Ray Strode [Fri, 15 Dec 2017 15:05:34 +0000 (10:05 -0500)]
background: don't leak wall clock when background changes
The background code allocates a GnomeWallClock when its first created,
but neglects to drop a reference to that clock at destroy time.
The undestroyed clocks lead to a timerfd leak that eventually prevents
the shell from functioning.
Florian Müllner [Sun, 19 Nov 2017 19:35:25 +0000 (20:35 +0100)]
system: Align submenu labels with parent
All section titles use an icon, while all corresponding submenu items
are plain-text. As a result, labels in submenus aren't aligned with
the labels in the parent, which makes them harder to read and look
unbalanced. Address this by adding additional whitespace to submenu
items to account for the additional elements in the title.
Since commit ef1e27966db2 turned DashItemContainer into an StWidget,
the destroy() method overrides the ClutterActor method, which is at
the very least bad style. Instead, follow the usual pattern of using
a ::destroy handler.
dash: Make sure item labels are only destroyed once
Labels are currently destroyed from both animateOutAndDestroy()
and destroy(), which now (rightfully) triggers a gjs warning. As
the label is created unconditionally since commit 36e5ae4a250,
mirror that and always release it in destroy() and hide it
elsewhere.
Florian Müllner [Sat, 25 Nov 2017 06:15:39 +0000 (07:15 +0100)]
build: Stop detecting features automatically
Auto-detect options add some convenience for platforms where a
particular feature isn't available - systemd on non-Linux OS comes
to mind - but the downside is that it is easy to accidentally build
without a desired feature. We consider the latter much more serious
nowadays, so turn our auto-detect options into regular boolean
options.
status/keyboard: Reset menuItems and Label objects on change
In the current code it could happen that we've menuItems and indicatorLabels
for sources that aren't anymore around, because in case a source is removed
we don't cleanup the their container objects.
Also, we should nullify InputManager's _currentSource when sources change
or it might point to some invalid data again.
So it could happen that we try to access an invalid menuitem or label
if a source change happens mentioning a source that has been deleted.
Carlos Garnacho [Thu, 16 Nov 2017 16:27:12 +0000 (17:27 +0100)]
overview: Protect ::drag-end handlers
These end up emitting item-drag-end/window-drag-end pretty much
without checks. Given the MetaDnd object may end up emitting
::drag-leave as a result of the plugin ending its grab, this
would result on spurious emission of those events and subsequent
warnings.
For extra paranoia, the _inDrag variable has been split into
_inItemDrag/_inWindowDrag so we can't cross the streams.
Florian Müllner [Mon, 23 Oct 2017 08:06:18 +0000 (10:06 +0200)]
windowAttentionHandler: Follow app policy for attention notifications
While window attention notifications are created by the shell itself
rather than applications (most likely as a result of focus stealing
prevention), users still commonly link them to the application for
which they are shown. It makes therefore sense to follow the appropriate
policy set by the user rather than showing them unconditionally.
Florian Müllner [Thu, 5 Oct 2017 08:06:10 +0000 (10:06 +0200)]
dialog: Reset initial focus on destroy
Since 0b02f757f862 we track the button that should have key focus
when the dialog is opened. However when the dialog is reused, the
button may get destroyed - clear the initial focus in that case to
allow setButton() to set a new one.
main: use SA_NODEFER to track signals to being able to raise from alarm
After we receive one of the tracked signals, in case we get stuck inside
the gjs_dumpstack () call, we use an alarm to raise the previously emitted
signal, however without using SA_NODEFER, the raise inside the alarm handler
will be ignored.
To avoid to handle new signals caused by the handler calls, once we get the
first signal, we just ignore them all as they could only lead to dirty traces.
Also, cleaning up a bit the code, and disabling the shell log handler
in dump_gjs_stack_alarm_sigaction since this might lead to a new
gjs_dumpstack () request.
main: call gjs_dumpstack on aborts and traps and optionally on segfaults
In order to debug issues triggered by JS code we might need to
see the stack of it, this is not normally visible in static stack traces,
thus we need to call gjs_dumpstack () before dying the process.
