Right now we set a subtitle of "" when the port is invalid. This
leads to the "Port" title getting center aligned vertically.
We also keep the port sensitive even when it's empty and there is
little value in copying it.
This commit changes the subtitle to " " so it serves as a
placeholder to keep the "Port" title top aligned. This commit
also makes the port insensitive in that case so the user never
copies the placeholder space.
Closes: https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/2934
Right now we set a subtitle of "" when the port is invalid. This
leads to the "Port" title getting center aligned vertically.
We also keep the port sensitive even when it's empty and there is
little value in copying it.
This commit changes the subtitle to " " so it serves as a
placeholder to keep the "Port" title top aligned. This commit
also makes the port insensitive in that case so the user never
copies the placeholder space.
Previously, events from accountsservice would be used to change the
state of the AdwNavigation in ways that don't correspond to the event
that happened. For example, deleting an account would pop the top page
off the stack, even if that page didn't belong to that account.
Especially buggy is the behavior of replacing the currently visible page
with the contents of an account that just changed. systemd-homed changes
the user record whenever authentication happens successfully. So, user
Foo might be trying to edit user Bar, type in their password at the
polkit prompt, and end up looking at a broken version of their own
settings page again: the title would be "Bar", there'd be no list of
users, and hitting the back button would take Foo back to the same
settings page they're currently looking at.
This commits refactors the handling of the accountsservice signals to
fix all the bugs
Fixes https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/2911
650p is way too much and makes the bottom bar to
appear when there is still space available in
the headerbar. Use 450p so the bottom bar will only
appear when the headerbar doesn't have enough
space available.
Follow-up from a4224d9e21
When the window width is small, the view switcher
labels get very ellipsized, making them difficult
to read.
Use AdwBreakpointBin to show the view switchers
at the bottom at the page (like in the Mouse & Touchpad
panel) when the window width is small.
- Use halign correctly ("left" and "right" aren't
valid values) and use hexpand in the
"Verify Encryption" button, since halign needs
hexpand set to "true"
- Ditch the spacer box, as it causes the page contents
to not scale correctly when the window width is small
- Add can-shrink property to "Generate New Password" and
"Verify Encryption" buttons so they can shrink when
the window width is small
- Also, set the margin-top only in the GtkBox to avoid
redundancy
Closes https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/2933
- Fix the activatable-widget from port_row,
which was pointing to copy_address_button
- Remove a11y labels from copy buttons
because Orca also reads tooltips
- Remove .property style class from entry rows
since this class is only meant to be used
in AdwActionRow and AdwExpanderRow[1]
- Set the copy buttons as suffix widgets
directly, because wrapping them in a GtkBox
is unnecessary
- Fix indentation
[1] https://gnome.pages.gitlab.gnome.org/libadwaita/doc/main/style-classes.html#property-rows
- Fix the activatable-widget from port_row,
which was pointing to copy_address_button
- Remove a11y labels from copy buttons
because Orca also reads tooltips
- Remove .property style class from entry rows
since this class is only meant to be used
in AdwActionRow and AdwExpanderRow[1]
- Set the copy buttons as suffix widgets
directly, because wrapping them in a GtkBox
is unnecessary
- Fix indentation
[1] https://gnome.pages.gitlab.gnome.org/libadwaita/doc/main/style-classes.html#property-rows
Show the full VRR (Variable Refresh Rate) range of a monitor in the
refresh rate expander row when a variable refresh rate mode is
selected and the minimum refresh rate of the monitor is known.
The code currently assumes that if the fingerprint is non-null it's
valid, but gnome-remote-desktop can also set the fingerprint to an
empty string if there's no certificate yet.
This commit makes sure to handle both NULL and empty fingerprints.
If the username and password get set then the certificate fingerprint
isn't updated even if that lets the service start.
This commit fixes things by doing a round trip through the service after
setting the username/password to
1. Make sure the change actually worked
2. Trigger our fingerprint dialog setting code that currently only runs
on reading username/password
Remote session support needs to be able to start systemd units
and also manage gnome-remote-desktop.
This commit adds a polkit policy to accomodate both subactions
with one overarching action that only needs to be unlocked one time.
This change will update the UI and internal reference names to use
the "Desktop Sharing" term, which refers to sharing a user's screen
with remote connections.
This differentiation is being introduced now so that we can avoid
conflict with "Remote Session", which will refer to headless/dedicated
remote desktop sessions.
Mockups
https://gitlab.gnome.org/Teams/Design/settings-mockups/-/raw/master/remote-desktop/remote-desktop.png
See #2827
The Proxy row shows its value as subtitle, which is
inconsistent with other rows where the subtitle describes
the row title/what the row is.
Show the proxy value as a CcListRow secondary-label to fix
such issue.
Reference:
cfb7cae4a9
Currently, the generic "Failed to log into domain" error is shown when
joining domain with invalid hostname. It is hard for user to guess what
is going on. Let's fail with "That hostname didn’t work" instead.
Related: https://gitlab.freedesktop.org/realmd/realmd/-/merge_requests/39
Currently, the previous cc_realm_manager_discover calls are not cancelled
when validating domain field. This might lead to the situation when valid
domain might be mislabeled as invalid and vice versa. Let's simply cancel
the cancellable to avoid this situation.
The `source_object` parameter for the callback functions called from
the `realm_join_as_user` and `real_join_as_admin` resp. `realm_join_as_owner`
functions is not `CC_REALM_OBJECT` type as one would expect, but
`CC_REALM_KERBEROS_MEMBERSHIP` type. Let's modify the code to ensure that
it is `CC_REALM_OBJECT` type.
The implementation is based on the cc-add-user-dialog.c codes, but it is
not a pure copy&paste. It has to be changed to fit the skeleton and reflect
changes from the mockup. I've also tried to simplify the code where possible.
It also fixes some memory leaks.
- Make all buttons as suffix widgets,
which makes unnecessary to wrap them in
a GtkBox that is set as a suffix widget
- Don't set a margin end to the Wi-Fi
icon[1]
- Remove halign from the buttons, because
it isn't necessary
- Remove halign from the buttons, as it
isn't necessary because halign works only
if hexpand is set to TRUE
- Add margin end to the spinner so it
doesn't appear misaligned with the other
widgets at the end of the row
[1] In latest mockups, the margin between
the Wi-Fi icon and network name is also
shorter: cfb7cae4a9
This allows us to open subpages of the "System" panel. The arguments/
parameters are expected to match the panel's subpage tags (as in
AdwNavigationPage.tag).
With this, a future change could reintroduce desktop files for the
subpages, and just change the Exec line to accomodate the new format.
Graphics labels under system details did not have enough padding
between them similar to other labels in the panel
Add a default spacing of 10 to children rows of gtk_box that sets the
graphic rows dynamically to fix the issue
Fixes https://gitlab.gnome.org/GNOME/gnome-control-center/-/issues/2916
