W3C home > Mailing lists > Public > wai-xtech@w3.org > September 2009

[aapi] RE: picking up on some comments in IRC on ARIA

From: Cynthia Shelly <cyns@exchange.microsoft.com>
Date: Wed, 2 Sep 2009 19:57:26 +0000
To: Maciej Stachowiak <mjs@apple.com>, David Bolter <david.bolter@gmail.com>
CC: Steven Faulkner <faulkner.steve@gmail.com>, HTMLWG WG <public-html@w3.org>, W3C WAI-XTECH <wai-xtech@w3.org>
Message-ID: <F8D41EF9671C524696D1930053C122912237FDF3@DF-M14-05.exchange.corp.microsoft.com>
I agree with David that it would be very valuable to have someone from Apple involved in documenting the HTML to API mappings.  It may be the case that it's not practical to have direct mappings from element to API.  I personally think that it would be much simpler to understand and use for web developers if that were possible, and that it should be a goal.  However, even if that is not possible, it is critical that the mappings be consistent across browsers on any given platform. 

In the current situation, with AT products going directly to the DOM and having widely divergent user experiences, Web application developers who wish to do accessible development find themselves with a truly unmanageable test matrix of (platform * browser * AT), where there are multiple AT products in different countries and where many of them are quite expensive.  Having a consistent DOM to API mapping would pave the way for AT products to use the APIs rather than the DOM more, and give those that do use the DOM a better idea of how certain types of HTML should behave. It will also give users an experience that is more consistent with the native applications on their platform, and more consistent across web applications. The greatest benefit is to developers, however, who will be able to test against the APIs at runtime (using existing platform tools) and, because the relationship between element and API is known, will be able to a fair bit of testing using validation and other static analysis mechanisms.

Maciej, your input on this work would be very valuable.  The AAPI call is currently at 8am Pacific on Fridays, but needs to move due to some conflicts.  We've been looking at the same time on Tuesdays to accommodate Steve in the UK and David (Eastern, I think).  I know you have trouble with early calls (as do I).  Is there time that would work for you?

-----Original Message-----
From: wai-xtech-request@w3.org [mailto:wai-xtech-request@w3.org] On Behalf Of Maciej Stachowiak
Sent: Wednesday, September 02, 2009 12:38 PM
To: David Bolter
Cc: Steven Faulkner; HTMLWG WG; W3C WAI-XTECH
Subject: Re: picking up on some comments in IRC on ARIA


On Sep 2, 2009, at 7:51 AM, David Bolter wrote:

> On 02/09/09 5:57 AM, Maciej Stachowiak wrote:
>>
>> On Sep 2, 2009, at 2:47 AM, Steven Faulkner wrote:
>>
>>>
>>>
>>> <othermaciej> I'm not sure a strict mapping to accessibility APIs  
>>> makes sense, because it would make it impossible to put any novel  
>>> and clever heuristics on the UA side instead of the AT side --
>>> Can you explain this further? if the UI is not mapped to  
>>> accessibility APIs Assistive technology has to pull this info from  
>>> the DOM, which is something you suggested previously in the alt=""  
>>> vs role="presentation" discussion was not desirable for voiceover.
>>
>> What I mean is, I don't think it makes sense to define a mandatory  
>> standard mapping to accessibility APIs for all possible HTML  
>> elements and attributes.
>>
>> I do think that browser should communicate with AT by mapping to  
>> accessibility APIs (and that's exclusively the way Safari/WebKit  
>> talks to VoiceOver).
>>
>> But the browser should be free to implement heuristics for poorly  
>> marked up content on the browser side, so the API mappings can't be  
>> mandated by spec. In particular, when using Safari with VoiceOver,  
>> some of the heuristics are implemented on the WebKit side before  
>> mapping to the accessibility API. That lets us put less browser- 
>> specific logic in VoiceOver. But to do that, we need freedom on how  
>> exactly we map particular markup to the accessibility API.
>
> We do this in Firefox as well. We look at native markup and ARIA,  
> and ultimately what gets exposed to platform API is based on  
> heuristics. Maciej, it would be great to have you or someone else  
> from Apple involved in our efforts to make sure we use similar  
> heuristics. Please see: http://www.w3.org/WAI/PF/aria-ua-task-force
>
> We're trying to make this all work with the least amount of pain for  
> AT and web developers.

I don't have the bandwidth for this, but I do regularly encourage  
anyone at Apple who is working on ARIA to participate, as their time  
permits.

Regards,
Maciej
Received on Wednesday, 2 September 2009 19:58:10 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:16:06 GMT