Commit 021fce8c (datetime: Don't show Auto Timezone setting
without location services) added code that unconditionally disables
the "automatic-timezone" dconf key while loading the datetime
panel view. Make this code conditional, so it only executes if
HAVE_LOCATION_SERVICES is not defined by the compiler.

Fixes #3157

Signed-off-by: Daniel DeGrasse <daniel@degrasse.com>

When the window is small, the rows show a broken icon in the reveal
button, because the "view-layout-symbolic" icon can't be found.
To fix this, replace such icon with "view-reveal-symbolic", as per
latest mockups[1].

[1] https://gitlab.gnome.org/Teams/Design/settings-mockups/-/commit/3a5fe844dde31326e9951ba248238b113a134923

(cherry picked from commit 2f394dfd)

The icon `view-reveal-symbolic` has been in a-i-t for years now.
Previously, the `view-layout-symbolic` was broken as it wasn't added
under the right path.

Now that we use `view-reveal-symbolic`, we can remove the older asset.

Currently it looks like we crash if the permission variant is a non-null
strv containing no strings.

Probably fixes #3168


(cherry picked from commit fa994e08)

In uses of CcBackgroundPaintable, the parent GtkPicture's scale-factor
is bound to that of CcBackgroundPaintable. Binding this property will
always result in a property setting of scale-factor. Currently, that
results in a cache update, even if the new scale-factor is the same as
it was.

By checking if that is actually the case, we can avoid any unwanted
cache updates.

This avoids the situation where the user has a scale-factor other than
1, as the paintable gets constructed with scale-factor 1, which then
causes another cache update once the scale-factor is set from the parent
GtkPicture usually.

Instead, we can just initialize the paintable with the correct
scale-factor which is known from any parent widget. It will now only
cause a cache update in the rare case that the scale-factor is changed
while this panel is open.

Otherwise they are not available during construction and the init
scale-factor other than 1 will still cause a cache update.

The "size" string property is only used for debugging purposes. But
updating it is costly, as it uses libgnome-desktop calls on the
background which may involve file IO.

Therefore, we should only call update_size whenever we actually want to
return it. None of the functions that can return or use the size string
are currently in use, so this should not add any time anywhere.

Mutter takes a key repeat interval of 0 to mean disabled, so we should
limit the range to not get to 0.

https://gitlab.gnome.org/GNOME/mutter/-/blob/0104fbe577594bcc7fdc965be323be08e4bd35d0/src/wayland/meta-wayland-keyboard.c#L500

Closes #2409


(cherry picked from commit 0cf22e6e)

It was referencing the wrong source, namely the main list box instead of
the treeview. Also, the callback should return TRUE, otherwise GTK tries
to load the empty invalid URI.

Fixes #1184


(cherry picked from commit 9f238c09)

Clicking on a color device row expands it and shows the buttons.
However, expanding it with the arrow does not. This is confusing. Also,
unexpanding it with the arrow again does not hide the buttons.

We can fix this by having an expanding row select itself, and an
unexpanding row unselect itself.

Fixes #11


(cherry picked from commit a1c9ba80)

Let's append ".desktop" to its app ID before launching GNOME Software.

Fixes #1022


(cherry picked from commit f1f36405)

Fix statically defined translatable strings, missed being passed to _().

cc #3189

If the user brings a stylus into prox over the window which does not
have an ID known to libwacom, the program will crash due to a NULL
dereference here by `libwacom_stylus_is_erase`. The caller already
knows how to handle a NULL, so simply pass the NULL along if libwacom
was unable to find a stylus.

Closes: https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/3211
(cherry picked from commit 47d3bf94b27dabe45455d9bdb5aec9a6eeb67b61)

To reflect the user choice rather than the system state. The system
state is communicated by the top banner that will indicate that the
setting will take effect on the next login.

Fixes #2542

(cherry picked from commit c08a9206)

(cherry picked from commit 42dd0fb2)
(cherry picked from commit b8631dcb)

https://www.freedesktop.org/software/systemd/man/latest/hostnamectl.html#hostname%20%5BNAME%5D

Hostnames can be fully qualified.

Otherwise we always replace dots with hyphens, even when the desired
result is a dot separated FQDN.

Various system services expect a FQDN, such as Idm/IPA.

Related to #2918


(cherry picked from commit 2d36acd9)

So that in locales with long labels we don't endup linebreaking the row
title.

