At the moment a session mode either allows extensions or it doesn't.
If it allows extensions, then the entire available list of
configured extensions get enabled as soon as the session mode is
entered.
Since enabling or disabling extensions is an all or nothing situation,
the code tracks whether extensions are already enabled when entering
the session mode, and if so, avoids iterating through the extension list
needlessly. It does this using a boolean named _enabled.
In the future, the extensions themselves will be given some say on
whether or not they should be enabled in a given session mode. This
means, the configured extension list may contain extensions that
shouldn't be enabled for a given session mode, and the _enabled boolean
will no longer be appropriated.
This commit drops the _enabled boolean optimization.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1967>
If a user hits escape twice really fast when coming back to
their machine to unlock it, they made end up getting presented
with a non-functional unlock screen that doesn't show their
user icon and doesn't ask for a password.
This is because showPrompt assumes that if an auth prompt already
exists, it's ready to go. That may not be true, if it's in the
process of getting torn down at the time because it's in the middle
of a cancel animation.
This commit solves the problem by ensuring the auth prompt is always
in a fresh reset state before showing it.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1999>
If the the unlock dialog gets finished before an auth dialog is
created, the code currently creates one just to tell it to finish.
This commit changes the code to skip creating the auth dialog in
that case.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1999>
Commit dd2cd6286c restricted callers of the screenshot methods to
portal implementations, gnome-settings-daemon and gnome-screenshot.
That restriction does make sense for the actual screenshot methods,
but `PickColor` is actually used by GTK in its color picker (and
therefore may be called from arbitrary applications).
Fix this by unrestricting access to `PickColor` again. Considering that
the method is always interactive, it's not very privacy/security-sensitive
anyway.
https://gitlab.gnome.org/GNOME/gtk/-/issues/4283
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1990>
Previously we used `get_image()`, which returned `cairo_surface`,
in order to create a `ClutterContent` with a screenshot of the
`MetaWindowActor`. This added a roundtrip from GPU to CPU memory.
Instead, use the new `paint_to_content()` API which lets us use a
`CoglTexture` directly as source of our `ClutterContent`.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1879>
This signal announces the preferred commit mode of the preedit text when
the input context is reset. Keep this mode around, and ensure to honor this
mode (e.g. maybe commit the preedit string) whenever the input method would
be reset.
This is delegated to the internal layers, so propagate this mode via
clutter_input_method_set_preedit_text().
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1929>
The Eval() method is unarguably the most sensitive D-Bus method
we expose, since it allows running arbitrary code in the compositor.
It is currently tied to the `development-tools` settings that is
enabled by default. As users have become accustomed to the built-in
commands that are enabled by the same setting (restart, lg, ...),
that default cannot easily be changed.
In order to restrict the method without affecting the rather harmless
commands, guard it by the new MetaContext:unsafe-mode property instead
of the setting.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3943
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1970>
Since touchpad smooth scroll events with source finger are handled by
the swipeTracker, the workspace scroll handler can focus on discrete
events.
Thanks to Mutter emulating discrete scroll events, see
meta_seat_impl_notify_scroll_continuous_in_impl in meta-seat-impl.c,
it is safe to ignore smooth scroll in the workspace scroll handler
and handle exclusively discrete events.
In addition, once high-resolution scroll events land in Mutter [1], a
mouse will be able generate non emulated smooth scroll events that
should be ignored in favour of the discrete scroll events.
Otherwise, a single mouse wheel click will scroll through multiple
workspaces at once.
[1] https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1959>
The keyboard is placed outside of the screen when shown and then slides
in via a transition that changes the translate-y property. This
translation does not affect the allocation of the Keyboard actor and as
such does not trigger any of the signals LayoutManager is connected to
to update the input region. This means the input region remains at the
original position of the actor outside of the screen and as a result on
X11 clicks will go through to the underlying window.
There was a workaround for this by queuing a relayout at the end of the
transition, but this stopped working due to optimizations avoiding
unnecessary allocation changes.
This updates that workaround to toggle the visibility of the actor
instead, which is the other signal that LayoutManager reacts to. Once
ClutterActor provides better ways to react to transforms this can
hopefully be removed entirely.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4556
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1955>
Sometimes when an icon is dragged and dropped in Dash, one Favourite app icon
can appear to the right of Dash separator. This can happen when available system
resources are low and the PlaceHolder destroy animation is delayed and the
corresponding container child is still present. With this commit the separator
position is calculated correctly, even when the animation is pending.
Fixes#3966, #3799
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1804>
eee2ccac fixed window opacity being changed outside of transitions but
resulted in the final step of the transition, that would set the window
to fully opaque, being missed.
This change now moves the opacity change entirely out of allocation and
instead follows the adjustment changes directly, which still fixes the
original bug and ensures the opacity is also applied for the last step
in the transition.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4561
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1957>
GJS added a console module that extensions may start using. To ensure that
extensions using console.log() and similar functions don't show up as
'Gjs-Console' in users' system logs, we should call setConsoleLogdomain()
with 'GNOME Shell'.
This GJS API addition is only accessible using ECMAScript Modules (ESM),
this commit moves shell startup to a small init.js module and adapts
CI jobs to either handle or ignore it.
We can drop the .jscheckignore file when future versions of SpiderMonkey
allow for compile checks without validating module specifiers.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1958>
!1940 added support for soup 3, including a fallback to soup 2.4
where the newer version isn't available.
Unfortunately it missed that libgweather has a hidden soup dependency,
and now gnome-shell fails to start if a weather location has been set
up and soup 3 is available.
We don't have a good way to detect that case, so hide the soup 3 support
behind a build option. Distributors are expected to switch it at the
same time as libgweather.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1966>
Unlike for the old Soup.SessionSync/SessionAsync classes,
Soup.Session:ssl-use-system-ca-file already defaults to true.
