Carlos Garnacho [Wed, 6 May 2020 08:57:54 +0000 (10:57 +0200)]
shell: Check we are in a systemd environment before starting systemd units
Avoid starting/stopping the gsd-xsettings systemd unit if the shell itself
was not started via systemd. In the lack of a user session manager, we
have no means to neatly start/stop services, so should rely on the good
ol' gnome-session to do this for us.
This changes the return value meaning slightly, TRUE means "service did
start", FALSE otherwise. The error is only raised if we ought to start,
but it produced an error somehow.
shell-app: Add PrefersNonDefaultGPU support to shell_app_launch()
Read the "PrefersNonDefaultGPU" key in desktop files to figure out
whether the application prefers running on the discrete GPU, or the
default GPU, and apply that.
Update the "Launch..." contextual menu to allow launching on
the default GPU if the application "prefers [the] non default GPU".
Ting-Wei Lan [Sun, 3 May 2020 15:26:39 +0000 (23:26 +0800)]
ibusManager: Fix the panel after porting to Promises
Commit 764527c8c9b7659901eb6296a6859ae2b0eabdb8 not only ports this file
to Promises but also changes the behavior of _initPanelService method.
Instead of always calling _updateReadiness when _panelService is ready,
it only calls it when get_global_engine_async succeeds.
The only callers of _updateReadiness are _initEngines and
_initPanelService. Assume that _initEngines completes first. Its
_updateReadiness call keeps _ready as false and it is expected for
_initPanelService to change it to true. However, since
get_global_engine_async fails because there is no active engine,
_initPanelService never calls _updateReadiness. Therefore, all setEngine
calls do nothing because _ready is false, and the input method panel
never shows. Users are unable to use any input method even if they can
see that ibus-daemon is already running.
Fix the issue by changing it back to the old behavior.
Jonas Dreßler [Thu, 30 Apr 2020 20:23:52 +0000 (22:23 +0200)]
st/theme: Unref CRStylesheet after removing old theme nodes
Since we now remove all theme nodes on a stylesheet change (ie.
StTheme's "custom-stylesheets-changed" signal) instead of only
invalidating them, those nodes may not be accessed anymore as soon as
"custom-stylesheets-changed" is emitted.
It turned out though that when comparing them to the newly generated
nodes in `st_widget_recompute_style()` using the
`st_theme_node_paint/geometry_equal()` functions, the properties of the
old nodes will still be accessed, causing a crash since the
CRDeclarations are already freed.
To fix that, keep the reference to the CRStylesheet, which owns the
CRDeclarations used by the theme nodes, around a bit longer, so it's
still possible to access the CRDeclarations inside the
"custom-stylesheets-changed" signal handler. This allows us to compare
the old theme nodes to the new ones since the CSS properties of both are
still valid.
gjs has gotten less forgiving about missing getters/setters, and
commit 6aa1b817 missed the missing getter in the base policy class.
Most notifications use a policy subclass that already provides a
getter, but at least Main.notify() and friends don't; unbreak them
by fixing the base class.
Jonas Dreßler [Tue, 28 Apr 2020 21:26:11 +0000 (23:26 +0200)]
main: Unset the right prevFocus actor after the focus stack got shifted
When a modal that's not on top of the modalActorFocusStack gets popped,
we shift the focus stack as described in popModal() to ensure the chain
remains correct. That however destroys the association of a modal actor
and its prevFocus actor on the focus stack, because the prevFocus actors
are now moved to different entries of the stack.
Now when a prevFocus actor gets destroyed, we don't handle that case
correctly and search for the modal actor that was associated with the
prevFocus actor before the stack was shifted, which means we end up
unsetting the wrong prevFocus actor.
So fix that and search the stack for the prevFocus actor which is being
destroyed instead to unset the correct entry.
Thanks to Florian Müllner for figuring out the actual issue and
proposing this fix.
Jonas Dreßler [Sat, 25 Apr 2020 11:17:13 +0000 (13:17 +0200)]
st/theme-context: Also remove theme nodes on stylesheet changes
StThemeNodes are built around the assumption that they're "immutable",
that means they are created once with certain parameters (that resolve
to certain css properties) and then replaced with new ones in case those
parameters changed.
Changes to the internal information stored by theme nodes (i.e. the css
properties, font names or the cached scale-factor) are not all handled
the same though: For changes to the font or the scale-factor we remove
all theme nodes from the cache and let the widgets which are on stage
generate new theme nodes. For changes to the css properties/the
stylesheet, we invalidate the properties of all theme nodes but keep
them in the cache using `_st_theme_node_reset_for_stylesheet_change()`.
