Re: UAIG requires checkable object attribute for aria-pressed

NVDA says "checked" or "unchecked" when it sees "checkable" object
attribute on the object. If that was a checkbox or checkable menuitem
then it works perfect. But it seems confusing in case of toggle
button. I think NVDA can workaround that but the point was whether
"checkable" is a good description for toggle button behavior because
toggle button doesn't seem as something that can be checked or
unchecked, it's rather pressed or not pressed.

Basically toggle button is exposed as a toggle button role (compare to
pushbutton role used by ordinal buttons) and it seems AT don't need
any extra info to handle it.

Thank you.
Alex.


On Fri, Jan 4, 2013 at 7:05 AM, Joseph Scheuhammer <clown@alum.mit.edu> wrote:
>> More notes -- pressed, selected and expanded are similar in that the lack
>> of them indicate an item is not pressable/selectable/expandable.
>
>
> I think that's a key point.
>
> Paraphrasing from the UAIG:
>
> - If aria-pressed="true", then IA2 and AT-SPI expose a toggle button role
> and set STATE_PRESSED.
>
> - If aria-pressed="false", then expose a toggle button role and *clear*
> STATE_PRESSED.
>
> Note that in the latter case, the state is cleared.  There is no "pressed =
> false" in IA2 nor AT-SPI.  STATE_PRESSED is like a bit in a bit vector.
> Thus, an AT that looks only at the states can not tell that the accessible
> could be pressed if the state is cleared. Hence, the additional information
> provided by the object property checkable:true indicates that the accessible
> can change its state. At least, that's my best guess at a rationale for
> storing that object property in the accessible.  (Next question -- why not
> pressable:true? or is that not available in IA2 or AT-SPI?)
>
> Still, it means that an AT has to infer that "checkable" means "can be
> pressed" in this specific instance, and not "can be checked". And, the way
> to determine that is by looking at the role -- it's a toggle button.  (But,
> if it's a toggle button, doesn't that entail that it's pressable?)
>
> In any case, that's my stab at what the rationale was for adding the
> checkable object property.  But, it's pure speculation and inference on my
> part.
>
>
> --
> ;;;;joseph.
>
>
> 'A: After all, it isn't rocket science.'
> 'K: Right. It's merely computer science.'
>              - J. D. Klaun -
>
>

Received on Saturday, 5 January 2013 05:29:58 UTC