Sam Hewitt [Thu, 11 Jul 2024 16:53:04 +0000 (14:23 -0230)]
lookingGlass: Clean up the stylesheet and improve visuals overall
- shuffle stylesheet around for readability
- change names and classes of elements to be consistent
- added style classes to window list elements so they can be styled
- make sure all elements follow OSD style even in light theme
- clean out some old styles
lookingGlass: Include ScrollView in inspector styling
All styling currently happens on the ScrollView child, which
means that instead of a scrollbar inside the dialog to move
the content, we have a scrollbar next to the dialog that moves
the dialog as a whole.
Fix this by simply moving the style classes to the up-most parent.
Now that we no longer provide a portal implementation ourselves,
the only expected callers are the GNOME portal implementation
and (since recently) the remote desktop daemon.
Make that explicit and restrict callers to these senders, like
we do for other D-Bus APIs.
Jonas Dreßler [Sun, 22 Jan 2023 18:29:23 +0000 (19:29 +0100)]
screencastService: Add gstreamer pipeline for h264 encoding on hardware
Try h264 on intel and AMD hardware using VA-API, using vapostproc
to do the color conversion using specialized hardware. We're not adding
a vaapi pipeline using the color conversion GL shader here, as vapostproc
color conversion should be better suited for anything needed by vaapi
based encoders/decoders.
A capsfilter is used to enforce the use of DMAbuf (memory:DMABuf) with
the new explicit modifier API (format=DMA_DRM).
Also, we're using the new va* gst elements instead of the vaapi* ones
here, as the vaapi* elements are deprecated, while the va* elements are
actively maintained.
Florian Müllner [Sat, 29 Jun 2024 19:21:29 +0000 (21:21 +0200)]
popupMenu: Center-align switches
The default align of FILL was fine with the asset-based switches,
as the image would still keep its fixed height. But after the switch
to a pure CSS implementation, the alignment now results in stretched
switches. Fix this by explicitly center-aligning the widget.
Jonas Ådahl [Fri, 28 Jun 2024 21:01:12 +0000 (23:01 +0200)]
tests/fittsy: Await leisure before checking fittsiness
Not until we've waited for leisure can we make precise assumptions about
otherwise animated actors sizes and positions, so do that before
checking fittsiness.
This should fix the flakyness of the fittsy test case.
Philip Withnall [Fri, 28 Jun 2024 16:06:31 +0000 (17:06 +0100)]
shell-util: Fix a memory leak of a floating variant on error
If `params` is floating when provided by the caller, and if one of the
early return paths is taken from this function, the call into GIO never
sinks and takes ownership of the `GVariant`, and hence it’s leaked.
Always explicitly sink the input variant as soon as the function is
called, then it can’t be leaked.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/3396>
workspace: Initialize overview windows correctly in constructor
Currently overview windows added in workspace constructor lack monitoring
the skip-taskbar property. All other windows added after this have
this monitoring.
This uses existing code to correctly add the signal and removes all redundant
code.
It also insures all transients are handled correctly on initialization using
the existing code.
Keyu Tao [Thu, 20 Jun 2024 15:09:27 +0000 (23:09 +0800)]
workspacesView: Make active workspace above siblings
If there's a window flowing over the left edge in a workspace, the
animation of showing or hiding overview in the left adjacent workspace
would glitch, as the window would appear or disappear suddenly.
This commits tries to fix it by making sure the active workspace are
above other workspaces when updating workspaces or active workspace
changed in workspacesView.
Jonas Dreßler [Thu, 6 Jun 2024 15:13:27 +0000 (18:13 +0300)]
st/theme-node-drawing: Factor in border radius in update_resources()
The reduced border radius from st_theme_node_reduce_border_radius() depends
on the allocation size, and when it changes, has_large_corners can change.
has_large_corners is used to determine whether to pre-render using cairo,
so that means a change to the allocation size can influence whether the
pre-render using cairo should happen or not.
On changes to the allocation size we don't want to call render_resources()
though, but just update_resources() instead. And since we know now that
changes to the allocation size might cause pre-rendering to get enabled
or disabled, we need to do the "should pre-render" checks in
update_resources(), too.
So factor out those "should pre-render" checks into a new function
st_theme_node_maybe_prerender_background(), and then call that function from
both render_resources() and update_resources().
This fixes rendering artifacts with borders or rounded corners after changes
to the actor size.