Fixes #3179 for gnome-47 and gnome-46. For gnome-48 alpha and beyond we
use AdwToggleGroup which doesn't have this problem.

See b6cfaa22


(cherry picked from commit 4a1ff843)

Fixes: dac9065d ("wacom: Drop list to handle pads")
(cherry picked from commit 794a7f5e)

If our stylus device is the libwacom fallback device forcibly disable
the Map Buttons button even if we do have a pad. mutter will not
actually show an OSD anyway because the layout file doesn't exist (or it
will crash, see mutter#3722).

Since the libwacom fallback device is a built-in one (and those don't
have pads) so not showing this button has no effect on the vast majority
of devices that make use of this feature - the few external tablets with
a pad can be added to libwacom to make this problem go away.

Closes: https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/3212
(cherry picked from commit 5d731ced)

The EKR is an external TV remote-like device. For a long time only Wacom
had that device, now other manufacturers also have similar devices.

Let's use libwacom to fetch the device class and handle all remotes the
like the EKR before.

This is slightly inefficient as we now construct the device and then
dump it immediately if it is not a remote. The alternative is
constructing a libwacom device from the vid/pid but our experience with
Huion devices has shown those may not be unique. So it's a lot easier to
be inefficient here and let libwacom take care of the crazy.

Note that libwacom_get_class() is deprecated and may cause warnings.
I can be replaced with an integration flag check once released, see
libwacom#779.

Closes: https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/3216
(cherry picked from commit 00568b38)

The set_value_for_combo_row helper will insert a new item if the
chosen value was not already present in the list, and inserts it in
sorted order.
However commit a0d49cb3 broke the position where the new item is
inserted, making it always the last element.
Because it then tries to select the row corresponding to the correct
sorted position, it would select the wrong row.

The idle-delay and lock-delay settings are read from gsettings, which
the user might have set to values not present in the enum model.

Ideally we would add the unknown value to the model, like equivalent
idle-delay setting from the power panel. But these widgets are set-up
with AdwEnumListModel's and changing that would require a refactoring.

When CcListRow is insensitive, the arrow is still white, which
creates an odd state, as if the row isn't properly insensitive.

Add the .dim-label style class to the arrow in this case, so the
row looks more uniform.


(cherry picked from commit 104a6d82)

The Network panel uses addition props/signals to track the NM state, so
let's wait for those too, which could maybe improve races in the tests.

See #1768

(cherry picked from commit e7501355)

Even with the previous commit, one test still keeps failing
intermittently. For now, let's disable that test.

Closes #1768

(cherry picked from commit 6bfdc675)

This reverts commit 2d36acd9.

(cherry picked from commit 0197dc14)

The previous reverted commit tried to allow FQDN hostname entries by
allowing periods in them. Although that is fine, the tested hostname
"Jawoll. Ist doch wahr!" now fails, as that should be turned into the
static "jawoll-ist-doch-wahr" per recommendation of
https://www.freedesktop.org/software/systemd/man/latest/org.freedesktop.hostname1.html

Instead, we should probably check if an entry with periods is actually
an FQDN. In that case, we can pass it as a static hostname.

The check is maybe not completely up to the standards, but should catch
all common cases.

(cherry picked from commit 40ce21e4)

These were tested with `hostnamectl hostname --static [NAME]`. The
unmodified ones are indeed allowed by `hostnamectl`, the other two are
not.

Also, don't break on an emptly line in the text file, as we can just
skip it.

(cherry picked from commit a13269c3)

Whether libwacom knows about a tablet has no effect on whether the
tablet works, only whether we know how to map it, which styli are
available, etc.

Instead of pretending there is no tablet connected let's use the
libwacom fallback tablet instead. As of libwacom 2.11 that defaults to a
built-in tablet (the vast majority of devices these days) which means we
can at least map this to an output, calibrate it etc.

(cherry picked from commit b82621a3)

This backport avoids breaking the string freeze for gnome-46.

Update to upstream version '46.5'
with Debian dir 0901608824c323262605254fc2424d96997dc71b

Dirk Su authored Oct 14, 2024 and Alessandro Astone committed Nov 21, 2024

lp: #2084405

ChunAn Wu authored Nov 04, 2024 and Alessandro Astone committed Nov 21, 2024

LP: #2084407

The Yaru-specific resources are only registered when the last-used
theme was Yaru. If g-c-c was started with another theme, we would
unref the Yaru resources one too many times.

LP: #2085610