Re: Fwd: [orca-list] Issues with Qt (was Re: Another call for testing master)

Mandag 11. november 2013 10.04.30 skrev Joanmarie Diggs:
> Hey all.
> 
> I'm doing some *major* refactoring in Orca. As part of this, many many
> sad hacks are being removed from Orca -- and exposing bugs in
> accessibility implementations. Below is the first (couple) found.
> 
> Whether or not these issues persist in Qt5, I haven't had the
> opportunity to check yet. But if anyone has some free time to give the
> Qt a11y implementation a closer look, that would be great.
> 
> Lastly, where would y'all prefer these get collected? Here? Issue
> tracker(s)??

Please file such bugs under https://bugreports.qt-project.org/ .
This is good feedback and I'll make sure that this kind of problem is fixed in 
Qt 5 at least. If I find the time I'll also try to fix it in Qt 4.

Greetings,
Frederik


> 
> Thanks in advance!
> --joanie
> 
> -------- Original Message --------
> Subject: [orca-list] Issues with Qt (was Re: Another call for testing
> master)
> Date: Mon, 11 Nov 2013 09:55:09 -0500
> From: Joanmarie Diggs <[email protected]>
> To: Peter Vágner <[email protected]>, [email protected]
> 
> Hey Peter.
> 
> On 11/11/2013 05:04 AM, Peter Vágner wrote:
> > What I have noticed is that QT apps need additional testing after this
> > refactor.
> > For example treeviews are not anounced while using VLC. It's not
> > critical, content is there but the fact treeview is in focus is no
> > longer announced.
> 
> That is due to a Qt bug combined with a removed hack in Orca. Objects
> that claim focus need to have two accessible states:
> 
> * focusable
> * focused
> 
> At least in the case of the tree in question, both of these states are
> absent. In order to not introduce a regression, I have just created a
> Qt-specific toolkit script in Orca and will put Qt-specific hacks there.
> On a related note, it might not be possible to put Qt-specific hacks
> anywhere else (e.g. an app script). Because at least in the case of VLC,
> we're not getting any accessible app name. Here's the debug.out output
> from the event in question that shows both problems:
> 
> vvvvv PROCESS OBJECT EVENT object:state-changed:focused vvvvv
> OBJECT EVENT: object:state-changed:focused             detail=(1,0,[tree
> 
> | ])
> 
>     app.name='' name='None' role='tree' state='active editable enabled
> sensitive showing visible manages descendants' relations=''
> Script for event:  (module=orca.scripts.default)
> TOTAL PROCESSING TIME: 0.0039
> ^^^^^ PROCESS OBJECT EVENT object:state-changed:focused ^^^^^
> 
> > I haven't yet tested other more complicated QT vidgets such as
> > comboboxes, tab controls etc
> 
> If you have a chance to do so, I'd appreciate it. Based on the above, I
> fear there may be other accessibility issues on the Qt side that Orca is
> no longer hacking around in its default script.
> 
> Thanks for the testing!
> --joanie
> _______________________________________________
> kde-accessibility mailing list
> [email protected]
> https://mail.kde.org/mailman/listinfo/kde-accessibility