Re: [orca-list] more on inaccessibility when run as root

[Peter Vágner <pvdeejay gmail com>]

Hello,

I'll be playing with this more however I think we are talking about 
different issues.

One issue is that sometimes occassionally apps running as root are slow 
to launch and they are not accessible. Most likelly this is the issue 
Mike is addressing with the patch.

The other issue is that non gnome gtk based desktop enviromments such as 
mate and xfce are accessible when they are launched after the 
accessibility support is turned on in their respective control pannels. 
However when launching graphical apps with root privileges while running 
these enviromments apps running as root are *never* accessible with orca.
I don't know how to replicate this with polkit, however I have just 
found a workaround on how to make apps ran via sudo accessible in such a 
case.
Add this line into your sudoers file by using visudo command
Defaults env_keep += "GTK_MODULES"
Then running for example
sudo pluma
will give you accessible text editor running as root.

As I have just said I'll play with this more.

On 07.12.2015 at 19:26 kendell clark wrote:
> hi
> I'll test it. Do I simply clone the at-spi2-atk branch and apply the
> patch there?
> Thanks
> Kendell clark
>
>
> On 12/07/2015 12:14 PM, Mike Gorse wrote:
> > Hi all,
> >
> > I've seen something similar; currently, when deciding whether to accept
> > a connection from an application such as orca, at-spi2-atk tries to
> > determine whether the user making the connection is the same user
> > running the application being connected to. If the application being
> > connected to is running as root, then it tries to figure out whether its
> > parent process is owned by the user making the connection. It seems that
> > this doesn't always work.
> >
> > I am considering applying the attached patch. It would be helpful if
> > someone is able to test it.
> >
> > Thanks,
> > -Mike
> >
> > On Mon, 7 Dec 2015, Peter Vágner wrote:
> >
> > > Hello,
> > >
> > > I have already tried looking into this however I don't have a full
> > > understanding of the accessibility stack so I was not yet able to come
> > > up with a solution / workaround / whatever that might improve the
> > > situation.
> > >
> > > What I have understood so far:
> > > - This is not orca's fault since it has already been explained.
> > > - I haven't noticed enviromment variables which may control this on my
> > > system while testing with gnome.
> > > - It's not an at-spi nor ATK's fault as it's installed and launched
> > > the same way in gnome and also in mate. I assumed we might miss some
> > > dbus configuration files or other gnome specific features when
> > > launching gui apps as root.
> > >
> > > Perhaps someone can remember /root/.orbitrc file where we were able to
> > > tune this back in the corba days.
> > >
> > > I believe if I was able to understand how gnome does it I might be
> > > able to replicate this on other desktops.
> > >
> > > Greetings
> > >
> > > Peter
> > >
> > > On 07.12.2015 at 16:35 kendell clark wrote:
> > > > hi all
> > > > Since I'm trying to improve the accessibility of the mate desktop, I've
> > > > been digging into why apps being launched as root are inaccessible to
> > > > orca. Joanie pointed out that generating a debug log wouldn't help
> > > > because orca isn't recieving any events so can't process them. I think
> > > > I've found a reproduceable way to encounter the issue. Steps to
> > > > reproduce. 1: Log into mate in your favorite login manager. 2: make sure
> > > > orca starts. 3: run an app that requires root privs, example gparted. 4:
> > > > Log out and log back into a desktop such as gnome or unity which this
> > > > does work. Do not restart, but press the log out button in mate's menus.
> > > > then try to launch the same app you just did using mate. What should
> > > > happen is that orca should then read the app. But this doesn't work.
> > > > This leads me to think this is an at-spi, rather than an orca issue. But
> > > > I'll need help from someone smarter than me to tell for sure. If anyone
> > > > can confirm or at least get close, I'll file a bug against at-spi2-core
> > > > so that this can get fixed. Note that this happens in cinnamon also
> > > > sometimes but not always. I think it may have something to do with the
> > > > fact that when logging out, the copy of at-spi2-registryd isn't killed,
> > > > instead a new instance is spawned, which can sometimes lead orca to
> > > > start but not speak because you can only have so many instances going at
> > > > once. Why orca doesn't connect to the existing one I don't know but I'm
> > > > sure it tries to.
> > > > Thanks
> > > > Kendell clark
> > > > _______________________________________________
> > > > orca-list mailing list
> > > > orca-list gnome org
> > > > https://mail.gnome.org/mailman/listinfo/orca-list
> > > > Orca wiki: https://wiki.gnome.org/Projects/Orca
> > > > Orca documentation: https://help.gnome.org/users/orca/stable/
> > > > GNOME Universal Access guide:
> > > > https://help.gnome.org/users/gnome-help/stable/a11y.html
> > > > Log bugs and feature requests at http://bugzilla.gnome.org
> > > _______________________________________________
> > > orca-list mailing list
> > > orca-list gnome org
> > > https://mail.gnome.org/mailman/listinfo/orca-list
> > > Orca wiki: https://wiki.gnome.org/Projects/Orca
> > > Orca documentation: https://help.gnome.org/users/orca/stable/
> > > GNOME Universal Access guide:
> > > https://help.gnome.org/users/gnome-help/stable/a11y.html
> > > Log bugs and feature requests at http://bugzilla.gnome.org