So be a bit more consistent and handle changes to the css-properties/the
stylesheet stored by StThemeNodes the same way as changes to the font or
scale-factor.
Carlos Garnacho [Sat, 25 Apr 2020 16:31:11 +0000 (18:31 +0200)]
st: Ensure to reset all widget theme nodes
Theme node invalidation stops at unmapped widgets, and widgets
that forget to chain up to the default ::style-changed implementation.
This may leave stale nodes that were invalidated on
StThemeContext::changed, but are still set on widgets, and maybe
used for CSS property lookups.
Make sure that theme node invalidation happens always by moving
propagation outside the vfunc, and ensure the theme nodes are reset
across the full actor tree. Emission of ::style-changed, and obtaining
a new theme node may be delayed till when the actor is mapped.
The theme node is also cleared after unparenting an actor to avoid
stale references.
This ensures that all widgets get their theme node cleared after
stylesheet changes, instead of maybe being left with a theme node
that's been cleared of all its properties.
cleanup: Don't linebreak before closing parentheses
Otherwise recent versions of eslint want "dangling" commas,
which is at least ugly considering that most functions don't
allow adding arguments at leasure.
overviewControls: Remove transitions before updating adjustment value
The workspace switcher blocks state updates while the indicator is
animating. Since commit 9c1940ef9db5 the indicator is considered to
be animating when the workspace adjustment's value doesn't equal the
active workspace.
There is one case though where this breaks badly: When a workspace
is inserted before the active one, the adjustment's upper and value
properties are changed without transitions. But if that change happens
while there's an ongoing transition to the previously active workspace,
the value gets out of sync with the active workspace and we end up
blocking state updates indefinitely.
Fix this by removing any transitions before setting the adjustment
value.
Commit d76162c1c047c removed the ability to pass an extension UUID to
the Extensions app, when we moved the dialog to a portal and made
gnome-shell use it instead of spawning the extensions app.
However that missed that many extensions called out to the app to
open their own prefs.
While extensions are encouraged to switch to the new openPrefs()
convenience method added in commit 8030d9ad323e, restore the old
behavior with a small script under the old gnome-shell-extension-prefs
name that either calls out to the portal or launches the app.
We override the :visible property for the keyboard actor, but don't
provide a corresponding setter. The property is therefore read-only
on the javascript level, and any attempt to set it will fail.
Jonas Dreßler [Thu, 12 Mar 2020 12:37:53 +0000 (13:37 +0100)]
altTab: Set allocation before allocating children
Just as with the last commit, we should not break the assumption made by
Clutter that parents have their allocation set before their children get
allocated, so fix that here, too.
In this case we have to fix it by chaining up to the parent vfunc
override and updating the allocation once more before allocating the
`this._label` child.
Jonas Dreßler [Tue, 10 Mar 2020 10:05:13 +0000 (11:05 +0100)]
boxpointer: Call set_allocation before allocating child
It's important to update the allocation of the parent before allocating
its children, it's an assumption we make in a lot of places.
This broke resource scale calculation for boxpointers and their
children when multiple monitors with different scales are used and the
primary monitor is not positioned at x=0, y=0.
Milan Crha [Thu, 2 Apr 2020 07:24:28 +0000 (09:24 +0200)]
calendar-server: Improve performance by properly using ECalClientView
The previous code always restarted whole ECalClientView when it received
any changes in it, which could sometimes lead to constant repeated restarts
of the view.
Jonas Ådahl [Tue, 21 Apr 2020 14:13:20 +0000 (16:13 +0200)]
js/main: Inhibit remote access when depending on session mode
The session mode determines whether the screen casting should work or
not, but until now only dealt with the built in screen cast, not the
ones using PipeWire. Add the newly added API for inhibiting remote
access when the session mode says screencasts are not allowed.
The template has been updated to rewrite the manifest to use the
checked out tree, so it's no longer necessary to pass additional
build arguments (which are now ignored) or generate translations
before the build (it's already in the manifest).
Andy Holmes [Thu, 23 Apr 2020 23:06:36 +0000 (16:06 -0700)]
JS: migrate from the global `window` to `globalThis`
As of mozjs68 (gjs-1.64) `globalThis` is recommended over `window` and
it makes more sense in this context anyways. Migrate the few instances
of `window` we use and adjust the eslint configuration.