Balló György [Thu, 20 Jun 2024 12:59:34 +0000 (12:59 +0000)]
extensions-app: Link to GitLab Issues instead of New Issue URL
When not logged in, the New Issue URL only redirects to an unhelpful
login page while the Issues page allows you to search for existing
issues and still offers a "New Issue" button.
Switch to render nodes. Most of the changes follow the pattern of
replacing a cogl_framebuffer_draw_*() call by (1) creating a paint
node, and (2) calling the corresponding clutter_paint_node_add_*()
API.
StWidget now overrides ClutterActor.paint_node(), instead of paint.
All subclasses that overrided the paint vfunc also are changed to
override paint_node.
The entry point for CSS rendering, st_widget_paint_background(), now
receives a root ClutterPaintNode, and propagates it to the rendering
machinery.
gdm/util: Cleanup CredentialManager on successful verification
CredentialManagers emit a user-authenticated signal with a token
upon a successful authentication. This marks the service as preempting
and will be used for the next login/unlock request. However, the
preempting service is not cleaned after a successful verification.
As a consequence, it will be picked on a second unlock request.
This doesn't happen after a succesful login request because there
is a transition from the gnome-shell process owened by GDM to the
one owned by the user.
Clean the preempting service on a successful verification, like we
do when the PAM conversation stops.
Florian Müllner [Thu, 6 Jun 2024 13:31:34 +0000 (15:31 +0200)]
shell/app: Remove some dead code
`g_desktop_app_info_launch_uris_as_manager_with_fds()` was added
in GIO 2.57.2. We now depend on at least 2.79.2, so we no longer
need a fallback path for the case the function isn't available.
Florian Müllner [Thu, 13 Jun 2024 13:43:51 +0000 (15:43 +0200)]
ci: Also store gvc subproject in artifacts
After the subproject moved to a meson wrap, its checkout is no
longer managed by gitlab.
While meson will take care of checking out required submodules
during setup, this will not happen for `test` or `dist`, so store
the checked out code in artifacts.
Jonas Dreßler [Fri, 31 May 2024 12:58:13 +0000 (15:58 +0300)]
overviewControls: Make sure thumbnails expandFraction is bound to visibility
We set the expand-fraction property of the workspace thumbnails on changes
to their should-show property. The should-show property defaults to true, and
the expand-fraction property defaults to 1. The should-show property gets
updated in the constructor of the ThumbnailsBox though, and it might become
false (that is the case when starting up the shell) during construction.
In this case we're not yet listening to "notify::should-show" in the
ControlsManager, and therefore forget to update the expand-fraction
accordingly. Fix that and ensure the expandFraction is always set to the value
we expect after hiding or showing the thumbnails (we call _updateThumbnailsBox()
from _update(), which we call in the constructor).
This should be the proper fix for the problem that 9bf550da8 (which caused
the workspace thumbnails to never show) originally tried to fix.
Fixes: 9bf550da88 ("overviewControls: Set expandFraction for ws thumbnails to 0 when hidden") Closes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/7668
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/3323>
Jonas Dreßler [Thu, 9 May 2024 08:03:23 +0000 (10:03 +0200)]
portalHelper: Use default_size instead of set_size_request for size
GtkWidget.set_size_request() enforces a minimum size, while
GtkWidget.set_default_size() simply sets the default size.
The docs of set_size_request() say "In most cases,
gtk_window_set_default_size() is a better choice for toplevel windows", and
in our case it doesn't seem necessary to prohibit the window from having a
smaller size, so switch to using the default_width and default_height
properties.
There are cases where it makes sense to have fillPreferencesWindow be
async, so await it. One such case is when using .ui files from resources
with (GTK) templates. When extension developers load the resource in
`fillPreferencesWindow` and then import the js file that registers the
classes, there will be an error at the moment that there is no UI
provided. Similarly, make getPreferencesWidget async optionally.
Philip Withnall [Tue, 28 May 2024 17:05:10 +0000 (18:05 +0100)]
messageTray: Ensure _updateDatetimeId idle callback is cancelled
If the idle callback is not explicitly removed when the notification is
destroyed, it may be invoked after destruction, and try to modify the
notification after it’s been freed.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/3352>
Philip Withnall [Tue, 28 May 2024 17:08:27 +0000 (18:08 +0100)]
shell-camera-monitor: Fix warning about strict aliasing
Changing type when casting a double pointer is not strictly allowed in
C. In order to fix compilation with `-fstrict-aliasing`, don’t use
`g_clear_pointer()` for clearing this pointer where the type expected by
the destroy notify function does not match the type stored in the
`ShellCameraMonitor` struct.
