Re: aria-activedescendant changes to the aria-implementation guide

Hi Joseph,

Thanks for the quick reply. Just thinking aloud here:

The way" doing the default action" generally works on Firefox is by
simulating a mouse click, allowing a lot of cases in the wild to "just
work", including aria-activedescendant patterns. Attribute changes are
indirect, and related to the web developers handling of click. Note: I think
the fact that the spec you mention talks about AT changing attributes might
confuse some.

I think an AT user purposely invoking doDefaultAction is a simple case, and
user expectations will usually be correct. I think the user expectations for
attempting to set focus are different. If a user tries to programmatically
set focus to a checkbox they don't want to actually check/uncheck it do
they?  (Not sure if you were implying that)

Cheers,
David

On Thu, Apr 28, 2011 at 1:06 PM, Joseph Scheuhammer <clown@alum.mit.edu>wrote:

> Hi David,
>
> You wrote:
> > The fact "there is no reliable mechanism to notify" a web
> > "application" that the aria-activedescendant has changed is not enough
> > of a reason to further complicate ARIA for authors ...
>
> I don't see how it complicates things for authors.  There being no way
> to be aware of changes is quite liberating.  It means that authors need
> not (can't?) look for them.  Nothing for authors to do here.  And, it's
> likely they don't bother looking anyway.  Didn't you point that out 100
> years ago? :-)
>
> >  ... and browser implementations.
>
> I have no opinion here (well, not much of one), and defer to you.
>
> I gather your suggestion is to recommend that authors not use
> aria-activedescendant, ever?  Could that be tempered to:  don't use
> aria-activedescendant until such time as there is a way to notify
> scripts of externally made changes.  There being "no reliable mechanism
> to notify a web application of changes" is the crux of the problem, and
> it's larger than just aria-activedescendant.
>
> Somewhat related:  the spec says [1]:
>
> > When a web application maintains a local representation of
> > accessibility information through WAI-ARIA roles, states, and
> > properties, the user agent MUST provide a method to notify the web
> > application when a change occurs to any of the related states or
> > properties in the system accessibility API.
>
> It goes on to suggest that "... Many state and properties can be changed
> by assistive technologies through existing accessibility APIs by
> responding to a default action event."  Would that work here for
> aria-activedescendant, without being overly complicated?  Using a
> checkbox as an example, the user agent wouldn't directly modify a that
> checkbox's aria-checked value.  It would instead tell the checkbox to do
> its default action, which would lead to the associated script
> "naturally" updating checkbox's aria-checked state.
>
> > Joseph, remember talking about this 100 years ago ;)
>
> Ah memories.  The good 'ol days. ;-)
>
>
> [1] http://www.w3.org/WAI/PF/aria/conformance#ua_domchanges
>
> --
> ;;;;joseph
>
> 'I had some dreams, they were clowns in my coffee. Clowns in my coffee.'
>                     - C. Simon (misheard lyric) -
>
>