Re: [AAPI] role=""

Hi all,

For the user agent guide we are building a full table of ARIA role, and 
HTML element mappings. I think we might want an additional column for 
role->state mapping rules; but at least for now we will capture this in 
the text. I agree on not removing the readonly state for roles with no 
native platform mapping as a rule of thumb. It sounds like we have 
consensus on treating an empty role attribute as an absent role 
attribute; we'll go ahead and push this change in FF. Note: it really 
shouldn't happen often and is an author error; barely worth discussing 
but thanks all for the feedback!

cheers,
David

On 20/08/09 10:21 PM, Alexander Surkov wrote:
> Btw, I didn't find any allusion in ARIA implementation guide to the
> rule "remove readonly state bit coming from native markup once ARIA
> role is used and there is no aria-readonly attribute" we follow in
> Firefox implementation. This rule might have sense but it sounds it's
> reasonable not to spread it to roles having no mapping to AT API role.
> Concerning initial question I think empty role should be treated as
> role attribute would be absent.
>
> Alex.
>
>
> On Fri, Aug 21, 2009 at 2:13 AM, David Bolter<david.bolter@gmail.com>  wrote:
>    
>> Sorry to reply to myself.  I feel compelled to connect more dots here.  What
>> generally happens is that for unknown roles (e.g. "", "gargleblaster"),
>> Firefox does not expose the state readonly (assuming interactivity), which
>> is a state at least one screen reader uses to decide whether or not to
>> create a virtual buffer (create if readonly).
>>
>> cheers,
>> D
>>
>>
>> On 8/20/09 1:58 PM, David Bolter wrote:
>>      
>>> Hi Victor,
>>>
>>> Hahaha! Seriously though, I guess the main thing is whether the browser
>>> should map it to something like an IA2_ROLE_UNKNOWN or not. If we take the
>>> example brought to our attention by James Teh, if an author were to do this:
>>>
>>> <body role="">
>>>
>>> If we mapped this to a role on the native platform side, the AT would
>>> probably not create a virtual buffer, assuming that it is an interactive
>>> element, which is probably not the case.
>>>
>>> cheers,
>>> David
>>>
>>> On 8/20/09 1:50 PM, Victor Tsaran wrote:
>>>        
>>>> Hmmm, wouldn't that be the case anyway? What happens now?
>>>> We could also define a role of "" whose purpose it would be not to do
>>>> anything.<LOL>
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: wai-xtech-request@w3.org [mailto:wai-xtech-request@w3.org] On
>>>> Behalf Of Benjamin Hawkes-Lewis
>>>> Sent: Thursday, August 20, 2009 10:40 AM
>>>> To: David Bolter
>>>> Cc: wai-xtech@w3.org
>>>> Subject: Re: [AAPI] role=""
>>>>
>>>> On 20/08/2009 18:36, David Bolter wrote:
>>>>          
>>>>> If an author (strangely) includes a role="", I'm thinking we can treat
>>>>> it the same as if the role is not specified at all. Thoughts?
>>>>>            
>>>> Makes sense to me.
>>>>
>>>> Would be nice if the WAI-ARIA spec defined that behavior. :)
>>>>
>>>> --
>>>> Benjamin Hawkes-Lewis
>>>>
>>>>          
>>>
>>>        
>>
>>
>>

Received on Tuesday, 25 August 2009 18:21:25 UTC