In Soup3, the behavior was made unconditional and the property
removed.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1940>
Quoting the documentation:
In libsoup 2.44 and later, you can set the session's “proxy-resolver”
property to the resolver returned by g_proxy_resolver_get_default() to
get the same effect. Note that for "plain" SoupSessions (ie, not
SoupSessionAsync or SoupSessionSync), this is done for you automatically.
libsoup 2.44 was released in 2013, so we can safely assume that Soup is
new enough to handle this for us.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1940>
The code that handles extracting extension archives currently uses
an awkward double-callback system. We can do significantly better
by using an async function and exceptions.
Partially based on code from Marco Trevisan.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1940>
Move all remaining bits to the new coding style before making
further changes.
The let → const changes are selectively done to the bits that'll
still be around at the end of the patch series.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1940>
Which better communicates what we are checking for, and is a little more
elegant than repeatedly writing:
```
Main.overview.animationInProgress && Main.overview.visibleTarget == false
```
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1440>
We hide our own "New Window" item if the app itself includes a
"new-window" action. That means that the separator between the
built-in item and desktop actions introduces a small inconsistency
depending on whether a "New Window" item is provided by the desktop
action or ourselves. There's no good reason for that from a user's
perspective, so remove the separator.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
The context menu in the overview includes actions for managing
favorites. Add those to the AppMenu class, but make it another
option as the actions would be slightly weird in the top bar menu.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
The top bar menu always corresponds to a running app, so it made
sense to include the 'Quit' item unconditionally. That won't be
the case for the overview context menus, so handle app state changes
and show/hide the item as necessary.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
Since commit fd0da9606f, we only show the "Open Windows" section
if there are at least two windows. That's another subtle difference
with the overview context menus, but while limiting the section to
a minimum of two windows makes sense where the menu always represents
a running app, it is useful to show the section even for single windows
in the dash/app grid.
Account for both uses cases by adding a corresponding option to the
constructor.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
There's a subtle difference between the top bar menu and the app
icon context menus in the overview regarding the "Open Window"
section.
The former includes skip-taskbar windows, the latter doesn't. It
clearly doesn't make sense for the context menu to include windows
that aren't shown in the overview, but skip-taskbar windows are
likely also less useful in the top bar menu.
Just settle on the behavior of the context menus and move on.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
This is the expected behavior when activating a window or app. Until
now we could rely on the menu being hidden in the overview, but as
that is about to change, make sure the menu behaves as expected.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
App windows are ordered by recency, so a focus change (correctly)
triggers the ::window-changed signal. If we rebuild the section
immediately in response, the activating item will be destroyed
before the menu's ::activate handler, with the result that the
menu remains open.
Defer the section update in that case to allow the menu to process
the ::activate signal first.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
For now the menu is only used in the top bar, where we can assume
that it exists "forever". This won't be the case when we start
reusing it elsewhere, so make sure we clean up after ourselves.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
There is a big overlap between the app menu in the top bar and the
context menu of app icons. It makes sense to unify the two both
from a design- and from a code perspective, so split out the more
modern one into a separate module as basis for a shared class.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1948>
There was a potential issue when suspend was inhibited and immediately
uninhibited again before the creation of the inhibitor has finished.
Then the new inhibitor would be kept active.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1927>
ScreenShield::_syncInhibitor() was closing (and recreating) the
inhibitor everytime it was called, even if no change was needed.
This gets called in various places, including on property changes in
the login1 dbus object. These happen by the time logind already started
suspending at which point new inhibitors can no longer be created. It is
only waiting for existing inhibitors to be closed, so closing the
inhibitor without a new inhibitor will cause the suspending to proceed
immediately if no other inhibitors are present. This can also happen
before the lock screen is shown, which will then complete after resume.
Fix this by keeping track of the expected inhibition state and only
create or close inhibitors if there was a change to that.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3736
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1927>
We have initTranslations() for binding an extension's
gettext domain, but nothing to help with using gettext
from an extension.
Such help would be useful though, as an extension that
calls textdomain() like a normal application would
inadvertently changes the default domain for the whole
gnome-shell process.
Instead, extensions have to use domain-specific versions
of the gettext functions:
```js
const Gettext = imports.gettext.domain('my-extension');
const _ = Gettext.gettext;
```
Make this a bit easier by adding those functions directly
to the extensions object when initTranslations() is called,
then expose helper functions for calling them.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/2594
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1941>
The value can not actually be null at that point, but it's too
hard to spot for tools like coverity:
- before setting view, we chain up to the parent's acceptDrop()
- that calls _canAccept(), which we override and check for the view
Still, and extra ? doesn't hurt, and hopefully will make the tooling
happy.
CID 351269
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1950>
There is now a location portal that provides a similar role
as our agent. Settings supports that portal in its application
panel, which allows users to revisit their choice later.
Unfortunately it uses a different permission store table, so
any permissions granted (or denied) through our agent won't
show up there.
To change that, switch to the same table as the portal/Settings.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1945>
On startup desktop-icons-ng Wayland windows have skip-taskbar==FALSE so
initially pass `_isOverviewWindow` and they get added to the overview,
which is confusing to users. However almost immediately after that they
get `meta_wayland_client_hide_from_window_list` and are removed from
future overviews. So now we respond to `notify::skip-taskbar` immediately
and prevent desktop-icons-ng appearing in the startup overview too.
This is messy and ideally we'd like to know the window type immediately
on creation, but that option only exists in X11 and not Wayland.
https://gitlab.com/rastersoft/desktop-icons-ng/-/issues/137https://bugs.launchpad.net/bugs/1936643
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1925>
At least the last stable release of gjs has an issue with async
handlers for custom (i.e. defined in JS) GObject signals.
The handler still works, but evoking it prints the following warning:
JS ERROR: Error: Could not guess unspecified GValue type
We can avoid the warning by using the addAction() convenience
method, which makes for a small nice cleanup anyway.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4531
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1938>
The PointerWatcher passes the current pointer position to every watch,
so we don't have to query the pointer again scrollToMousePos() is used
as watch callback.
While that should be the only use, be conservative and still query the
pointer when called without arguments.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1932>
Right now the Show Password menu is shown unconditionally for password
entries, ignoring the org.gnome.desktop.lockdown disable-show-password
key.
