the pattern aria-*.  This allows the ARIA states and properties
to be controlled by the PFWG specification, including future  

.. this approach could be used in HTML5 and in extended DTDs or
otherwise-verifiable supersets of HTML4.

* by Namespaces in XML

The Role Attribute Module specification will give @role a 'home'
namespace from which it can be cited using the Namespaces in XML
Recommendation technique.

.. this approach could be used in DIAL and other applications of XHTML.
These dialects are likely to be used primarily to feed middleware, as
there is no immediate prospect of this technology being picked up
widely by browsers, soon.


The discussion in the Hypertext CG in December revealed no
show-stopping problems with this approach.

I personally don't understand remarks to the effect that
ARIA-embedding approaches "have been narrowed down to one"
but the dual approach notwithstanding, there seems
to be no problem for **HTML5, per_se** adopting the first one
and only the first one.

Life is not so simple for SVG.  But that's because they
want to swim in a middleware world as well as over the wire
to the browser.



> On Thu, 2008-01-17 at 20:22 +0000, Gregory J. Rosmaita wrote:
>> aloha!
>> 1) i have marked action 23 as "pending review" with the following
>> comments:
>> <q cite="http://www.w3.org/html/wg/tracker/actions/23">
>> 2008-01-17 19:41:44: this action is delayed due to the pending  
>> release of
>> new public drafts of the ARIA documents
> When do you expect them?
>>  which will render this action item
>> moot; a new action item will be opened by GJR when the new public  
>> drafts
>> are released and test pages are updated to reflect any changes in  
>> the new
>> drafts;
> Let's not bother with all that; let's just keep ACTION-23 updated  
> until
> the tests are agreed by both WGs.
>>  it is for this reason that i have renamed the action item
>> "coordinate tests using ARIA"
> Tweaking the label as we go is fine.
> Rich, I see you put "In the next release I want to see full ARIA
> support" in your response to the question about publishing
> the HTML 5 spec as a W3C working draft. Should I expect
> you at the teleconference tomorrow at noon boston time?
> I'm trying to figure out what priority to put on my
> own action in this area:
>  Show how <div aria="something"> works with URI based extensibility
>  http://www.w3.org/html/wg/tracker/actions/11
> In the 10 Jan teleconference, I heard that the aria designs
> have been narrowed down to one, but nobody had a pointer
> to details.
> http://www.w3.org/2008/01/10-html-wg-minutes.html#item04
> p.s. Gregory, please include more novel keywords in the subject
> and not just boilerplate/admin stuff like "action-nn updated".
> -- 
> Dan Connolly, W3C http://www.w3.org/People/Connolly/
> gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Wednesday, 23 January 2008 21:25:05 UTC