Re: Need to discuss: normative spec for exposing ARIA via APIs

        This part wasn't logged ... commitments to work on the various 
specs proposed:
 
1. HTML+ARIA to DOM API  : commitment by Aaron Leventhal (IBM) and Jonas 
Klink (Google)
2. HTML DOM (with ARIA) to MSAA/IA2   : commitment by Aaron Leventhal 
(IBM) and Jonas Klink (Google)
3. HTML DOM (with ARIA) to ATK/ATSPI     : commitment by Aaron Leventhal 
(IBM) and Jonas Klink (Google)
4. HTML DOM (with ARIA) to MSAA/UIA Express   : no commitments yet
5. HTML DOM (with ARIA) to Mac OSXA    : no commitments yet

Any commitments to work on these specs from some other ARIA implementors?

- Aaron



From:
Richard Schwerdtfeger/Austin/IBM@IBMUS
To:
Michael Cooper <cooper@w3.org>, "wai-xtech@w3.org" <wai-xtech@w3.org>
Cc:
Aaron M Leventhal/Cambridge/IBM@IBMUS, Alfred.S.Gilman@ieee.org, 
alice.liu@apple.com, annevk@opera.com, Beth Dakin <bdakin@apple.com>, 
hsivonen@iki.fi, James Craig <jcraig@apple.com>, Jonas Klink 
<klink@google.com>, marcsil@windows.microsoft.com, simonp@opera.com, Tony 
Ross <tross@microsoft.com>
Date:
10/29/2008 11:04 PM
Subject:
Re: Need to discuss: normative spec for exposing ARIA via APIs



Michael raises an important point. I am replying to Michael and xtech to 
get the discussion logged and we continue the discussion.

Rich


Rich Schwerdtfeger
Distinguished Engineer, SWG Accessibility Architect/Strategist
blog: http://www.ibm.com/developerworks/blogs/page/schwer

Michael Cooper ---10/29/2008 04:06:12 PM---I'm getting a little concerned 
that ideas are maturing in this discussion, but not all the right people 
are necessarily at the

Michael Cooper <cooper@w3.org> 
10/29/2008 04:05 PM



To

Richard Schwerdtfeger/Austin/IBM@IBMUS

cc

Jonas Klink <klink@google.com>, Aaron M Leventhal/Cambridge/IBM@IBMUS, 
Alfred.S.Gilman@ieee.org, alice.liu@apple.com, annevk@opera.com, Beth 
Dakin <bdakin@apple.com>, hsivonen@iki.fi, James Craig <jcraig@apple.com>, 
marcsil@windows.microsoft.com, simonp@opera.com, Tony Ross 
<tross@microsoft.com>

Subject

Re: Need to discuss: normative spec for exposing ARIA via APIs





I'm getting a little concerned that ideas are maturing in this discussion, 
but not all the right people are necessarily at the table yet. This thread 
isn't even taking place on an archived list. As I mentioned in an earlier 
reply, there was some agreement on directions related to this documented 
at http://www.w3.org/2008/10/24-role-minutes.html, and all the people 
involved in that discussion would want to participate in this one as well.

Could I ask that this discussion be moved to the wai-xtech@w3.org list? 
We'll need a recap of what's been discussed so far. If you prefix your 
subject line with "[Role]" it will help establish continuity with the 
discussion that took place in the HTML breakout session last week. Anybody 
on the cc list to this message who is not on the xtech distribution, 
please let me know and I'll add you.

Again, I'm not able to offer substantive help with this process until next 
week. But if someone could look after the above I think it will help us to 
be in a better place for moving forward.

Michael

Richard Schwerdtfeger wrote: 
That is great. This is looking like a normative spec which defines:

HTML+ARIA to DOM API
HTML DOM (with ARIA) to MSAA/IA2
HTML DOM (with ARIA) to ATK/ATSPI
HTML DOM (with ARIA) to MSAA/UIA Express
HTML DOM (with ARIA) to Mac OSXA

This would include how events are generated on each platform. 

Unless anyone here believes one of these is not ready. If not, we could do 
a second revision of the spec. when they are. 

I would not try to tackle SVG as we have no proven solution. The SVG 
working group may be upset but we could say that this can be done later.

