I'm resending the mail to the proper Qt/KDE list.

----------  Forwarded Message  ----------

Subject: Re: Enabling accessibility stack by default in Qt4/Qt5
Date: Tuesday 01 September 2015, 18:50:43
From: Lisandro Damián Nicanor Pérez Meyer <perezme...@gmail.com>
To: debian-qt-...@lists.debian.org
CC: Samuel Thibault <sthiba...@debian.org>, debian-

On Tuesday 01 September 2015 01:39:51 Samuel Thibault wrote:
> Hello,
> As discussed at DebConf [1,2,3], we would like to make the accessibility
> stack enabled by default, so that all a user who needs it has to do is
> just to start orca (while at the moment she has to find an option in
> the control panel, and logout/login again, thus closing all running
> applications...).
> For Qt4, this is implemented in the qt-at-spi package, which has been
> developped mostly between 2009 and 2013.  There doesn't seem to be much
> activity there since early 2013.

Kind of. See the last part of the first message in:


And the following comments. I'll try to get at it as soon as possible, but not 
before the gcc5 transition ends.
> For Qt5, this is implemented directly in the Qt core, and seems to be
> actively maintained and improved.


> So it seems safer to enable accessibility in Qt5 than in Qt4, but for
> now it seems to me that most Debian applications are still using Qt4,
> so people will mostly need Qt4 accessibility. 

Well, that's changing. And will change a lot after the gcc5 transition.
We are retiring qt4's webkit in Stretch, and will do our best to try to get 
most apps ported to Qt5 (ideally killing Qt4 in the process).

> Considering that we
> are just at the beginning of the Stretch development, perhaps we can
> try to enable accessibility by default in both Qt4 and Qt5, see how
> well this goes, and at worse not enable it by default for Qt4 if the
> possibly-revealed bugs become really not solvable?

I like the idea, but ~1 month after the gcc5 transition ends. We are also 
pushing KDE+Qt5 to the archive and have quite some problems with the 
transition, so it would be better to let the transition "dust" settle down 
before adding yet more sources of bugs. After that we should definitely give 
it a try IMO.

> The way I see it is simply by making the at-spi2-core package define
> Qt5) to 1 from a script in /etc/X11/Xsession.d.
> Is it OK with Qt people?

Sounds like a good starting point. I see that /etc/X11/Xsession.d/90qt-a11y is 
provided by at-spi2-core. I think the a11y stuff in Qt5 is provided by 
libqt5core5, please feel free to file a wishlist bug to have a similar script 
provided by it.

Now we should also consider Wayland and even pure console. Qt5 has been 
improving in this side (yes, it can draw a GUI without X and without the need 
to recompile anything). I don't know if a11y is present in those cases, but if 
it is it would also be good to check them.

Kinds regards, Lisandro.


Lisandro Damián Nicanor Pérez Meyer

Lisandro Damián Nicanor Pérez Meyer

Attachment: signature.asc
Description: This is a digitally signed message part.


Reply via email to