WirelessSecurityWPAEAP is a GtkWidget owned by the CEPage8021xSecurity
widget, which is supposed to "unparent" it on "dispose" (since parents
hold a reference to child widgets). Instead we were calling
g_clear_object on it.
Fixes#1671
Weak references are added when logical monitors are created, however we
don't remove them when destroying the display.
This means that if a monitor survives to the display finalization
(because may be referenced elsewhere) it will make g-c-c to crash
during its finalization, as that will trigger the weak reference
callback that will try to access to the already-finalized display.
getlogin() can fail for several reasons as detailed in the man page, and
the current behaviour is a segmentation fault when it fails with NULL,
such as due to an unset loginuid.
* Check return value for error and act accordingly.
* Change to getpwuid(getuid())->pw_name, which is less likely to error.
Instead of using a plain "flat" button positioned at the window corner,
let's use GtkWindowControls to wrap the close button and get the
default window control styling and alignment.
Fixes#1737
Make sure to keep a reference to the GoaObject of the account around so
that the GBinding that synchronises each switch widget and the account
properties don't get finalized on startup.
#0 0x00007ffff73c6a20 in g_binding_finalize () at /lib64/libgobject-2.0.so.0
#1 0x00007ffff73d3d22 in g_object_unref () at /lib64/libgobject-2.0.so.0
#2 0x00007ffff73c68e8 in weak_unbind () at /lib64/libgobject-2.0.so.0
#3 0x00007ffff73cf117 in weak_refs_notify () at /lib64/libgobject-2.0.so.0
#4 0x00007ffff72acd6c in g_data_set_internal () at /lib64/libglib-2.0.so.0
#5 0x00007ffff73d0195 in g_object_real_dispose.lto_priv () at /lib64/libgobject-2.0.so.0
#6 0x00007ffff73d3c44 in g_object_unref () at /lib64/libgobject-2.0.so.0
#7 0x00007ffff72b6793 in g_hash_table_remove_all_nodes.part () at /lib64/libglib-2.0.so.0
#8 0x00007ffff72ba723 in g_hash_table_unref () at /lib64/libglib-2.0.so.0
#9 0x00007ffff753403d in g_dbus_object_proxy_finalize () at /lib64/libgio-2.0.so.0
#10 0x00007ffff73d3d22 in g_object_unref () at /lib64/libgobject-2.0.so.0
#11 0x0000000000402d08 in glib_autoptr_clear_GoaObject (_ptr=0x5d59f0) at /usr/include/goa-1.0/goa/goa-generated.h:3265
#12 glib_autoptr_cleanup_GoaObject (_ptr=<synthetic pointer>) at /usr/include/goa-1.0/goa/goa-generated.h:3265
#13 on_application_activate_show_account_cb (application=0x49f2f0, argv=<optimized out>) at ../../../../Projects/jhbuild/gnome-control-center/panels/online-accounts/gnome-control-center-goa-helper.c:360
Closes: #1721
In the previous design of this panel, it made sense to show the
currently selected monitor because all editing widgets were in
the same page. With the new design, however, the monitor options
were moved to a separate page, and that page already shows which
monitor is being edited.
Remove the colored background of selected monitors.
Instead of updating the titlebar of the monitor preferences page in
the row clicked callback, update it in set_current_output(). This
ensures that the title of the page is always in sync with the monitor
it's displaying.
PpPpdOptionWidget and PpIppOptionWidget both use combo boxes for
certain types of selections. With GTK4, combo boxes no longer
support scrolling[0], which in turn causes problems setting some
things in the PpOptionsDialog[1].
This replaces instances of GtkComboBox with GtkDropDown which do
support scrolling. This change was applied to both PpIppOptionWidget
and PpPpdOptionWidget as both are used in PpOptions dialog.
Since the configuration values passed to CUPS can no longer be stored
in a GtkTreeModel alongside the displayed values, some logic changes
to update_widget_real in PpPpdOptionWidget to maintain the reference
to the ppd_option_t so the selected index can be mapped to the
configuration value.
[0] - https://gitlab.gnome.org/GNOME/gtk/-/issues/3674
[1] - https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/1704
Those are not always present in the device string.
Guidance was taken from the usage on vendor websites.
NVIDIA actually has the rights to GTX™ but doesn’t seem to use it,
in contrast to RTX™.
This made everything AMD lower case after the second word
and is likely unintentional for something like EPYC™.
e.g. was
AMD Ryzen threadripper 1950x 16-core processor
AMD Radeon rx 580 series
Previously, when a list of VPN plugins was loaded once, it will not load
again the next time you try to add a VPN connnection. This causes an
issue that if you keep gnome-control-center open and install a new VPN
plugin, the new VPN plugin won't show up in the VPN plugin list.
Fix that by always load all avilable VPN plugins when showing the
"Add VPN" dialog.
If an user opens an connection editor and then closes it, repeat this
multiple times causes lots of unused hidden editor windows.
Fix that by destroying the connection editor instead of hiding them
when closing.
The hostname/device name dialog has only a GtkEntry. So a user
navigating with only a keyboard should be able to apply their
changes by pressing "Enter".
This went mistakenly missing in the recent UI updates. Since this
panel is still accessible through search despite tablet availability,
we should have a proper empty state view when moving to the panel
with no tablets available.
Fixes: https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/1684
Turning off the service doesn't mean we need to disable the RDP backend,
the setting can stay enabled in case one wants to manage it directly via
systemd.
We were checking for the existance of an image file as avatar when
setting up the main avatar widget in the page, but the same wasn't
done for the avatars displayed in the "Other users" list. These
were AdwAvatars simply initialized to use the user initials.
Now we load the user image (if it exists) and it will fallback to
initials when an image isn't set.
Fixes#1658
