]> dgit.raspbian.org Git - gnome-shell.git/log
gnome-shell.git
11 years agopanelMenu: Consider vertical margins when computing max-height
Florian Müllner [Fri, 13 Feb 2015 19:52:11 +0000 (20:52 +0100)]
panelMenu: Consider vertical margins when computing max-height

When opening a panel menu, we set it's max-height to the available
work-area height to keep menus with scrollable content from growing
outside the monitor. However a menu that extends all the way down
to the bottom edge does not look great either, so also take margins
into account here.

https://bugzilla.gnome.org/show_bug.cgi?id=744498

11 years agotelepathyClient: Fix a translator comment
Florian Müllner [Fri, 6 Feb 2015 23:22:47 +0000 (00:22 +0100)]
telepathyClient: Fix a translator comment

Ouch, classy - we are telling translators to use a '24h' time
format for the '12h' time format string.
Luckily, only a handful of translations actually followed the
comment (de,hu,id,is,kk,nb,nl), and most of the corresponding
locales do not support 12-hour format anyway (only is_IS, at
least on Fedora).

11 years agolookingGlass: Remove ugly hack
Florian Müllner [Sat, 14 Feb 2015 15:51:20 +0000 (16:51 +0100)]
lookingGlass: Remove ugly hack

We don't need to hard-code any magic offsets to hide the top corners
and border, we can just not add them in the first place.

11 years agoUpdated Turkish translation
Muhammet Kara [Sat, 14 Feb 2015 07:47:33 +0000 (07:47 +0000)]
Updated Turkish translation

11 years agoappDisplay: Do not duplicate "New Window" action
Florian Müllner [Fri, 13 Feb 2015 01:03:33 +0000 (02:03 +0100)]
appDisplay: Do not duplicate "New Window" action

We assume that applications that export a 'new-window' action can open
a new window, so we add an appropriate entry to the context menu.
However this duplicates functionality if the application already
exposes the action via the desktop file - don't add our own entry
in that case.

https://bugzilla.gnome.org/show_bug.cgi?id=744446

11 years agomain: Fix structured_log() call
Florian Müllner [Fri, 13 Feb 2015 22:02:06 +0000 (23:02 +0100)]
main: Fix structured_log() call

Meh, I should have caught that one in review ...

11 years agoDrop use of libgsystem, move single API call into shell-global.c
Colin Walters [Fri, 13 Feb 2015 09:27:27 +0000 (04:27 -0500)]
Drop use of libgsystem, move single API call into shell-global.c

See https://mail.gnome.org/archives/desktop-devel-list/2015-February/msg00121.html

We presently only indirectly link to it via this one call; until GLib
gains structured logging, simply import the code to wrap the journal
logging in an introspectable API into Shell.

https://bugzilla.gnome.org/show_bug.cgi?id=744457

11 years agoUpdated Czech translation
Marek Černocký [Thu, 12 Feb 2015 09:00:48 +0000 (10:00 +0100)]
Updated Czech translation

11 years agorun-js-test: Remove old code to shove X11 events through
Jasper St. Pierre [Thu, 12 Feb 2015 00:31:13 +0000 (16:31 -0800)]
run-js-test: Remove old code to shove X11 events through

11 years agorecorder: Remove unused field
Jasper St. Pierre [Wed, 11 Feb 2015 08:59:51 +0000 (00:59 -0800)]
recorder: Remove unused field

11 years agoscreenshot: Use MetaCursorTracker to fetch the cursor position
Jasper St. Pierre [Wed, 11 Feb 2015 08:58:42 +0000 (00:58 -0800)]
screenshot: Use MetaCursorTracker to fetch the cursor position

ClutterInputDevice will use the last place it knows the cursor to be,
which isn't always up to date when running under X11.

11 years agoUpdated Indonesian translation
Andika Triwidada [Wed, 11 Feb 2015 08:06:31 +0000 (08:06 +0000)]
Updated Indonesian translation

11 years agoAdd an OSD monitor labeler exposed on DBus
Rui Matos [Mon, 26 Jan 2015 14:52:10 +0000 (15:52 +0100)]
Add an OSD monitor labeler exposed on DBus

This DBus API is intended to be used by gnome-control-center's
displays panel to show monitor labels.

Each output (i.e. hardware monitor) identified by its
org.gnome.Mutter.DisplayConfig API ID has at most one label. On
mirrored setups, all the labels for outputs corresponding to the same
logical monitor (i.e. showing the same contents in the same mode) are
shown together.