StShellEntry now honors that key for its peek icon, and there's little
reason for the peek icon and the menu to ever be out of sync.
This commit forces the menu and the icon to always be in sync, and
so makes the menu lock down work automatically, too.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/687>
The behavior when switching workspaces now with the touchpad gesture is
very very weird, it almost always swipes to the last workspace instead
of the next one.
So revert this change again and only swipe a single page per gesture. We
can enable long swipes again when we've figured out a proper way to
detect what the user wants (which is going to be quite challenging), see
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4355.
This reverts commit dfae3281b9.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1933>
ClutterVirtualInputDevice::notify_keyval() expects time to be in
microseconds but Clutter::get_current_event_time() returns the time in
milliseconds. On Wayland the generated event triggers all the warnings
in MetaDisplay::sanity_check_timestamps() due to the event time
seemingly being in the past.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1926>
Dragging a window preview in the overview is supposed to change the
opacity of the dragged actor. This however fails for minimized windows,
because Workspace::allocate() also changes the opacity of those. The
allocation gets triggered by removing the window actor from the
workspace when starting the drag. Avoid this by only changing the
opacity during the overview transitions.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4292
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1847>
At the moment user switching functionality is controlled by two
settings:
org.gnome.desktop.lockdown disable-user-switching
for the panel when the session is unlocked, and
org.gnome.desktop.screensaver user-switch-enabled
for the unlock dialog when the session is locked.
Having the lockdown setting not apply when the screen is
locked is counterintuitive.
This commit makes the unlock dialog honor both settings.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1833>
Since this is a bin and not a button, and it doesn't have any click/keyboard
handling on its own (as in that case it needs a parent button wired like in
the messages list controls), it is confusing that it has can_focus set to
True. To avoid any confusion, this commit removes it without breaking anything
since it had no real use.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1920>
This one was moved to Meta.Context as well.
I don't know why the `debugexit` command quit with an error code
(it dates back all the way to commit 98bd590a5d). Terminating on
request by the user doesn't sound like an error, so don't replicate
that particular behavior.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1917>
Before this, creating a separate instance of `Location.Indicator` failed
because it tries to create export the same DBus path.
This is useful for extensions adding panels on multiple monitors. But
it also seems like a cleaner design to separate the indicator widget
from the logically separate role as a Geoclue agent.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1919>
We build controls layout using the whole monitor vertical space as
available, however extensions or external apps in X11 may reduce the
workarea size horizontally and the shell should always take care of it.
Given that we're already assuming that the allocation is monitor-based
and that we're adjusting it to the workarea, we can just make it more
explicit by using a workarea box that is used as the allocation area.
As per this, we also apply the same logic of applied to the vertical
dimension to the horizontal one.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1892>
To compute the available height for the layout we're currently using the
panel position, while this works for the current and default setup, the
shell may be configured to use a different workarea, so we should rely on
it to compute the available space, instead of a specific widget.
So get the current monitor index for the current view and use its coordinates
instead.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1892>
The background group is currently allocated taking care of the workarea
x, y offset but not of its width/height and this may lead to building a
wrongly sized workspace view when the workarea size is not matching the
monitor size (like when there are struts set).
So, take care of the difference between the workarea and monitor
absolute end coordinates to allocate the background scaled content box.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1892>
We currently disable all network items on both the lock- and login
screen. While it makes sense to be very restrictive on the lock screen,
there are some (fringe) use cases for being more permissive on the
login screen (like remote home directories only accessible via VPN).
There's precedence with the power-off/restart actions to be less
restrictive on the login screen, and since we started respecting
the `network-control` polkit action, it's possible to restore the
old behavior if desired.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1874>
NetworkManager installs a `network-control` polkit action that can
be used to disallow network configuration, except that we happily
ignore it. Add it to the conditions that turn a network section
insensitive.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1874>
Setting up a connection for an Enterprise WPA(2) encrypted wireless
network requires Settings. That's not available when windows are
disabled via the session mode, so filter out affected entries.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1874>
The item launches the corresponding Settings panel when activated, which
doesn't work when windows are disabled by the session mode. Rather than
failing silently, turn the item insensitive.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1874>
Unaccelerated deltas make sure the gesture works the same regardless of how
fast the fingers move; this is what we were already doing for scrolling.
Remove the swipe multiplier as the deltas already match scrolling other than
the 1/10 multiplier Clutter applies to scrolling specifically.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1763>
Since commit 3fb02843, we no longer skip allocation for
results that don't fit the width, and give them a 0x0
allocation instead.
That has the unintended side effect of those children now
being available to keynav. There are cases where we want
0-sized actors to be part of the focus chain (e.g. FocusTrap),
but this isn't one of them, so explicitly exclude 0-sized
children from keynav.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4470
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1916>
Destroying the source from an action callback will result in the
notification being destroyed twice:
- source.destroy() destroys all its notifications
- a notification destroys itself after an action
was activated
This results in unwanted log spam when attempting to dispose the
notification for a second time.
There is actually no good reason for destroying the source explicitly,
as sources already self-destruct with their last notification.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4457
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1908>
Currently we only mark the banner as removed if it is destroyed
while in SHOWN or SHOWING state, but not if we're already HIDING
(for example in response to `NotificationBanner::done-displaying`).
If this happens, we'll try to destroy the notification again at
the end of the transition, which leads to (harmless but annoying)
log spam since Notifications were turned into GObjects (that are
disposed when destroyed).
Address this by always marking destroyed banners as removed, while
still only triggering a state update while shown (or in the process
of being shown).
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4457
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1908>
These do some some signals connection and monitor layout tracking
that we forget to disconnect when the magnifier is disconnected.
This causes warnings if enabling the magnifier, disabling it,
switching to another tty, and back. (Presumably also happens for
monitor hotplugs).
This signal should just stay connected as long as it's needed,
so add a setEnabled() method on the CrossHairs actor that we
can hook on the magnifier.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1902>
Some popovers like the ibus candidate popover may be requested while
showing an unredirected fullscreen window. Disable unredirection
while the popover is visible so that it can actually be shown.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1900>
When implementing a D-Bus method synchronously, regular JS methods are
used. That is, whatever the method returns is passed as return value
to the method invocation.