Michael, how many platforms do we actually need to support in the first 
implementation guide? I don't want to try and boil an ocean if someone is 
not ready.

Rich


Rich Schwerdtfeger
Distinguished Engineer, SWG Accessibility Architect/Strategist
blog: http://www.ibm.com/developerworks/blogs/page/schwer

"Jonas Klink" ---10/29/2008 12:34:07 PM---I personally (and Google) see no 
reason why we should not be able to move forward with describing the 
deliverables at this poin
"Jonas Klink" <klink@google.com> 
10/29/2008 12:33 PM



To

"Michael Cooper" <cooper@w3.org>

cc

Aaron M Leventhal/Cambridge/IBM@IBMUS, Alfred.S.Gilman@ieee.org, 
marcsil@windows.microsoft.com, Richard Schwerdtfeger/Austin/IBM@IBMUS, 
simonp@opera.com, hsivonen@iki.fi, "Beth Dakin" <bdakin@apple.com>, 
alice.liu@apple.com, "James Craig" <jcraig@apple.com>, annevk@opera.com, 
"Tony Ross" <tross@microsoft.com>

Subject

Re: Need to discuss: normative spec for exposing ARIA via APIs





I personally (and Google) see no reason why we should not be able to move 
forward with describing the deliverables at this point. Even though we are 
not implementation-wise quite there yet, I do think we have a clear 
picture of where this is heading.

- Jonas

On 10/28/08, Michael Cooper <cooper@w3.org> wrote: 
FYI the HTML WG had a breakout session on a topic very much the same as 
this. While we didn't take careful notes, what we have is at 
http://www.w3.org/2008/10/24-role-minutes.html.

My feeling coming out of that session was that we had some additional 
ideas for next directions. The issue of people who can represent the 
various accessibility APIs remains.

My own involvement in this process is restricted for this week, due to 
conflicting responsibilities. I plan to come back to this next week and 
hopefully offer substantive support from there on.

Michael


Aaron M Leventhal wrote: 

What's the next step to bring this into the W3C? 

Apple, Opera and Google have indicated they are not far enough along in 
implementation to make deep contributions to such a spec. As I see it, 
that leaves Microsoft and Mozilla as the two main drivers. 

Unfortunately, Microsoft and Mozilla use different APIs to extend beyond 
MSAA. By itself is only enough to support ARIA widgets. Mozilla uses 
IAccessible2, and Microsoft uses UIA Express (correct?) to get beyond 
these limitations. We can standardize on things like object and text 
attributes used to expose ARIA properties, but there will be some places 
where API specifics will differ. 

On Linux, I'm not sure if there is a browser vendor outside of Mozilla 
very focused on supporting ARIA via ATK/AT-SPI. However, it is at least 
very close to IAccessible2. 

And, if we want to have anything meaningful for how ARIA is exposed on a 
Mac, we will need input from Apple, Google Opera, who have expressed that 
there is no one full time working on ARIA support, and that it is not 
clear when they'll be ready to solidly contribute this information. 

Here's a chart that describes my understanding of where each vendor is 
focusing in efforts to expose ARIA: 
Mozilla: MSAA+IA2, ATK, OS X Accessibility Protocol 
Microsoft: MSAA + UIA Express 
Google/Apple/WebKit: OS X Accessibility Protocol, MSAA(+IA2?), ATK? 
Opera: OS X Accessibility Protocol, MSAA(+IA2?), ATK? 

I'm sure we can work out a lot of issues as we go through the doc, but I'm 
not sure what kind of commitment can be made. In the current environment 
it doesn't look like we can do a satisfactory job of describing the 
mapping to all APIs. 

Can we describe the deliverables so that we can be successful even where 
API specifics aren't ready? 
Or do the individual API specifics need to be both normative and complete 
for every platform? I don't see why they do. 

- Aaron 

-- 
Michael Cooper
Web Accessibility Specialist
World Wide Web Consortium, Web Accessibility Initiative
E-mail cooper@w3.org
Information Page

-- 
Michael Cooper
Web Accessibility Specialist
World Wide Web Consortium, Web Accessibility Initiative
E-mail cooper@w3.org
Information Page[attachment "pic08967.gif" deleted by Aaron M 
Leventhal/Cambridge/IBM]