Intercepting signals SIGABRT, SIGTRAP (needed for catching fatal glib
errors) SIGFPE and SIGIOT by default, while introducing a new
'backtrace-segfaults' flag for the SHELL_DEBUG environment variable to
do the same on SIGSEGV and SIGBUS (this is a precaution to avoid that we
corrupt the stack for automatic errors trackers).
In any case after dumping the stack we raise the signal again, in order
to make the system aware of it.
bluetooth: Add missing 'BluetoothHardwareAirplaneMode' to D-Bus interface
This D-Bus property was never been added here, which caused inconsistencies
under some scenarios (e.g. coming back from suspend) if some devices were
previously paired, since _sync() would then make the bluetooth menu visible
unconditially, because of the proxied property evaluating to 'false'.
Adding this to the D-Bus interface makes sure that it's no longer undefined
and returns the right value, fixing the bug.
Florian Müllner [Sun, 15 Oct 2017 22:09:52 +0000 (00:09 +0200)]
popupMenu: Fix setting ImageMenuItem's icons via strings
Commit 28ca96064bcb added support for setting PopupImageMenuItem's icons
via GIcons as well as via strings. However as the check whether an object
implements the GIcon interface only works on GObjects, specifying an icon
name was broken. Fix that to actually allow both strings and GIcons.
Florian Müllner [Fri, 13 Oct 2017 14:17:46 +0000 (16:17 +0200)]
gdm: Remove separate greeter universal access icon
The idea behind always showing the icon on the login screen is that
the users' needs aren't known at that point. However we can achieve
the same behavior by including the 'always-show-universal-access-status'
key in GDM's presets, so drop the special-case.
Jonas Ådahl [Sat, 7 Oct 2017 00:26:24 +0000 (20:26 -0400)]
layout: Unset primary and bottom monitor when headless
We were handling being initially headless by only setting the primary
and bottom monitor if there was any primary monitor, then checking the
primary monitor reference before making calls assuming there was any
monitors.
What we didn't do was unset the primary and bottom monitor when going
headless, meaning that temporarly disconnecting a monitor while having
windows open caused an assert to be triggered due to various code paths
taking the path assuming there are valid monitors.
Unsetting both the primary and bottom monitor when going headless avoids
the code paths in the same way as they were avoided when starting
headless.
We are already closing top bar menus on session mode changes, but
as this behavior makes sense for any other menus as well - dash
context menus or the background menu for instance - just generalize
the behavior.
Commit 1c7a3ee61b1892 broke setting the initial key focus for default
buttons added via addButton(). Fix this by allowing the dialog class
to provide a different default widget to ModalDialog than the entire
dialog itself.
Florian Müllner [Sun, 20 Aug 2017 15:53:41 +0000 (17:53 +0200)]
workspace: Focus clone when highlighting
We don't use different hover- and focus indications for window previews,
so using keyboard navigation after hovering a clone often has surprising
results when the previous focus window wasn't the selected one. Address
this by simply moving the keyboard focus with the highlight.
Florian Müllner [Sun, 20 Aug 2017 09:19:39 +0000 (11:19 +0200)]
workspace: Fix tab navigation order
By default the focus chain uses the same order as the list returned
by clutter_actor_get_children(), which corresponds to the z-order.
This doesn't work well in the window picker, where clones follow
the stacking of windows to ensure a correct overview transition,
but previews are laid out purely based on space efficiency. As
a result, the order in which window previews are navigated when
tabbing around is essentially random. Fix this by providing a
focus chain implementation that is based on the visual layout
of the previews rather than the stacking.
Jonas Ådahl [Wed, 12 Apr 2017 06:46:54 +0000 (14:46 +0800)]
ui: Improve handling being headless
Don't assume there will always be a primary (logical) monitor, or any
(logical) monitor at all. This includes not allocating / layouting /
styling correctly when being headless.
The initial background loading will also be delayed until there are any
(logical) monitors connected.