However for asynchronous implementations, we need to explicitly return
a value to the invocation, otherwise the caller will wait until it times
out eventually.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1891>
The appDisplay used to be accessible via the viewSelector, but
that is gone now. Commit c09c070b15 adjusted the code, but the
replacement uses a Main.overview.appDisplay accessor that never
existed, whoops.
Fix this by exposing a new selectApp() method directly on the
overview, so we don't have to shuffle the appDisplay through
controls → overviewActor → overview.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1891>
With Xwayland started on demand, mutter/gnome-shell will spawn Xwayland
and GNOME settings daemon Xsettings.
Currently, gnome-shell uses a timeout of 5 seconds before canceling the
launch and declare failure to start X11 services.
While 5 seconds may seem like a lot of time, when this is executed on a
virtual machine part of a QA tests under load, this causes random
failures.
There is no good value here, we could increase the timeout but it will
always remain racy. Instead, remove the timeout, systemd itself already
has a timeout mechanism (with a much longer wait time).
Thanks to Benjamin Berg <bberg@redhat.com> for finding the root cause
and suggesting the fix.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1888>
Right now we pop the modal immediately when closing, so the key-release
event of the Escape key ends up in the focus window which may consume it
for leaving fullscreen mode or similar.
We can avoid that in many cases by keeping the modal grab until the end
of the transition, as the key-release event will then likely occur while
the grab is still in place (provided animations are enabled).
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1880>
If parsing the pipeline fails for some reason, we currently end up
with a zombie session that leads to a stuck recording indicator in
gnome-shell.
Instead, properly tear down the session to allow mutter and gnome-shell
to correctly update their state.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1878>
org.freedesktop.impl.portal.desktop.gnome will be used for the
GNOME implementation of desktop portals in the future, so make
sure GNOME Shell's implementation of the access portal won't
conflict.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1872>
Another trivial cleanup. This is in preparation for next commit, that
will expand on this if clause, and will make it a bit more complicated.
Using a variable here makes it easier to read.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1866>
Often, when using multiple monitors, the thumbnails box may be
destroyed before the 'hidden' signal of the Overview is emitted,
which leaves a bunch of lingering signal connections to the
workspace manager.
Call this._destroyThumbnails() in the _onDestroy handler, which
disconnects from workspace manager. This fixes the some of the log
spam that GNOME Shell produces with backtraces of destroyed actors.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1860>
On systems/setups not using systemd, a failure to start gsd-xsettings
with NOT_SUPPORTED does not denote a failure to start Xwayland, just
that we're not using systemd.
In that case, we should just ignore the error, otherwise it will prevent
Xwayland to start on such systems/setups without systemd.
Thanks to Rose Kunkel (@rosekunkel) for spotting the issue/suggesting
the fix.
Fixes: 019229c40e - windowManager: Return failure to start X11 services
Closes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4284
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1850>
Currently, handling of touch devices in the app grid is a bit awkward,
paging by dragging the view can only happen if started from the gaps
between icons, trying to drag from an icon will trigger DnD, and popping
up the menu takes over it all.
Instead, have the app grid actions play this game of rock-paper-scissors:
- Fast swipes on icons trigger scrolling, beats DnD and menu
- Slower press-and-drag on icons trigger DnD, beats scrolling and menu
- Long press triggers menu, beats scrolling, is beaten by DnD
This allows quick swipes to handle navigation, while still allowing the
fine grained operations. DnD, when triggered, dismisses the menu, if
shown.
This all could probably be nicer with a more stateful gesture framework,
we're not there yet though.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3849
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1774>
This setting (by default, 0) sets a time threshold in order to allow
DnD. If the drag is shorter than this threshold, DnD will be cancelled,
and event handling left to the next handlers.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1774>
The switcherPopups use _next() and _previous() to get the items
in the text direction. I. e. with LTR _next() gets the right item;
on RTL it gets the left item. This doesn't work well with RTL when using
the arrow keys since the text direction doesn't matter in those cases.
Pressing Left Arrow should still move left regardless of text direction.
So use the opposite methods.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/2547
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1835>
If for some reason the xserver failed to start, mutter/gnome-shell
should not attempt to connect to the X11 display otherwise it will
lock up in XCB.
An indication of such a premature failure of the xserver is when the X11
services fail to start.
Return the status so that the caller can cancel the connection in time
and avoid the lockup of mutter/gnome-shell in case of failure.
This, however, makes the X11 services a critical component to start
Xwayland, meaning that a failure to start those services for any other
reason than the xserver failing to start would still prevent Xwayland
and therefore X11 clients to run in Wayland. This is however a lesser
issue than mutter/gnome-shell locking up.
This basically reverts commit a96753f0 - "windowManager: X11 can work
without gsd-xsettings".
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1837>
Change behavior on pressing Super+Number in Activities/Overview mode
so that the overview mode is hidden and application can be used.
This makes it consistent with clicking icon in the dash.
Closes: #4212
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1839>
Previously the workarea was only set on construction and then never
updated. As a result the preferred width and height as well as the
allocation were based on an outdated workarea size when it changed after
construction. This for example was happening during the startup
animation, for which the WorkspaceLayout is constructed before the panel
is shown. This caused the workspace in the overview to be slightly
smaller when it is first shown and the overview closing animation to
not expand the workspace to the correct size the first time it is
closed.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3945
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3816
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1771>
In the nested session the startup animation sometimes fails, weirdly
that always happens to me when running a freshly built gnome-shell the
first time. The reason it fails is that mutter fails to aquire a pointer
grab from the xserver, XIGrabDevice() is unsuccessful.
A simple workaround for this race condition in the xserver is to just
grab the devices a bit later, that is after the startups animation
instead of before it.
This was also tested with disabled animations, and seems to work just as
well in that case.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1812>
Some drag actor parents might change their width to 0 after the drag
actor has been removed. If the drag is now canceled and the drag actor
is animated to return to the current parent position and scale, this
scale can not be determined and falls back to 1.0. This might be wrong
and can result in the drag actor being too large/small at the end of the
return animation.
