[Kde-accessibility] Fwd: An accessible Qt? I think not.

[Leo Spalteholz]

Oops, gmail decided to reply only to Gary, just forwarding to the list
for completeness.

On 11/15/05, Leo Spalteholz <leo.spalteholz at gmail.com> wrote:
> On 11/15/05, Gary Cramblitt <garycramblitt at comcast.net> wrote:
> > On Monday 14 November 2005 08:00 pm, Leo Spalteholz wrote:
> > > Specifically for Qt, support is quite poor (although Openoffice is far
> > > worse).
> >
> > Leo responded to me that this is Qt4.  That's very disappointing.  Leo please
> > do report these to Trolltech and keep us informed.
>
> Will do.  I brought up the issue, and they confirmed that there were
> some inconsistancies with MSAA support in menus, and promised to look
> at the issue in Qt 4.1.  I will nail down some more concrete examples
> of other problems and submit them to Trolltech.
>
> >
> > I can understand the menu problems you report had you said Qt3.  The QMenuBar
> > and QPopupMenu widgets contained the individual menu items in a QMenuData
> > structure that (I'm guessing) did not inherit the QAccessible interface.  In
> > Qt4, the menus have been unified into a QMenu widget with QAction items.  I
> > would have thought QAction would inherit the QAccessible interface and
> > provide the accessible information.  Perhaps it is because QAction inherits
> > from QObject, not QWidget that Inspect32 doesn't see the accessible info?  Or
> > perhaps Inspect32 doesn't know how to ask the QMenu widget for the individual
> > menu items?
> >
> > The listview problems are probably similar issues.  Are you using the
> > model/view architecture?
>
> I'm not sure, I was just testing it against various widgets in Qt
> Assistant that came with the windows open source version.  The same
> problems are evident in all Qt apps that I've tried, including such
> prominent ones as Opera.  Trolltech informed my that they use
> Inspect32 for testing, (because I brought up an issue I observed with
> AccExplorer 2.0) but it is of course possible that I am identifying
> certain behaviour as problematic that is not technically incorrect.  I
> am just comparing the behaviour to what I would expect to get from
> certain functions, and also what information other applications
> provide.
>
> >
> > I'm also a bit confused because I had understood there must be an accessible
> > plugin installed that intermediates between Qt and Inspect32. ???
>
> I'm not entirely sure what you mean, but there is a compile-time
> option for Qt to enable MSAA accessibility support.   It is on by
> default, so I assume that Qt is loading that plugin.
>
> >
> > My understanding is that oo.org does not support MSAA.  It does support the
> > Java Accessibility API and AT-SPI, so it isn't surprising at all that
> > Inspect32 would "fail miserably" on oo.org.
>
> Ah.  Thank you for clearing this up.  This means I need to do some
> research into the Java Accessibility API so that I can support
> OpenOffice with my accessibility client.
>
> Cheers,
> Leo
>