`window` will continue to resolve to `globalThis`, so this won't affect
extensions or other downstream users.
We track messages so that we can account for just added and removed
events instead of having to rebuild the entire list, however it's
also possible that the time or summary of an existing event changed.
Account for that by updating existing messages in-place.
Jonas Dreßler [Wed, 22 Apr 2020 15:27:54 +0000 (17:27 +0200)]
authPrompt: Grab key focus when making entry sensitive
We currently let the entry of the autoPrompt grab the key focus inside
setQuestion(), which is called from _onAskQuestion(), which is the
callback of the "ask-question" signal.
It seems that the "ask-question" signal isn't emitted again right after
the password-check failed, but a few seconds after that. Since we get
the "verification-failed" signal earlier than "ask-question" (right
after we know the check failed) and we also get a hint whether the entry
should be usable again with the canRetry argument, we can also grab key
focus to in the same step.
So do that by grabbing key focus when making the entry sensitive.
The additional function UnlockDialog.addCharacter() is only used at one
place, so we can simply remove it and call AuthPrompt.addCharacter()
directly. The AuthPrompt is shown right before that anyway.
Will Thompson [Sat, 25 Apr 2020 09:54:55 +0000 (10:54 +0100)]
systemActions: include ASCII alternatives in search index
With this change, "eteindre" matches "éteindre" (the French translation
for "power off"), consistent with search for applications where "e"
matches "é".
Matthew Leeds [Fri, 24 Apr 2020 21:42:52 +0000 (14:42 -0700)]
screenShield: Don't inhibit suspend during initial setup
Normally, we inhibit suspend while locking the screen. But in the
session mode used for gnome-initial-setup locking is not supported, so
in that case this inhibit call is pointless and should be avoided.
Without this patch you get the following error when you suspend and
resume during initial setup:
JS ERROR: Error getting systemd inhibitor: Gio.IOErrorEnum:
GDBus.Error:org.freedesktop.login1.OperationInProgress: The operation
inhibition has been requested for is already running
_promisify/proto[asyncFunc]/</<@resource:///org/gnome/gjs/modules/core/overrides/Gio.js:435:45
Matthew Leeds [Fri, 24 Apr 2020 00:42:16 +0000 (17:42 -0700)]
screenShield: Fix use of null this._dialog
On Fedora 32 if you close the laptop lid during gnome-initial-setup,
gnome-shell hits this error:
JS ERROR: Exception in callback for signal: prepare-for-sleep: TypeError: this._dialog is undefined
_resetLockScreen@resource:///org/gnome/shell/ui/screenShield.js:434:9
activate@resource:///org/gnome/shell/ui/screenShield.js:571:14
lock@resource:///org/gnome/shell/ui/screenShield.js:617:14
_prepareForSleep@resource:///org/gnome/shell/ui/screenShield.js:219:22
_emit@resource:///org/gnome/gjs/modules/core/_signals.js:133:47
_prepareForSleep@resource:///org/gnome/shell/misc/loginManager.js:198:14
_emit@resource:///org/gnome/gjs/modules/core/_signals.js:133:47
_convertToNativeSignal@resource:///org/gnome/gjs/modules/core/overrides/Gio.js:169:19
This is because _ensureUnlockDialog() hit its first early return. So
return early from activate() in that case, so this._dialog doesn't get
used while it's null.
appDisplay: Return the parent class' result in overrides
StButton returns CLUTTER_EVENT_STOP in various circumstances, but
AppIcon throws that away and returns CLUTTER_EVENT_PROPAGATE even
when it should stop.
Return the parent class' result instead of CLUTTER_EVENT_PROPAGATE.
Will Thompson [Tue, 21 Apr 2020 12:44:57 +0000 (13:44 +0100)]
test-theme: Check return value of chdir()
I noticed the following warning while building the .deb package for
gnome-shell on Endless OS:
../src/st/test-theme.c: In function ‘main’:
../src/st/test-theme.c:549:3: warning: ignoring return value of ‘chdir’, declared with attribute warn_unused_result [-Wunused-result]
chdir (cwd);
^~~~~~~~~~~
(Of course this is very unlikely to fail in practice.)
Philip Chimento [Tue, 21 Apr 2020 00:25:14 +0000 (17:25 -0700)]
messageTray: Make NotificationPolicy properties read-only
These properties are never written; in the base class they are always
their default values, and in the subclasses the getters are overridden.
