Re: [aria impl] platform mappings, and role vs state

Joseph Scheuhammer wrote:
> Hi David,
>
>> For example ...
>> <span role="checkbox" aria-checked="undefined">
>>
>> The role tells us that the object is checkable, and
>> the aria-checked="undefined" would normally indicate the object is not
>> checkable.
>
> This doesn't answer your question (giving the role the power to
> override the author-declared state); rather, noting that the situation
> as described is self-contradictory.  This is something a checking tool
> should detect and complain about.
>
> A way to look at it is why an author would intentionally write the
> markup as in your example?

Exactly.  So in trying to guess the user's intention... so as a rule of
thumb, we could just go with role over attribute since we gotta pick
something.  Thoughts on that?

>   Perhaps they think it's the way to declare a disabled checkbox (a
> checkbox that is uncheckable).  But, the markup in that case should be:
> <span role="checkbox" aria-enabled="false" aria-checked="false">.
>

Yes.

cheers,
David