To avoid this calculate the scale of the parent by recursively
calculating the product of its parents individual scales.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4017
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1790>
We want to clip the Workspace actor in the appGrid state of the overview
in order to make sure windows that overflow the monitor don't spill out.
So far we had commit b1970b95b8 for that.
Now since the last commit, window previews always slightly overflow the
allocation with their icons. That means a part of the window icon gets
clipped away as soon as the transition to the appGrid starts, which
looks weird.
Fix that bug in the transition by slightly extending the clip downwards
when animating between the window picker and the app grid state. The
extra height we extend the clip by is controlled by the overviewState,
which means we extend the clip by the full icon overlap in the window
picker state, but don't extend the clip at all when in the app grid
state.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1813>
We want to use as much space as possible for showing window previews in
the overview, and the title and close buttons of those windows are only
visible on hover, so we can show them above anything if we want.
On both primary monitors and secondary monitors, there's a certain free
space available towards the bottom edge of the monitor (on the primary
monitor we show the dash there, and secondary monitors just scale down
the Workspaces). We can make use of this by checking how much free space
there is available from the bottom edge of our allocation to the bottom
edge of the monitor, and then aligning the window previews to make full
use of this space.
So stop adding any padding to the edges of the Workspace, which will
make the windows a lot larger and completely fill the Workspaces
allocation.
The left, top and right monitor edges should always be far enough away
to accomodate the close button and hover scale-up of the window. Only
with the bottom edge of the monitor we have to be a bit more careful
(the overflowing height of the window title is quite big), so there we
check if enough free space is available. If there isn't enough free
space, we simply apply a bit of bottom padding again and shift the
window up.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1813>
Make the SecondaryMonitorDisplay a bit more similar to the
ControlsManager container on the primary monitor, and clip that widget
instead of the WorkspacesViews on secondary monitors.
This will allow us to overpaint the WorkspacesView allocation and paint
the WindowPreview overlays like the title and close button outside the
allocation with the next commit.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1813>
The current gstreamer pipeline performs quite bad on slower machines and
is dropping lots of frames, improve the pipeline by changing a few
things:
- Use threads for videoconvert and improve speed of videoconvert by
disabling some unneeded things
- Add a queue before the encoding step, this allows the encoder to work
at its own pace and will lead to a lot more stability
- Remove the fixed quantizer and only set a max quantizer, this helps
quite a bit with performance
- Change the deadline parameter of vp8enc to 1: This makes the encoder
go into real time mode, which will make it a lot faster
- Set cpu-used to 16, the maximum possible value.
- Set static-threshold to 1000, static-threshold is the motion detection
threshold, and while a value of 100 is recommended for screencasting in
the gstreamer documentation (see [1]), using 1000 appears to perform a
lot better and still outputs fairly good quality
- Set a larger buffer size than the default size, this seems to get a
bit more stability during high load scenarios
All in all, those changes make the pipeline drop no more frames when
recording at 30 FPS and 2K screen resolution. That was tested on a
fairly recent mobile core-i5 processor.
Also, because we now have two %T replacement strings for the number of
threads, we need to switch to replaceAll(). For that to work, we have to
put the %T matching expression into quotes.
[1] https://gstreamer.freedesktop.org/documentation/vpx/GstVPXEnc.html?gi-language=c#GstVPXEnc:static-threshold
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1633>
The original parent of a dragged actor might have been destroyed after
the drag has been started. When the drag is canceled and _dragOrigParent
is set, the code is trying to get the current position and size of it
when restoring. With a destroyed parent this however would result in a
crash.
This could happen for example when starting a drag on a window preview
while the overview is hiding and then releasing it once the transition
is done.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4024
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1817>
`ThumbnailsBox` listens for the `showing` signal from the `Overview` to
create its thumbnails and destroys them on the `hide` signal. Since the
`showing` signal can be emitted multiple times when switching between
the shown and hidden state without ever fully completing the transition,
this will cause `_createThumbnails` to be called multiple times, each
time adding another set of workspaces.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3819
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1828>
When starting a gesture to open the overview while a transition to hide
the overview is running, Overview._shown will be first set to false when
starting the transition and then to true by the gesture before the
`onComplete` callback is called. The `onComplete` callback in this case
is `Overview._hideDone()` which starts a transition to show the overview
again which also emits the `showing` signal. Since the gesture emits a
`showing` signal as well, this results in two consecutive `showing`
signals without a `hiding` signal in between.
This breaks the `searchControler` which adds a key press handler to
start the search on `showing` and removes it on `hiding`. So every time
this happens a new handler that will never be removed is added,
resulting in the first key press being repeated.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4004
Related: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3819
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1828>
For the screensaver service, it is quite normal that a consumer only
subscribes to the "ActiveChanged" signal without calling any methods.
The result is that we don't know about the consumer, and shut down
the service anyway after we hit the timeout.
If this happens, we break functionality like gnome-settings-daemon's
screen blanking on idle.
Fix this by simply disabling auto-shutdown for the service, which
also reflects the expectation that the screen saver service is
always running in a GNOME session.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4114
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1824>
It can happen that we get a problem report and a verification failure at
the same time. For fingerprint, a problem report can result in an
internal verification failure to be queued.
Remove this queued failure again if we got a failure already from GDM
directly.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1821>
At the moment a failure in a background service can lead to the
various verification signals getting disconnected, even though
we still need them for a foreground service.
This commit changes the code to only disconnect when we've run
out of tries.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1821>
At the moment we set the state of the auth prompt to failed any
time the user fails an attempt. But verification is still going
on until the user exhausts all attempts, so that's wrong.
This commit changes it to only set the state to failed when the
user is out of tries.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1821>
At the moment we treat a failure in any service as a signal to stop
tracking users responses to service questions.
This commit makes sure we don't stop waiting for answers if a background
service fails.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1821>
When we press a key with variants, we used to prevent an
early ::pressed, because a long press could show the options
popover, and the press be undone.