This will be necessary because GJS is adding checks to make sure that
readable properties always have a getter, writable properties always
have a setter, and that the variations of camelCase/snake_case are
handled correctly. It's supposedly backwards compatible, but that
assumes that code is not doing things like forgetting a setter on a
writable property. (If the missing setter had ever been called, it might
have led to a crash, which is why we've made this change.)
This is the minimally invasive patch which should work with both older
and newer versions of GJS. If you decide to require GJS 1.65.2, then
you'll also be able to remove the getters from NotificationPolicy as
well.
We want the spinner to be centered with regard to the entry, but
constraining the height breaks that:
1. clutter_actor_allocate() is called with the available size
2. clutter_actor_update_constraints() then adjusts that according
to the constraints
3. clutter_actor_adjust_allocation() applies the margin/expand/align
properties.
The issue there is that 2. reduces the allocation to the desired size,
so there is no more extra space to distribute in 3.
We can fix this by either constraining everything (and rely on the
cancel button's alignment) or limit the constraint to the width. The
latter seems more appropriate, given that the constraint is only used
to center the entry horizontally.
polkitAgent: Return GLib.SOURCE_REMOVE on timeout callback
resetDialog is used (also) as GLib timeout function, so while just returning
an undefined value will stop the timeout anyways, it's just cleaner to
use the GSource expected return values
Jonas Ådahl [Wed, 15 Apr 2020 09:16:32 +0000 (11:16 +0200)]
js/ui: Always use namespace for animation modes
Clutter.Animation doesn't contain any animation modes, they live in
Clutter.AnimationMode. The places we did `Clutter.Animation.WHATEVER`
just evaluated to `undefined`. Thus, use the correct namespace for the
animation mode enums.
Revert "app-cache: Fix cache for folder translations"
Existing folders as created by gnome-software (including the
default ones) all have the .directory suffix, so dropping
the suffix from the keys broke those folders.
This will simplify future work where the PaginatedIconGrid will be
an implementation of StScrollable, in which case we'll be able to
simply remove the StBoxLayout from there.
We currently always ignore the minimum width of the hint actor and
request/allocate the preferred width. This can be problematic with
labels with long text, where we should rather ellipsize the text
than allow the entry to grow indefinively.
Florian Müllner [Thu, 7 Nov 2019 11:46:47 +0000 (12:46 +0100)]
extensions-tool: Add --template create option
We now have everything in place to allow users to pick a template
other than the default one (although at the moment it's the only
one we have), so add a corresponding option.
Florian Müllner [Fri, 8 Nov 2019 23:20:08 +0000 (00:20 +0100)]
extensions-tool: Add template metadata
When we allow users to choose between different templates, we should
provide some context for each template to facilitate that choice.
Add that metadata in the form of a .desktop file, which allows us to
specify name and description, and is well supported by our translation
infrastructure.
Florian Müllner [Wed, 6 Nov 2019 09:06:11 +0000 (10:06 +0100)]
extensions-tool: Prepare for alternative templates
The template used when creating a new extension is intentionally
minimal, as the sample code in the old extensions-tool proved to
be annoying more often than not.
However as we support per-command options, we don't have to limit
ourselves to a single template, and can offer alternatives for
common use cases.
To prepare for that, namespace the existing template by moving it
into a subfolder.
Florian Müllner [Sun, 10 Nov 2019 02:00:37 +0000 (03:00 +0100)]
extensions-tool: Separate create metadata when prompting
When using the create command's --interactive option, we prompt for
any metadata that wasn't passed on the command line. As every prompt
is preceded by a short multi-line description, it is hard to follow
when everything is lumped together.
Improve legibility by separating all prompts by newlines.
extensions-tool: Handle NULL input when prompting for metadata
g_data_input_stream_read_line_utf8() may return NULL, for example
when interrupting the prompt with ^D. Handle that case and keep
prompting until we got a line.
Philip Withnall [Mon, 30 Mar 2020 11:40:11 +0000 (12:40 +0100)]
search: Hide search providers which are blacklisted by parental controls
If a search provider is installed by an app which is blacklisted for the
current user by their parental controls, don’t show it or results for
it.
Currently, this only filters ‘remote’ (not built-in to the shell) search
providers. This seems fine for now; in future it could be expanded to
also filter built-in search providers, if any of them end up needing to
be filtered.
No corresponding changes need to be made `remoteSearch.js`, because the
results of `loadRemoteSearchProviders()` are filtered in `search.js`.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/465
projects / gnome-shell.git / log