Re: Implementation bugs (present and anticipated) with aria-owns in tabular containers

Hi, Joanie.

On Tue, Mar 31, 2015 at 9:49 PM, Joanmarie Diggs <jdiggs@igalia.com> wrote:

> Hey Alex.
>
> On 03/31/2015 09:17 PM, Alexander Surkov wrote:
> > I think your example is valid and grid should work as if the sibling row
> > was a child of the grid. The browser have two options to make it
> > working: fix table interface implementation for this case or change the
> > accessible tree to reflect aria-owns in the hierarchy.
>
> Please tell me that the options are not mutually exclusive. ;)
>

Basically they are, but only because there's no reason to have them both
same time: either you do this or that way in order to make things working.


>
> Just in case you are suggesting they are mutually exclusive: Whether or
> not the tree is changed based on aria-owns, grids and tables are
> expected to implement the accessible table interface on those platforms
> that have such an interface -- and implement it correctly.
>

that's the point, it doesn't matter how browser implements aria-owns, the
table interface should just work.


>
> > I find 2nd
> > approach more prominent to implement, we have a bug for Firefox [1].
>
> Could you please clarify what you mean by prominent?
>

I meant it seems to be the easiest approach to make all stuff working
because aria-owns affects on events, group information, table interface and
whole bunch of other things. If aria-owns altered the tree then the browser
implementation can get all these stuff for free. If accessible tree
alteration doesn't harm the assistive technology then we should be in good
shape to make it. Iirc IE does this. So I lean towards to make same in
Firefox.


>
> Thanks!
> --joanie
>
> > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=1133213
>
>