Since the ClutterGrab rework, grabbing properly emits crossing events.
StButtons take a ClutterGrab as soon as they are pressed, so pressing
the close StButton of the WindowPreview takes a grab and causes a
LEAVE+key-focus-out event getting sent to the WindowPreview. This in
turn makes us hide our overlay (which the StButton is part of). We
automatically ungrab ClutterGrabs when hiding actors, so the StButtons
grab now gets released again, key-focus and hover state are updated
again and we emit an ENTER event to the WindowPreview. The preview now
tries show its overlay again and things explode because we re-enter the
mapping machinery.
For the LEAVE event we can break this cycle by detecting the GRAB_NOTIFY
crossing event of our own StButton and not reacting to that.
We should do the same for the key-focus-out event, but these don't pass
context information like a GRAB_NOTIFY flag yet, so just check the
current grab actor here.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3165
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2231>
Commit ba23279f1f was aimed at fixing a
bug where the layout is frozen so early that we don't have an existing
one to re-use, because no allocation cycle happened yet. It tried to fix
that by forcing the creation of a layout when needsLayout === true,
this turned out to be a bit too much, as it also forced creating a new
layout when the layout was frozen after closing a window (which would
set needsLayout to true).
To fix this regression and still avoid reintroducing the bug the
previous commit fixed, tighten the check a bit and only force creating
the new layout when there's no allocation yet. This makes freezing the
layout after closing a window work again.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2236>
The magnifier uses a PointerWatcher (which is based on a simple timeout
source) to update the zoom region based on the current mouse cursor
position 60 times a second. When updating the zoom region, it would also
hide mutters cursor using meta_cursor_tracker_set_pointer_visible().
Since a few months, mutter has decoupled the handling of input events
from the monitor refresh rate though, which means it's no longer
guaranteed that the cursor changes only 60 times a second (on higher
refresh rate monitors it actually never was). This means mutter might
show the cursor more often than 60 times a second, while we hide it only
60 times a second, leading to a flickering second cursor.
To fix this, implement the cursor-hiding by listening to
MetaCursorTrackers visibility-changed signal, and immediately hiding the
cursor again when it's shown.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2234>
We want to use the current event time for activating the workspace, and
this time is only available when calling global.get_current_time() or
Clutter.get_current_event_time() from the context of an event handler.
So instead of trying to get that time when the animation has finished
from the onComplete() handler, get it before and store it as a variable
to use in the onComplete() handler later.
This fixes an error message when switching workspaces using the swipe
gesture where MetaWorkspace complains about the 0-timestamp we pass it.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2233>
In testing on Fedora, gnome-shell crashes here:
JS ERROR: TypeError: DEFAULT_FOLDERS[folder] is undefined
This needs to be a "for of" loop, not a "for in" loop, because
`folders` is an array of the hash's keys, not the hash itself.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2238>
In a very un-StButton fashion, OSK keys are not interested in
altering the current focus state, as otherwise pressing those
will trigger a grab that will alter key focus, so the generated
key or IM state events will not be handled by the actor that
originally had the focus as long as the OSK key is pressed.
Despite being StButtons, OSK keys do already perform their own
press/release handling with internal state tracking, so it is
not a big stretch to simply consume the events, and update the
:active pseudo-class manually.
This makes OSK keys still look and behave as usual, but without
any grabbing shenanigans that might affect the focus state. This
makes all OSK keys work again.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4986
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2237>
In the case of bringing up the OSK while there is a grab (like, every
GNOME Shell entry), we used to special case event capturing so events
directed to the OSK would be let through.
When Clutter.Grab came around, events would be propagated only within
the actor hierarchy that holds the grab, which rendered this special
case just as useless as the OSK while a grab was hold. Since it wouldn't
be part of the grab hierarchy, clicking on the OSK would do nothing.
In order to let the OSK handle events, double down on the special case
and let it forward the event directly to the actor under the device,
instead of trying to let it through somehow. Since the actor under the
device are usually OSK buttons in this case, we don't need further
propagation to make it work, which makes the OSK functional again while
the shell holds a grab.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2237>
The screencast portal supports recording a single window,
and presents a list of open windows when that option is
selected. To allow updating that list when windows are
opened or closed, add a new "WindowsChanged" signal that
the portal can listen to.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2229>
The transition from the overview freezes the workspace layout at the
start of the animation, which means that the calculated window slots
remain the same while the workspace itslef grows. This causes the
windows to appear slightly shrunk in comparison to the workspace and
shifted to the top left. This is especially noticeable during the
beginning of the animation when there is more weight on the slots than
the original window position and if there are not that many open
windows.