In addition, this long press could move to one of the suboptions,
and be released there. For this case we also want this late
emission of the ::pressed signal.
This makes the "tap, drag, release" pattern work on the
regular OSK keys, in addition to the emoji panel.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1789>
This is already allowed for pointer events, but touch events still expected
that the touch begin and end happen on the same Key actor. Change this
behavior for touch events, this is necessary for the tap-drag-release
pattern to select key variants to work on all input devices.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1789>
At the moment the timed login feature is implemented in the user list.
If there's no user list, we don't show the indicator anywhere and
don't proceed with timed login.
This commit allows timed login to work when the user list is disabled.
It accomplishes this by putting the timed login indicator on the
auth prompt in that scenario.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1809>
When super is pressed again during the overview transition, we shift
up to the app grid. That means that the feature currently doesn't
work when animations are disabled (like in a VM), because there is
no transition in that case.
Address this by adding a time-based fallback in that case, i.e.
shift up when a second super-press occurs within 250ms after the
first one.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4121
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1811>
Using the callback directly as signal handler means that it will
receive the signal parameters (in this case: the StButton instance
that emits the signal).
This leaks an implementation detail that is harmless in the best
case, but can break dialogs when using bind() on the callback.
Avoid that trap by explicitly calling the callback without arguments.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4139
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1819>
For the primary monitor workspace thumbnail, we must keep the monitor
index in sync with what is currently the primary monitor index,
otherwise we might end up trying to move windows to non-existing
monitors.
For example, if the primary monitor index was 1 when the thumbnail box
was created, but later, the primary monitor index changed to 0, with the
other monitor being turned off, moving a window to one of the workspaces
on the workspace thumbnail, gnome-shell would attempt to move it to the
monitor with the index the primary monitor had in the past, with the
problem being that that monitor no longer exists.
Fix this by listening on the 'monitors-changed' signal on the layout
manager, and update the monitor index of the primary workspace
thumbnails box. Make sure to connect to the signal before creating the
thumbnails box, as the thumbnails box itself will listen to the signal
and recreate its actual thumbnails, and it must do this with the up to
date monitor index.
Closes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4075
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1810>
When OverviewControls goes from HIDDEN to APP_GRID, it constantly checks
if AppDisplay needs to be visible or not by checking the current overview
state is bigger than WINDOW_PICKER. Turns out in this case this check is
problematic, because when the current state trespasses WINDOW_PICKER, the
layout manager will have already positioned AppDisplay halfway to its final
position.
Use either the final or the current state, whichever is biggest, when updating
the AppDisplay visibility. It optionally allows passing the overview state
params to _updateAppDisplayVisibility() so that we avoid a few trampolines to
recaltulate the adjustment state.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1805>
Both app grid and window picker are now always visible in the overview,
so their handling of the PgUp/PgDown keys conflicts.
Resolve that by checking for the overview state instead.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1798>
Window previews looks slightly offset to the left and top right now
because we don't use the same padding on all edges of the Workspace. We
do that because the oversize and overlap of the window previews is
different on all sides (for example the bottom overlap is very large
because there we show the window title).
To make sure window previews are always perfectly centered on the
Workspace, only use the largest one of the oversize values as spacing
and padding, and add the larger one of the overlap values for the
vertical padding in addition.
With this, we now center the window previews on the Workspace while
never overpainting the allocation of that Workspace to show overlays.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3634
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1670>
This was forgotten when porting to GTK 4, leading to the following error
when user tries to copy the error message produced by an extension:
JS ERROR: TypeError: Gtk.Clipboard is undefined
_initActions/<@resource:///org/gnome/Shell/Extensions/js/extensionsService.js:255:31
run@resource:///org/gnome/Shell/Extensions/js/dbusService.js:177:20
main@resource:///org/gnome/Shell/Extensions/js/main.js:19:13
run@resource:///org/gnome/gjs/modules/script/package.js:206:19
start@resource:///org/gnome/gjs/modules/script/package.js:190:8
@/nix/store/fwnkwvhwm3kqck4fhkc5y5z853radggg-gnome-shell-40.0/share/gnome-shell/.org.gnome.Shell.Extensions-wrapped:7:17
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1795>
We only want to show the welcome dialog in the user session, not
on the login screen or during initial setup. We currently achieve
that by explicitly checking for those mode names, but there are
other modes like gnome-classic where the dialog is equally un-
helpful. Support those cases by adding a session mode property
that determines whether the welcome dialog should be enabled,
so that modes can opt in or out of the feature themselves.
(Both the 'gdm' and 'initial-setup' modes are based on the
'restricted' mode, so this change does not affect them)
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4026
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1793>
Usage of GET requests for checking updates was made deprecated
at website some time ago [1], but REST endpoint was
CSRF-protected until recently [2].
The body of update request may be big enough and thus does not
suitable for GET requests.
[1] 0b38da1b2b
[2] e3ab0c07dc
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1781>
`key` is an empty string in this case, causing `charCodeAt(0)` to return
`NaN`, which when passed to `Clutter.unicode_to_keysym` now generates an
error in gjs >= 1.67.3:
```
JS ERROR: Error: Argument wc: value is out of range for uint32
```
And the symbolic keys like Backspace, Enter and Caps Lock would have their
presses ignored.
Just skip the call to `charCodeAt` that will fail and allow
`Clutter.unicode_to_keysym` to return its usual error flag.
Fixes: https://bugs.launchpad.net/bugs/1918738
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1758>
The parameter to `ModalDialog.close(timestamp)` is optional. But when
invoked via the network dialog's Cancel button it was receiving an
implicit parameter value that's definitely not a timestamp:
```
[0x560f18af0c50 StButton.modal-dialog-linked-button:first-child hover ("Cancel")]
```
And as of today (or gjs >= 1.67.3) that's reported as an error:
```
JS ERROR: Error: Argument timestamp: value is out of range for uint32
popModal@resource:///org/gnome/shell/ui/main.js:638:12
popModal@resource:///org/gnome/shell/ui/modalDialog.js:206:14
close@resource:///org/gnome/shell/ui/modalDialog.js:179:14
```
and so you can't Cancel the dialog anymore.