The script was used by a distcheck hook to check for files that are
in git, but not the distributed tarball. We dropped autotools support,
so the script is no longer used - in fact, it isn't useful anymore,
as `ninja dist` generates tarballs from git itself rather than from
a separately maintained list of files ...
shell-app: Don't transition to STOPPED while still in the STARTING state
Similar to what it's done in _shell_app_add_window(), rely on calling
shell_app_sync_running_state() when removing windows too, to avoid
transitioning to STOPPED while it's still in the STARTING state.
This makes the logic compliant with the startup notification spec and
prevents stopping an application too early if it has a splash screen
that has been closed before the application's main window is shown.
On the screen shield, the only possible interaction is lifting the
shield. The on-screen-keyboard is not useful for that, and the drag
gesture from the bottom may in fact conflict with dragging up the
shield, so disable it.
messageTray: Check monitor's existence before trying to update the state
Not doing this will throw a backtrace when running on headless mode and
trying to show a notification, due to Main.layoutManager.primaryMonitor
being undefined, so it's better to return early.
Krzesimir Nowak [Fri, 18 Nov 2016 14:52:46 +0000 (15:52 +0100)]
notificationDaemon: Guard against invalid D-Bus object paths
When making any D-Bus call through the GDBus' proxy wrapper with an
invalid D-Bus object path, gnome-shell hangs.
Supposedly FdoApplicationProxy constructor should validate the passed
D-Bus object path and throw an error if the path is invalid. Since it
does not do that, we work it around by making sure that the deduced
D-Bus object path is valid or throw an exception if the path is not.
Krzesimir Nowak [Fri, 18 Nov 2016 14:52:46 +0000 (15:52 +0100)]
notificationDaemon: Consider hyphens for app ID to object path translations
Some application IDs contain hyphens, which are not allowed in D-Bus object
paths, so we need to update the translation by converting them to something
that's a valid object path. This is consistent with what GApplication does.
Florian Müllner [Thu, 10 Aug 2017 19:41:36 +0000 (21:41 +0200)]
meson: Use build_rpath instead of custom linker args
Meson now warns when specifying the rpath in linker flags instead
of using the newly added build_rpath option - now that we already
depend on the latest release, there's no reason anyway to not use
the nicer syntax.
shadow: Only use correctly sized textures for shortcut
When creating a shadow for a ClutterTexture, we currently use the
underlying CoglTexture directly instead of rendering the actor to
an offscreen buffer. This assumes that the CoglTexture is directly
suitable as shadow source, which isn't necessarily the case - it
may have a very different size than what is shown and scaled up or
down by the hardware. In that case we end up with a scaled shadow
texture as well, which messes up the desired blur effect - the
result will be too light when scaling up, or too sharp when scaling
down. To fix this, only take the shortcut when a ClutterTexture's
underlying texture has the correct size and fall back to offscreen
rendering otherwise.
Commit 289f982949c2 broke all remote providers when adding support for
non-auto-started search providers: Whether the provider should be
auto-started needs to be known in the constructor, so setting the
property on the constructed object doesn't work.
Ensure that key-above-tab for selecting window application is consistent
with the down key.
After focus an application in the switcher:
- if you press down, the window thumbnails are previewed, and first element
is selected
- if you press key-above-tab, the window thumbnails are previewed, however
the second element is directly selected.
Make both interactions always select the first element.
Ray Strode [Thu, 14 Sep 2017 18:46:13 +0000 (14:46 -0400)]
gtk-embed: ensure we only listen for window-created events once
If a tray icon gets a mapped and unmapped and the mapped again
in quick succession, we can end up with multiple handlers
listening for window creation events.
This commit tries to guard against that by only listening for
window-created events when we don't know the actor associated
with the icon.
Bastien Nocera [Thu, 24 Aug 2017 11:06:45 +0000 (13:06 +0200)]
remoteSearch: Allow search providers not to auto-start
This would be used by search providers which only operate on data in the
running instance, such as the terminal's search provider which finds the
shell in the tab matching the search text.
projects / gnome-shell.git / log