Unfreezing the layout for this transition is not possible, because it
would cause issues with newly opened windows abruptly changing the
layout when the animation happens after starting a new app.
This change instead tries to scale the frozen layout to the current
workspace size. While this is not entirely correct, because this scales
the spacing between the slots as well, it is still more accurate than
the completely unscaled slots.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4616
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1980>
We currently assume that any '::destroy' signal on a GObject type
has the semantics of the ClutterActor/GtkWidget signal, and should
therefore result in all signals being disconnected.
But we already have a case where the assumption doesn't hold: ShellWM
uses '::destroy' for the closing animation of windows, and the ShellWM
object itself remains very valid after the emission.
So rather than making assumptions about '::destroy', check objects
against a list of destroyable types that are explicitly registered
as such.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2226>
On some touchpads/laptops, the swipe gesture to open the overview can be
performed so fast, that it starts and ends in between two frames. Now
when this happens, and the gesture ended with too little movement to
confidently say the user intended to open the overview, we'll close the
overview again.
While closing the overview, we freeze the layout of the Workspace in
order to avoid changes to windows messing with the animation. This means
that in the case described above, we freeze the layout even before the
first frame of the opening animation happens. No frames being drawn also
means no allocations happening, and since we create this._layout in
vfunc_allocate(), this means that on the first allocation cycle of the
overview we'll see this._layoutFrozen = true, but will also not have
a this._layout nor this._windowSlots.
This creates an annoying visual glitch where for a split second all
the windows disappear (overview is visible but no WindowPreviews get
allocated).
To fix this, force creating a layout on the first allocation cycle, even
if the layout is currently frozen.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2203>
In the panel we listen for button presses on the panel in order to start
a drag op of the adjacent maximized window. With the recent changes
removing the source detail of events from Clutter, we now can no longer
access buttonEvent.source, but instead need to use
ClutterStage.get_event_actor().
Since that function expects us to pass a ClutterEvent and not a
ClutterButtonEvent, let's listen for the "button-press-event" signal
instead of connecting to vfunc_button_press_event(). This will give us a
ClutterEvent we can work with.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2225>
While we don't want to position overview elements according to
the work area, it does apply in the session when the overview
is hidden. That is, we should take it into account for the HIDDEN
state, so that transitions from and to the overview are correct.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2223>
The work area is really the space that is available to application
windows. Applying it to the overview is therefore questionable,
in particular given that
- X11 clients that affect struts aren't shown
- elements added by extensions have other means
to affect the overview layout
Not applying the work area to the overview also makes ignoring
actors' visibility for struts acceptable again: Assuming that
strut actors are only hidden when the monitor is in fullscreen,
freezing the work area instead of updating it for windows that
are fully obscured by a fullscreen windows makes perfect sense.
This reverts commits 81a1e294f8 and 2b074882f4.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2223>
The changes violated too many assumptions on the mutter
side. And even if those were addressed, changing the
work area when a window enters/leaves fullscreen isn't
great, because it will force an expensive relayout on
all other windows.
This reverts commit cd1102ff30.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2223>
There are cases where we want to connect to a number of signals
for the lifetime of an object, but also other signals for a
limited period (say: between show and hide).
It is currently not possible to use disconnectObject() for the
latter, because it will disconnect all signals.
To address this use case, add a small class that can be used as
a transient signal holder, while still benefiting from autocleanup
by proxying the real owner.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2221>
The 'destroy' signal currently doesn't work with connectObject(),
because the handler is only connected after the signal tracker's
own destroy handler, which disconnects all handlers.
Address this by using connect_after for the cleanup handler, so
that other destroy handlers run before it (unless they also use
ConnectFlags.AFTER, but well *shrug*).
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2221>
In order to make very short screen capture sessions more visible, let
the indicator remain visible, but a bit greyed out, for some seconds.
This makes it more obvious something was just capturing the screen.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2132>
Start using the new methods to simplify signal cleanup. For now,
focus on replacing existing cleanups; in most cases this means
signals connected in the constructor and disconnected on destroy,
but also other cases with a similarly defined lifetime (say: from
show to hide).
