[webkit-dev] Changing two unit tests: 'alt' as accessible name instead of description

[Chris Fleizach]

I don't think you want to return the altTag in stringValue(). That  
should only be for things that have actual strings (textfields, text  
rendered onto the screen)

Right now titleAttr is exposed for all elements in the ::title()  
method. Perhaps that is incorrect, but its hard to get this right,  
since so many developers
use title and alt incorrectly.

On the Mac side, AT software is able to get the titleAttr through an  
AXTitle attribute and the altAttr through AXDescription

AXDescription is supposed to be the description of the object that  
when that description is not displayed on the screen. This is  
basically what accessibilityDescription() does (which also
explains why altAttr is being returned in accessibilityDescription())

So in my naivety, it seems Gtk should exposed the ::title() method  
through some attribute, so that the AT program can take the title and  
the description and decide what to do with it.

On the mac side, there is also an attribute called AXHelp, which is  
more inline with what W3 says on how title should be used (as a  
tooltip). AXHelp thus does return the titleAttr in webkit.

Unfortunately, due to the general uselessness of AXHelp and tooltips,  
it is not something that is often heard by mac AT users.

Combine that with the misusage of the title attribute to label an  
element, creates a situation where the AT user would not hear the  
information the developer adorned their elements
with, in most cases.

Thus, titleAttr is returned as the AXTitle on the mac.


> Hi,
>
> As the result of working in bug 25524 (see [1] for more details), we
> have reached the conclusion (and not only to fix that bug) that  
> perhaps
> it would be good to change how the 'alt' attribute is exposed to ATs,
> switching from being the accessible description to be the accessible
> name. This would allow to always use (for every HTML element) the HTML
> core attribute 'title' for the accessible description, which is what  
> the
> already proposed patch implements so far.
>
> In other works, current patch proposal is as follows:
>
>  * Always expose 'alt' in images as the accesible name.
>  * Always expose 'title', for every HTML item, as the accessible
>    description.
>
> Problem is that this change, which affects to all the platforms (and  
> not
> only to GTK, despite of the bug name), would also mean to change two
> unit tests which are failing right now:
>
> - accessibility/img-aria-button-alt-tag.html
> - accessibility/input-image-alt.html
>
> Those tests are failing because they *always* expect the 'alt'  
> attribute
> for images to be exposed as the accessible description instead of the
> name, and that would need to be changed if this patch got accepted.
>
> Thus, as this change would affect to all the platforms and would  
> mean to
> change two regression tests, we'd love to hear more opinions from
> someone else on this topic before making a final decision.
>
> So... any thoughts?
>
> Thanks,
> Mario
>
> [1] https://bugs.webkit.org/show_bug.cgi?id=25524
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20090723/f3f14f42/attachment.html>