Re: [aapi] ARIAAttrModified?

Yeah.

This is all about http://www.w3.org/TR/wai-aria-practices/#aria-write

Maybe someone who is a big supporter of this part of the WAI-ARIA BP can
jump in with the best use cases?  Then we can look at cases.

Today web app developers rarely (if ever) listen for DOM mutation
events. I suspect part of the reason is the ability to get information
in other ways. Polling is one (yuck). In the case of an html checkbox,
they are far more likely to handle onclick of course, than pay attention
to "checked" mutation.  I suspect, the other reason for rarely listening
to DOM mutation events is poor support in some browsers. I suspect the
poor support is due to a performance strategy. These latter assumptions
are leading me to tabling this ARIA mutation idea (not mine).

I know, a lot of assumptions.

The biggest worry for me is, if we go through all the effort of creating
a mechanism for web app developers to know when an aria attribute is
modified, will they use it?

cheers,
David



Alexander Surkov wrote:
> Hello.
>
> On one hand I'm not sure semantic extension is a good way to get rid
> performance problems. On another hand I'm not sure it can be in real
> life when application (web page) author needs to listen ARIA changed
> but he doesn't need to listen DOM changes, I mean both ARIA and DOM
> mutation changes should be used in the same time.
>
>
> Alex.
>
>
> On Fri, Mar 6, 2009 at 12:53 AM, David Bolter <david.bolter@utoronto.ca> wrote:
>   
>> Hi All,
>>
>> During some discussion over IRC yesterday Olli Pettay suggested that a
>> performant way of implementing DOM attribute mutations might be to have
>> separate but similar API to DOMAttrModified.
>>
>> Note DOMAttrModified would serve the same purpose, but having the
>> browser create DOM mutation events for all mutations is going to slow
>> the web experience.
>>
>> ARIAAttrModified is tabled as a way of only creating mutation events for
>> aria attributes.
>>
>> Background: why do we need these mutations? In the case that an
>> aria-attribute is modified, not by the web developer (perhaps through
>> platform API), for
>> example aria-sort is set to "true", the web developer is supposed to
>> respond to that change (a dom attr mutation), and put his web
>> application into a state that corresponds with the aria-sort="true"
>> semantic.
>>
>> (see http://www.w3.org/Bugs/Public/show_bug.cgi?id=6653)
>>
>> If this is something we think is worthwhile, we should work with the
>> DOM3 group to make it happen.
>>
>> cheers,
>> David
>>
>>
>>

Received on Thursday, 5 March 2009 21:31:03 UTC