Make sure `ModalDialog.close()` receives an `undefined` timestamp it
knows how to handle.
Fixes: https://bugs.launchpad.net/bugs/1918666
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1757>
Right now we use a ClutterClone ghost of the panel simply to add some
spacing to the top monitor edge.
We can remove the panelGhost by adjusting the allocations of all the
parts of the overview ourselves inside ControlsManagers vfunc_allocate,
this should also work with extensions that move the panel somewhere
else.
This makes the initial relayout of the overview significantly faster,
because we now no longer have to relayout the whole panel in the
process.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1755>
The OverviewControls actor gets allocated a fixed size by its parent,
the OverviewActor, anyway, so it's pretty useless to go through the size
request machinery and add up all the sizes of items in the iconGrid,
coming up with a preferred size that's wrong anyway.
Instead simply return a min and preferred size of 0 in
get_preferred_height/width of ControlsManagerLayout.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1755>
Hiding actors allows excluding them from layout, so by hidding the
appDisplay in all the cases where the overviewAdjustment is not actually
showing it, we can save a lot of time on the first frame of painting
the overview because we no longer have to layout the whole appGrid.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1755>
We create a lot of BaseIcons for the appGrid, one for every app, and for
all of those we have to hop through JS to get the preferred width. That
makes it another obvious target for moving to C, so let's do that.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1755>
The logic that decides whether we should shift the window up when the
cursor rectangle overlaps with the keyboard rectangle doesn't work
properly right now, we want it to work like this:
- If the currently focused window is shifted up, keep it shifted up
until the cursor rect no longer overlaps the keyboard rect. To do that
comparison correctly, we need to adjust for the height the cursor rect
is shifted up by (keyboardHeight) and temporarily shift it down again.
- If the currently focused is not shifted up, we want to shift it up as
soon as the focus rect overlaps the keyboard rect. If that's not the
case, want still want to call _setFocusWindow(null) in order to shift
the previously focused window back down.
This fixes two issues: 1) We're currently shifting windows back down at
the wrong position of the cursor (that is y < keyboardHeight). 2) We're
not shifting down previously focused windows when focusing a different
window with the new focus in a specific region (y >= keyboardHeight &&
y + h < monitor.y + monitor.height - keyboardHeight).
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1760>
So far the FocusTracker of the OSK can only recognize grab ops on a
window, that is when the user grabs the window using a mouse or the
touchscreen and actively drags it somewhere.
Window can also be moved using keyboard shortcuts, fullscreen buttons or
other ways which don't rely on grabs. Start also supporting those window
movements by listening to the "position-changed" signal on the currently
focused window and emitting the new "window-moved" signal in that case.
Because the OSK sometimes moves windows by itself, we temporarily
disconnect from that new signal while we move the focused window in
_windowSlideAnimationComplete().
This also takes care of resetting this._focusWindowStartY on movements
of the window.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1760>
Commit 8526776b4a changed the OSK to use
the translation-y property of the MetaWindowActor when animating a focus
window, which broke two things:
1) It's not compatible with the obscured region culling we do for
windows in mutter. That's because MetaCullable strictly operates in
integer coordinates and thus has to ignore any transformations
(translation-y is a transformation). Because of this, during the
animation and gesture, window damage is now tracked incorrectly,
introducing painting issues. The best fix for this would probably be
factoring in transformations when tracking damage in MetaCullable, but
that's not feasible right now.
2) It broke the shifting up of maximized and tiled windows, likely that
is because they are positioned using constraints internally, and mutter
enforces those constraints every time meta_window_move_frame() is
called, not allowing the window to move somewhere else.
To fix both issues, go back to the old way of shifting the window for
now, using the fixed y position of the ClutterActor. To make sure the
drag-up gesture still works, store the initial y1 position of the window
and then use that as a reference point for all our animations.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1760>
I suggested it myself when reviewing
https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1668, so
here I am reverting that again...
The difference between calling _setFocusWindow(null) and simply
unsetting the focusWindow is that the former animates the window back to
its position before we shifted it up, while the latter simply "lets go
of the window".
In this case we actually want the latter because after the user grabbed
the window, we obviously should not animate it away right underneath the
users pointer/finger.
To ensure the same mistake doesn't happen again, add a small comment
explaining why this code is as it is.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1760>
If the actor is unmapped in the handler, the touch gesture will cancel.
Since we haven't reset the state yet, it will still work and will actually
cancel the gesture, so reset before that instead.
Fixes overview cancelling when trying to open it with a swipe.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1731>
When this class was written, all swipes in the shell were vertical, so it
made sense to make the default orientation vertical. This isn't the case
anymore, thus pass make it mandatory to specify orientation when creating
a tracker.
Change the property default values to horizontal as well to match Clutter
instead of the old shell design.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1731>
The focus rectangle of the OSK currently gets stored with an offset that
removes the global coordinates and makes it window-local coordinates.
This offsetting has been applied since the introduction of the
FocusTracker with commit fc5ab44704 and it
seems there's no real reason for it.
By removing this, we also emit position-changed when the window has
moved but the window-local coordinates stayed the same. We really want
to emit position-changed in that case because it might affect whether
the window needs to be shifted up.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1728>
Apparently some clients, including gtk don't "clip" the focus rectangle
to their window bounds when scrolling the focus outside the window. This
makes us shift up windows when the focus actually is no longer visible.
This issue needs fixing in GTK, it should probably stop reporting
focus changes when the focus moves outside of the visible view. We can
still do a little bit better on our side though and "clip" the rectangle
to the windows frame rect: If it moves out of the window, we simply stop
updating our focus rect.
The intersection check introduces a small problem though: Some clients
(for example gedit) will give us a cursor rect that has a 0 width or
height. This won't play well with graphene_rect_intersect()
(GrapheneRects never intersect if they are 0-sized), so we set the size
to 1 in case we get a 0-sized rectangle.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1728>
The FocusTracker keeps track of the currently focused window using its
internal this._currentWindow property. It will only pick up the focused
window though when receiving a "notify::focus-window" signal, so the
focused window that's set when the FocusTracker is created won't be
picked up.