At most, only one DBus client at a time is allowed to show labels.

https://bugzilla.gnome.org/show_bug.cgi?id=743744

11 years agoWorkspacesDisplay: update the primary monitor index too
Rui Matos [Wed, 4 Feb 2015 14:35:34 +0000 (15:35 +0100)]
WorkspacesDisplay: update the primary monitor index too

The primary monitor index might change so we need to update it too.

https://bugzilla.gnome.org/show_bug.cgi?id=743993

11 years agoStTextureCache: do not monitor files inside GResource
Cosimo Cecchi [Wed, 4 Feb 2015 20:23:30 +0000 (12:23 -0800)]
StTextureCache: do not monitor files inside GResource

This is wasteful, since GResource does not support file monitoring.
Further, doing so will trigger a fallback code path in GLib that polls
every second in a thread, which is doubly wasteful.

https://bugzilla.gnome.org/show_bug.cgi?id=744013

11 years agosystem: Fix visibility of suspend button
Florian Müllner [Wed, 4 Feb 2015 11:01:47 +0000 (12:01 +0100)]
system: Fix visibility of suspend button

It depends on the availability of suspend, not shutdown ...

11 years agoUpdated Slovak translation
Dušan Kazik [Sun, 1 Feb 2015 21:55:51 +0000 (21:55 +0000)]
Updated Slovak translation

11 years agoUpdated Esperanto translation
Kristjan SCHMIDT [Sun, 1 Feb 2015 03:16:13 +0000 (04:16 +0100)]
Updated Esperanto translation

11 years agoRevert "shell-app: Fix typo"
Florian Müllner [Fri, 30 Jan 2015 17:01:57 +0000 (18:01 +0100)]
Revert "shell-app: Fix typo"

This reverts commit ead4c2b863fe8221a27ff17d1515e8517caadaa3.

11 years agoshell-app: Fix typo
Florian Müllner [Fri, 30 Jan 2015 16:57:09 +0000 (17:57 +0100)]
shell-app: Fix typo

Commit 086fcd4a32 broke the build, use the correct variable to fix.

https://bugzilla.gnome.org/show_bug.cgi?id=743729

11 years agoshell-app: Only raise windows on the same workspace when activating them
Jasper St. Pierre [Fri, 30 Jan 2015 16:35:46 +0000 (11:35 -0500)]
shell-app: Only raise windows on the same workspace when activating them

For some reason, we never did this, even though this is what the comment
says we should be doing.

11 years agoUpdated Icelandic translation
Sveinn í Felli [Thu, 29 Jan 2015 15:20:59 +0000 (15:20 +0000)]
Updated Icelandic translation

11 years agoApp Picker: Prevent drag if favorite-apps is locked.
Murray Cumming [Tue, 27 Jan 2015 21:10:45 +0000 (22:10 +0100)]
App Picker: Prevent drag if favorite-apps is locked.

Because there's nothing (in single-monitor setups) that could
take the drop in this case.

* js/ui/appDisplay.js:
  AllView._loadApps(), FrequentView._loadApps(): Pass
  an isDraggable parameter when creating the AppIcons,
  depending on whether the favorite-apps key is locked.
  AppIcon._init(): Check for isDraggable in the params and
  do not create _draggable if it was specified, to prevent a
  drag from starting.
  AppIcon.popupMenu(): Check _draggable before trying to call
  fakeRelease on it.
* js/ui/dash.js: Dash._createAppItem(): Check AppIcon._draggable
  before trying to connect to its signals.

https://bugzilla.gnome.org/show_bug.cgi?id=741325

11 years agodash: Improve feedback when favorite-apps key is locked
David King [Thu, 11 Dec 2014 15:22:19 +0000 (15:22 +0000)]
dash: Improve feedback when favorite-apps key is locked

In a lockdown scenario, where the favorite-apps GSettings key is not
writable, hide the menu items for adding and removing favorites from the
dash menu. Additionally, reject drops to the dash for DND.

https://bugzilla.gnome.org/show_bug.cgi?id=741325

11 years agoPortalHelper: Make translator comment appear
Bastien Nocera [Wed, 28 Jan 2015 12:47:56 +0000 (13:47 +0100)]
PortalHelper: Make translator comment appear

Use "/* */" for the translator comment, so that xgettext can extract it.

https://bugzilla.gnome.org/show_bug.cgi?id=736182

11 years agoUpdated Hebrew Translation
Yosef Or Boczko [Mon, 26 Jan 2015 19:16:10 +0000 (21:16 +0200)]
Updated Hebrew Translation

11 years agoUpdated Icelandic translation
Sveinn í Felli [Mon, 26 Jan 2015 13:16:43 +0000 (13:16 +0000)]
Updated Icelandic translation

11 years agoUpdated German translation
GNOME Translation Robot [Sun, 25 Jan 2015 20:29:21 +0000 (20:29 +0000)]
Updated German translation

11 years agoAdded Icelandic translation
Sveinn í Felli [Fri, 23 Jan 2015 15:36:04 +0000 (15:36 +0000)]
Added Icelandic translation

11 years agologinDialog: fix reactivity of first user in user list
Ray Strode [Thu, 22 Jan 2015 19:00:01 +0000 (14:00 -0500)]
loginDialog: fix reactivity of first user in user list

After the login banner is shown and hidden, the first user
in the user list becomes non-reactive.  This is because the
banner is given an opacity of 0, but still allocated.

