The BG_RECENT_SOURCE (user_data) call was done before it was checked if is was
cancelled. At this point user_data is pointing to an object that has been deleted.
Fixes#729
Monitors may not always have a mode. This means we cannot look up the
mode on the fly, but instead need to have a heuristic to select a mode
if there is none.
Everything was in place already for this, except that one location was
still doing a fresh lookup, possibly causing a crash.
Fixes: #675
localed will not update the LC_* variables if they are not given. This
means that should the user switch the formats away from LANG and then
back, they will not be reset appropriately.
Simply always setting the variables even if the match will fix this.
Add a scheme to the address which we test for correctness
by g_network_address_parse_uri(). It does not work without it.
Use "none" scheme if user did not entered one.
Use port number 0 if user did not specify any.
Fixes#679
Empty is a valid value here. It means the setting is not set. Two places
need to be updated:
- the validation function
- when retrieving the value from the combo, empty strings need to be
mapped to a setting value of NULL
Closes#677
"Primary Display" and "Contains top bar and Activities" are visible when
you have >1 display connected and the "Join Displays" option is
selected. I can't work out how to see the "Active Display" string but it
sure looks translatable to me.
When the virtualization widgets are hidden, the following widets don't need to
move up. This seems to be due to the orignal author thinking this was required
or a bug in GTK rendering a double space between these rows.
They’re already listed in a section titled ‘VPN’, and the user is likely
to have included ‘VPN’ in the device’s name already.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Fixes: #701
As the hotspot dialog is reused, we have to reset the connection label
everytime it’s shown. Otherwise obsolete informations shall be shown
if there isn't any active connection
Setting markup within UI file might be bad as some translators
might break those markups.
Setting it in C source allows to exclude the markups from being
translated, and thus to not confuse translators.
642be83798 (note_608677)
When creating a Page, the dialog variable wasn't set and thus
dereferencing a NULL in a later code resulting a crash.
This was introduced in 14b9f65ba4
The other checkboxes on this page all have mnemonics. It seems unfair
for the background data usage one to be left out of the party.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
The Add Connection widgets from the connection
editor dialog weren't properly using the available
space, since they don't expand.
Fix that by making them expand and fill the available
space properly. It also avoids blank space by vertically
centering the list.
The colour icon blends with bright themes (such as Adwaita) making
it hard to distinguish.
The symbolic variant of the same asset contrasts fine with light
and dark background and matches the "symbolic" pattern of the other
two buttons shown in the dialog (the audio-volume-* icons).
Fixes#502
At the default window size, the printer name label is too wide and
collapses the UI into phone mode. (This is the only panel that shows up
in phone mode by default.) Giving the printer name fewer characters
avoids this.
In CUPS 1.7 httpConnect() and httpConnectEncrypt() were deprecated and
replaced with httpConnect2(). This checks if httpConnect2() is available
and if so, replaces the uses of the deprecated functions.
In the CUPS source code, httpConnect() and httpConnectEncrypt() are now
wrappers around httpConnect2(), so we make sure to use the same
arguments as in the CUPS source code so the two code paths are sure to
be identical:
2c030c7a06/cups/http.c (L412)2c030c7a06/cups/http.c (L477)
In https://bugzilla.gnome.org/show_bug.cgi?id=696766 it was decided to
build with _PPD_DEPRECATED defined, to suppress the warnings about the
deprecation of the CUPS PPD API, since there was no full replacement
yet. Unfortunately this didn't make it intact through the port to Meson,
so fix it here.
The previous code would have leaked all the UpDevice objects because no
free function was set on the returned GPtrArray.
Since we already depend on upower-glib 0.99.8 which is the release in
which get_devices() was deprecated and get_devices2() was introduced, we
can simply switch to get_devices2() which does set a free function on
the returned GPtrArray, stopping the leak.