Fix that by setting the _currentWindow during creation of the
FocusTracker.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1728>
Now that we got rid of the external calls to setCursorLocation(), we can
make that private. animateShow() and animateHide() weren't called from
outside anyway, so let's make those private, too.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1728>
What this signal does is fire when a window was grabbed. A receiver
might want to do something special when a window was grabbed, whereas
"reset" can mean anything. Rename the "reset" signal to
"window-grabbed".
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1728>
Instead of interpolating our workspace and layout boxes for each child
using clutter_actor_box_interpolate(), use our Util.lerp() function and
stay in JS land instead.
This is quite a large performance improvement since it avoids
heap-allocating a new ClutterActorBox for every child. With this, we're
finally at a duration of 1.0 ms to allocate the Workspace with 20
windows.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
Move the get_preferred_width/height() and allocate() vfunc
implementations of WindowPreview to C, subclassing the C GObject from
JS.
This gets us another significant performance gain, allocating a
workspace with 20 windows now only takes 1.2 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
It turned out that getting properties and saving them to a variable
outside of loops instead of accessing them everytime inside the loop can
have significant impact on performance, so do that in Workspaces
vfunc_allocate().
Here the impact is not that large, about 0.05 ms with 20 open windows,
that still seems worth it though.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
These checks aren't needed since Clutter should enforce this for us
already and skip the implicit transition when possible. This gets our
time spent in vfunc_allocate() down to 2.0 ms with 20 windows
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
Setting four properties is more expensive than calling two C functions,
so move to set_origin()/set_size() calls for our ClutterActorBox
handling.
This gets us down to an average time of 2.1 ms spent in vfunc_allocate()
with 20 windows
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
We can save a little bit of time in this loop by iterating directly
over the windowSlots array instead of iterating through children and
then performing a search for the windowSlot. This saves more time and we
only spend 2.2 ms instead of 2.3 ms in vfunc_allocate() now.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
This layout manager is used quite often and the time we spend calling
it's allocate and get_preferred_width/heigth functions increases with
every open window.
We can save a lot of precious time during the layout cycle by moving
this layout manager from JS to C, thus avoiding the overhead of
trampolining between C and JS land.
In a measurement where the average time spent in vfunc_allocate() of the
Workspace actor was measured while opening an overview with 20 windows,
the average time spent went down from 3.1 ms to 2.3 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1743>
Right now we always recreate the icon of the appMenu when calling
_sync(). This will relayout the panel everytime we open the overview,
because we call _sync() in that case.
We can easily avoid that by only recreating the icon actor in case the
app that's opened actually changes. This gets us close to doing no more
relayouts of the panel when opening the overview.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1733>
We already set the AppMenuButton to be non-reactive and transparent when
we hide it. Hiding it completely using clutter_actor_hide() will
additionally make it no longer affect layout and thus queue a relayout.
Since we hide the appMenu in the overview and we want to avoid
relayouting the panel when entering and leaving the overview, don't
completely hide the AppMenuButton to avoid queueing this relayout.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1733>
This updates the use of bind_cairo_surface_property for the changes
from d7cb2eeebc. Now this function returns a StImageContent and not
an Actor like the following code expects, so wrap it in a StIcon.
Also the function lost its size argument.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1718>
If the window wasn't in _windowSlots, the index is -1, so the last
element of the array is removed, leading to
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3547
I logged the stack trace from removeWindow() and it seems that when you
move the window from the second monitor to another workspace like shown
in https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3547,
removeWindow() is called twice for it, presumably once for the second
monitor workspace, and then once for the first monitor workspace for
some reason. It is during that call that _windowSlots doesn't contain
the window, so instead the last element (index -1) is removed, leading
to the animation bug.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1727>
We block state updates while the indicator for the active workspace
is animating. To track that, we check whether the scroll-adjustment's
value matches the active workspace index. That works as long as the
adjustment's value changes after the active workspace, but not when
switching workspaces via SwipeTracker which only changes the active
workspace after transitioning to the new scroll value.
To fix that, update the indicator on workspace changes as well.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1716>
Animating items of the iconGrid involves calling a few more C functions,
which is quite slow. Especially calling ClutterActor.set_easing_delay()
is slow because we override that function in JS to adjust for the
animation slow-down factor. So add a small class variable to make sure
we only animate the icons of the grid when we actually need it.
This makes the average time spent in vfunc_allocate() of the iconGrid go
down to about 0.7 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1713>
It's quite slow to access class variables in JS, especially when they're
backed by GObject properties. To avoid accessing them in every iteration
when we're looping through the children of iconGrid, store those values
to another variable and reuse that inside the loop.
This shaves off another 0.2 ms from iconGrids vfunc_allocate(), getting
the average time spent in that function down from 1.3 ms to 1.1 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1713>
Using a preexisting array to iterate over is much faster than iterating
over the actors children using a "for ... of" loop, that's because the
latter calls into C functions to get the next actor all the time.
Also, stop using array deconstruction here, it turned out that this is
extremely expensive. When profiling this part of the code, it turned out
that only 9% of the time spent in _getChildrenMaxSize() is spent calling
the get_preferred_height/width() methods. When not using array
deconstruction, this time increased to 22%, still not great, but a lot
better.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1713>
We need to access the visible children of a page in inside
vfunc_allocate(), and since getting those children is quite slow (it
involves iterating over all the children of the actor) let's avoid that
and cache the array instead.
This reduces average time spent in vfunc_allocate() of the iconGrid from
1.6 ms down to 1.4 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1713>
We call this._getChildrenMaxSize() from the allocate() vfunc of
IconGridLayout. Since the function is quite expensive, it slows the
layout process down a lot, so instead of re-calculating it on every
relayout, cache the max size of children.
This makes the average time spent in vfunc_allocate() of the iconGrid go
down from 2.3 ms to 1.6 ms.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1713>
