| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This currently affects only the xsettings plugin, which needs GDK to use
the x11 backend. It already calls gdk_set_allowed_backends ("x11") for
this purpose; however, if GDK_BACKEND=wayland (or any other non-x11
backend possibly) happened to be in the environment, GDK would fail to
initialize at all. This would prevent the xsettings plugin from starting
up.
Big thanks to Olivier Fourdan for figuring this out!
Bug: https://bugzilla.redhat.com/show_bug.cgi?id=2022283
Bug: https://bugs.debian.org/1008992
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
geocode-glib 3.26.3 was released today, with optional soup3 support.
When built against soup3, it generates a geocode-glib-2.0 pkg-config
file rather than geocode-glib-1.0.
Support both versions, with a preference for the libsoup3 version.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
| |
This is now handled dynamically since commit e2d268eb00, so we should
not tie dconf setting and Xsetting automatically here. Doing so, we
are clobbering the dynamic value on startup, making it only effective
on later changes.
Fixes: https://gitlab.gnome.org/GNOME/gnome-settings-daemon/-/issues/682
|
| |
|
|
|
|
|
|
|
| |
The gnome-shell magnifier uses AT-SPI caret and focus events to follow
keyboard navigation. While GTK emits those as long as it can connect to
the a11y bus, other toolkits/applications such as Qt or Firefox check
for the IsEnabled property which follows this GSettings key.
Fixes: https://gitlab.gnome.org/GNOME/gnome-settings-daemon/-/issues/688
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
python-dbusmock will also stop importing all session services that are
installed on the host system by default (as is already the case for
system services). With this change, the dconf service cannot be
activated anymore unless we explicitly enable the service in the
environment.
Add the appropriate call to enable the service, guarded by a check that
the method exists to be compatible with older dbusmock versions.
See: https://github.com/martinpitt/python-dbusmock/pull/108
|
| | |
|
| | |
|
| |
|
|
|
| |
This is broken comparing with stable release numbers, like 42.0. This
version is also released, so it sounds better to simply depend on that.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This setting used to be modified by gsd-keyboard at runtime, but it no
longer does. We want to leave this setting in a pristine state before
we lend control to the user in order to avoid setting leftovers make
GTK and others use the unintended IM module.
Since the setting is actually staying on the same schema/path, there is
no nice mechanism that would help us in doing a one-time port, so rely
on a file at ~/.cache to make this happen once. In the common case, it
just adds one stat() more at startup.
After this migration is done, the gtk-im-module setting can be considered
in full control of the user.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Settings now directly uses systemd for managing the gnome-remote-desktop
service, thus we don't need to have gsd-sharing do it for us anymore.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Right now, gsd-keyboard and gsd-xsettings have a strange relation
where the first sets the gtk-im-module dconf setting for the latter
to read the setting and forward it through XSettings.
Since this detection is highly X11 specific, make it happen in the
Xsettings daemon itself, from the relevant setting and device presence.
This makes users still able to shoot themselves in the foot by changing
the setting, X11 clients able to be told to switch to ibus if it turns
out necessary, and Wayland clients unaffected otherwise.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/4443
|
| | |
|
| |
|
|
| |
gnome-shell itself now handles the screencast key too.
|
| |
|
|
|
| |
gnome-shell itself now handles the screenshot keys. This is required to
easily provide same-frame response to pressing a screenshot key.
|
| |
|
|
|
|
| |
Otherwise there is a race condition that can cause the FD to be closed
twice. Avoid this by signalling EOF through a separate variable and
always closing the FD from the main thread.
|
| |
|
|
|
|
|
| |
The lid-switch is inhibited at startup, and a number of tests rely on
this inhibition to have been released again. Unfortunately, it seems
that sometimes we were not clearing all messages properly. Either way,
explicitly waiting for the uninhibit that happens at startup is sane.
|
| |
|
|
|
| |
Those usually get port names from UCM. Add common port names from
there.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Despite the debounce logic introduced in
60621b90bc3154dc905ed90780a7f19eb553ab26, GSD kept warning me of the
battery level every time my Logitech K600 keyboard (via Unifying
receiver) disconnects/reconnects.
Added another condition: when battery level drops to
`UP_DEVICE_LEVEL_NONE`, the previous entry in the hash table would not
be overwritten. This avoids duplicated warnings when transitioning
NONE->LOW->NONE->LOW.
|
| | |
|
| | |
|
| |
|
|
| |
gweather_location_ref()
|
| |
|
|
|
|
| |
Use the new GWeather major version.
The only change consists in replacing GWeatherTimezone with GTimeZone.
|
| |
|
|
| |
Ignored in meson < 0.60.0, deprecated since 0.60.1 and fatal since 0.61.0.
|
| | |
|
| |
|
|
|
|
|
| |
Some versions of mutter were not setting the use as active when the lid
was opened. Skip the test in case the current mutter version does not
contain the (latest) fix. This will also disable the test for older
versions that work, but that is fine.
|
| |
|
|
|
|
| |
The others are not erroring out correctly for close_range syscalls,
causing issues when spawning. So use the "crun" runners and disable
"asan" as there is no overlap.
|
| | |
|
| |
|
|
| |
(cherry picked from commit d058db6a091326962562a816169e345a50034217)
|
| |
|
|
| |
(cherry picked from commit 6868a4a102902834f6c193b3a321053e2b82afb0)
|
| | |
|