This doesn't change signal connections that only exist for a short
time (say: once), handlers that are connected on-demand (say: the
first time a particular method is called), or connections that
aren't tracked (read: disconnected) at all.
We will eventually replace the latter with connectObject() as
well - especially from actor subclasses - but the changeset is
already big enough as-is :-)
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1953>
The module exports a `addObjectSignalMethods()` method that extends
the provided prototype with `connectObject()` and `disconnectObject()`
methods.
In its simplest form, `connectObject()` looks like the regular
`connect()` method, except for an additional parameter:
```js
this._button.connectObject('clicked',
() => this._onButtonClicked(), this);
```
The additional object can be used to disconnect all handlers on the
instance that were connected with that object, similar to
`g_signal_handlers_disconnect_by_data()` (which cannot be used
from introspection).
For objects that are subclasses of Clutter.Actor, that will happen
automatically when the actor is destroyed, similar to
`g_signal_connect_object()`.
Finally, `connectObject()` allows to conveniently connect multiple
signals at once, similar to `g_object_connect()`:
```js
this._toggleButton.connect(
'clicked', () => this._onClicked(),
'notify::checked', () => this._onChecked(), this);
```
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/1953>
The default folders used to be created by gnome-software, as that was
where folder management used to be implemented. Since then, folder
management via drag and drop was implemented in the shell, and the
gnome-software code was removed.
The only bit that still involves gnome-software are the default folders
that are created on first run. Given that everything else has moved to
the shell, it makes sense to take over that part as well.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4948
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2110>
It does not make sense that the target actor is both destinatary
and content of the events being sent, so this API call is going away.
Since the event can be sent entirely unmodified (more so, it will
become immutable/readonly in the future), avoid creating a copy
since it does not matter sending one or other struct.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2216>
With the porting of gnome-bluetooth to the new GListModel API the
behavior regarding removing adapters changed: It now no longer
guarantees to emit "device-removed" signals for the paired devices when
the adapter gets removed.
This means we need to do that ourselves now, so clear the list of
connected signals when the default adapter changes.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2214>
Turns out this broke showing the bluetooth menu entry for adapters which
rely on the had-devices-setup property being set while turned off. These
adapters are completely removed from the system by the firmware after
powering them off, so in that case there is no default adapter anymore,
although we still want to show the menu.
This reverts commit aaf47167b5.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2214>
We actually don't get a time from the xdndHandler when it emits
drag-end, so we fail right now when calling
workspaceManager.get_workspace_by_index(time).
Fix it by getting the time ourselves instead.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2211>
Only workspace groups that are associated with a workspace have a
background. As a result, when restacking window previews we end up
passing `undefined` to clutter_actor_set_child_above_sibling()
instead of null, triggering a warning.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2209>
We have made good progress on object literals as well, although there
are still a lot that use the old style, given how ubiquitous object
literals are.
But the needed reindentation isn't overly intrusive, as changes are
limited to the object literals themselves (i.e. they don't affect
surrounding code).
And given that object literals account for quite a bit of the remaining
differences between regular and legacy rules, doing the transition now
is still worthwhile.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2200>
It's unfortunate that we cannot fully embrace template strings, but
we aren't limited to "legacy" formatting either; replace the last
remaining places where we still use string concatenation to reduce
the difference between regular and legacy style a bit.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2200>
This commit adds support for Home and End keys to move
to the first and last workspace respectively.
Previously only Page_Up and Page_Down were recognized
to move one workspace at a time in overview mode.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2201>
The event passed to formatEventTime() is reused at a later point.
Therefore, we are not allowed to manipulate the event directly.
This fixes an issue where the user clicks on a multi-day all-day event
the second time before the event gets garbage collected and the event
then is one day shorter.
Fixes 528ee01fef
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2184>
WorkspacesDisplay connects to key-press-event on the stage to switch
workspaces when page up or down is pressed and nothing else intercepts
these keys. This means that it is still possible to switch workspaces
while they are hidden behind the search.
So only allow these keybinding while the WorkspacesDisplay is reactive
which gets updated by ControlsManager depending on whether there is an
active search or not. Also set it as initially reactive, because
otherwise the keybindings would only work after performing an initial
search.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2204>
