- From: Joseph Scheuhammer <clown@alum.mit.edu>
- Date: Fri, 20 Apr 2012 11:59:19 -0400
- To: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
- CC: alexander Surkov <surkov.alexander@gmail.com>, wai-xtech <wai-xtech@w3.org>
Hi Leif,
You wrote:
> Alexander or someone,
>
> is it possible to test Firefox' longdesc implementation somewhere,
> somehow? I mean, not via some a11y inspector but in a 'real' way?
Only a partial answer: Since FF's job is to take the information in the
markup, and publish relevant aspects of it to the a11y API, FF's job is
done when it has so published. In the case of @longdesc, FF represents
the information as an action in the a11y API, specifically, an action to
go to the url provided.
The client of the a11y API are ATs, and it's up to them to use the
information in the API. Thus, if you don't want to use an a11y
inspector, the other way to test it is to launch some AT and see what it
does with the information. ATs include screen readers, onscreen
keyboards, voice control systems, among others; and whether any
particular AT makes use of this specific a11y action is up to that AT.
To see if they do, and how, you'd have to consult their documentation,
or find an expert user.
To satisfy my own curiosity, I used the Linux a11y inspector,
Accerciser. For an <img> with a @longdesc, Accerciser shows a
"showLongDesc" action and provides a "Perform Action" button. Clicking
that button causes FF to open the page referenced by the url in a new
tab. That implies that *if* an AT invoked this a11y action, the result
would be the same.
I also looked into what Orca does in this case, including a brief
discussion with some of the developers. Orca ignores the action, or
perhaps more accurately, didn't even know it was available.
Hope that's useful.
--
;;;;joseph.
'A: After all, it isn't rocket science.'
'K: Right. It's merely computer science.'
- J. D. Klaun -
Received on Friday, 20 April 2012 16:00:00 UTC