[Pkg-xfce-devel] Bug#760740: Replacing or taking precedence over an existing AT-SPI process

Samuel Thibault samuel.thibault at gnu.org
Sun Nov 30 22:11:17 UTC 2014


Luca Saiu, le Fri 28 Nov 2014 19:21:08 +0100, a écrit :
> When working to solve bug 760740 (at least on bleeding-edge versions of
> lightdm and lightdm-gtk-greeter) I found a different workaround than the
> one employed by Samuel: setting the xserver-share option to false in
> lightdm.conf lets AT-SPI die after the Lightdm greeter session
> terminates and the X server exits; I suppose restarting the X server
> wouldn't be unacceptable,

I guess that there are some reasons why this is left as an option :)

> Anyway all of this feels a little fragile.


> I would like to allow AT-SPI to be started with some --replace version
> so that the current accessibility bus can be replaced by a new one;
> or, if the process can't be killed, at least the current user could
> specify her own new process.

Well, that could also pose other problems when run as the proper user
too: in case the bus is not working for some reason, that'd create
another at-spi bus, and then we'd have applications which are not
accessible to the screen reader because they are not running on the same
bus.  I'm not sure which is less confusing for the user.

> Orca does that in a fine way at a different level, connecting to AT-SPI;
> What I'm imagining is doing the same when AT-SPI connects to dbus.  Is
> there some fundamental reason why my idea is wrong or unfeasible (for
> example, permissions)?  Do you happen to have some pointers for me to
> look at to better understand the issue?  Is this worth the trouble?

I would say this is worth discussing directly with AT-SPI maintainers.
We usually try to avoid introducing patches in the Debian packages.


More information about the Pkg-xfce-devel mailing list