This introduces no functional changes.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/3351>
Florian Müllner [Tue, 28 May 2024 16:25:15 +0000 (18:25 +0200)]
ci: Bump ci-fairy template
The new version allows @users.noreply email addresses by default,
which seems like something we should allow at least for small
drive-by contributions.
js/environment: Always use Shell.util_spawn_async functions if possible
If no child setup is provided then all the shell extensions that use
GLib.spawn_async should actually use the shell spawning utils since we
are supposed to always restore the default nofile rlimit on launched
children.
shell-util: Add async-signal-safe wrappers for GLib.spawn_async implementations
In the shell code we often use GLib.spawn_async to launch processes
with a GSpawnChildSetupFunc implementation in JavaScript to reset the
mutter nofile rlimit in the new child process.
However, this is highly unsafe to do because this implies that the child
setup function code is executed in gjs where a lot of allocations are
done and even more not-async-signal-safe code is executed, in fact
leading to dead-locks as reported in the past.
To prevent this, declare a new functions that do the same of the GLib
counterpart but without providing a GSpawnChildSetupFunc that is instead
implemented in the C-side doing the cleanup that mutter requires without
allocations or async-signal-unsafe code.
shell-app: Do not allocate memory during GSpawnChildSetupFunc
The child context setup function used as launch GSpawnChildSetupFunc
uses g_object_get that internally potentially allocates memory, making
it not async-signal-safe and so not something that is safe to use in
between fork and exec, so just use the simpler getter here.
Note that the current implementation of app_child_setup() is safe
already as sd_journal_stream_fd, dup2 and close are so.
Florian Müllner [Thu, 23 May 2024 17:11:42 +0000 (19:11 +0200)]
ci: Hook up release-module
In the future, the module will automate uploading the release
tarball. We already use the CI pipeline to generate the tarball,
so it's easy to hook up the module and provide some testing
before the module goes into production.
A feature that works for server side decorated windows only that
only adds slightly more convenience over the regular "move" action that
is still there.
Keeping it around would complicate some cleanups in upstream Mutter
so just drop it.
Sebastian Keller [Tue, 21 May 2024 23:45:44 +0000 (01:45 +0200)]
style: Don't use different offset for the HC app grid running indicator
The offset in high contrast mode was moving the dot too close to the
text making the spacing between the text, dot and hover highlight appear
visually unbalanced. Also it was making it more likely to overlap
descenders in the text.
Sebastian Keller [Wed, 22 May 2024 00:04:53 +0000 (02:04 +0200)]
style: Shift dash running indicator dot up a pixel in high contrast mode
With the dot now being placed correctly at the bottom of the hover
highlight it is now also overlapping the 1px inset shadow used as
outline in high contrast mode. To avoid this, shift the dot up
accordingly.
Sebastian Keller [Sun, 17 Mar 2024 11:43:03 +0000 (12:43 +0100)]
appDisplay: Set running indicator dot offset via CSS
The offset was given as a hardcoded (physical) pixel value and did not
take scaling into account. This lead to it being shifted closer towards
the icon as the scale increased.
This now replaces the hardcoded value with a CSS property which
automatically includes the scale factor.
Further this allows simplifying some calculations that previously were
trying to counteract the hardcoded offset using margins by using the
intended offset directly.
With this the dot in the dash is also now placed at exactly the bottom
of the hover highlight, while previously there was an unintentional 1px
space.
Florian Müllner [Mon, 20 May 2024 18:22:11 +0000 (20:22 +0200)]
workspaceThumbnail: Don't rely on allocation to compute scale
We currently compute the target scale from the allocated height,
which means that it only becomes available after thumbnails have
been allocated at least once at their expanded height.
As the minimap is initially hidden, this only happens after it is
expanded for the first time, which means the corresponding transition
is not animated.
In order to allow for a fix, compute the target height ourselves
to allow the scale computation to work independently from the
expand factor.
Florian Müllner [Tue, 14 May 2024 23:55:40 +0000 (01:55 +0200)]
authPrompt: Consume cancel key event
The auth prompt currently propagates all key presses, even the
Escape press that is used to cancel it.
On the lock screen that means that the same event that cancels
the prompt (and switches back to the clock) is *also* propagated
to the handler that activates the prompt on key press.
That handler doesn't do anything when the prompt is already visible,
which is the case when the transition to the clock is animated.
However when animations are disabled, canceling the prompt will
result in a new prompt getting created immediately, and the login
screen is stuck on the prompt.
Fix this by not propagating key events that are used to cancel
the prompt.
projects / gnome-shell.git / log