This commit fixes that by hiding the banner explicitly.

https://bugzilla.gnome.org/show_bug.cgi?id=743370

11 years agologinDialog: reset greeter proxy on auth prompt reset
Ray Strode [Thu, 22 Jan 2015 18:39:46 +0000 (13:39 -0500)]
loginDialog: reset greeter proxy on auth prompt reset

Normally when a user uses the login screen to log in, the
login screen gets killed and the user session takes over
the display.

This doesn't happen for wayland sessions, though. Instead,
the login screen gets reset, and the wayland session is started
on another VT.

The greeter proxy object needs to be recreated after this reset,
since it's associated with state no longer coupled to the login
screen after the reset.

This commit moves greeter proxy creation to happen at reset time.

https://bugzilla.gnome.org/show_bug.cgi?id=743371

11 years agoUpdate Aragonese translation
Daniel Martinez [Thu, 22 Jan 2015 18:24:13 +0000 (19:24 +0100)]
Update Aragonese translation

11 years agodocs: Update for renamed file
Florian Müllner [Wed, 21 Jan 2015 15:01:57 +0000 (16:01 +0100)]
docs: Update for renamed file

11 years agoBump version to 3.15.4
Florian Müllner [Wed, 21 Jan 2015 14:11:52 +0000 (15:11 +0100)]
Bump version to 3.15.4

Update NEWS.

11 years agowindowManager: Kill a comment
Florian Müllner [Mon, 19 Jan 2015 22:40:07 +0000 (23:40 +0100)]
windowManager: Kill a comment

The code described by the comment was moved away in commit eda27d51,
so it is not misleading at best. It wasn't too useful to begin with,
so kill it off rather than moving it to the correct place ...

11 years agoworkspacesView: Remove obsolete constant
Florian Müllner [Thu, 18 Dec 2014 17:14:04 +0000 (18:14 +0100)]
workspacesView: Remove obsolete constant

11 years agoUpdated Friulian translation
Fabio Tomat [Tue, 20 Jan 2015 21:15:02 +0000 (21:15 +0000)]
Updated Friulian translation

11 years agoRemove the dependency on XTest
Jasper St. Pierre [Mon, 19 Jan 2015 00:49:13 +0000 (16:49 -0800)]
Remove the dependency on XTest

We haven't been using XTest since edd66c4.

11 years agowindowManager: Allow moving a window above the top workspace
Florian Müllner [Thu, 15 Jan 2015 23:13:58 +0000 (00:13 +0100)]
windowManager: Allow moving a window above the top workspace

When using dynamic workspaces, a new workspace will be appended
when moving a window down to the last (empty) workspace. It makes
sense to extend the behavior in the opposite direction, and prepend
a new workspace when moving a window up from the first workspace.

https://bugzilla.gnome.org/show_bug.cgi?id=665764

11 years agowindowManager: Update active workspace after inserting a new one
Florian Müllner [Thu, 15 Jan 2015 23:32:27 +0000 (00:32 +0100)]
windowManager: Update active workspace after inserting a new one

New workspaces are inserted by shifting all windows on workspaces
below the insertion position down. As a result, when the new
workspace is inserted before the active one, we end up with
the illusion of a workspace switch. Instead, activate the workspace
on which the windows from the active one ended up.

https://bugzilla.gnome.org/show_bug.cgi?id=665764

11 years agowindowManager: Don't shift OR windows when inserting a workspace
Florian Müllner [Fri, 16 Jan 2015 00:19:37 +0000 (01:19 +0100)]
windowManager: Don't shift OR windows when inserting a workspace

We are not supposed to mess around with OR windows, so don't try
to shift them to a different workspace. This fixes a warning with
newer versions of mutter.

https://bugzilla.gnome.org/show_bug.cgi?id=665764

11 years agowindowManager: Split out workspace insertion from thumbnails
Florian Müllner [Thu, 15 Jan 2015 22:03:24 +0000 (23:03 +0100)]
windowManager: Split out workspace insertion from thumbnails

We will soon allow to insert a new workspace by other means than
DND in between workspace thumbnails, so move the relevant code
to a new windowManager method.

https://bugzilla.gnome.org/show_bug.cgi?id=665764

11 years agoLower slider scrolling step to 2%
Adel Gadllah [Fri, 16 Jan 2015 17:24:33 +0000 (18:24 +0100)]
Lower slider scrolling step to 2%

https://bugzilla.gnome.org/show_bug.cgi?id=742648

11 years agoshell-screenshot: Disable unredirection while taking screenshots
Adel Gadllah [Thu, 4 Dec 2014 14:38:28 +0000 (15:38 +0100)]
shell-screenshot: Disable unredirection while taking screenshots

Otherwise we grab wrong contents.

https://bugzilla.gnome.org/show_bug.cgi?id=741114

11 years agoStTextureCache: adapt to MetaWindow changing icon prop type
Rui Matos [Mon, 12 Jan 2015 20:21:39 +0000 (21:21 +0100)]
StTextureCache: adapt to MetaWindow changing icon prop type

Mutter change is
https://git.gnome.org/browse/mutter/commit/?id=af7f51b992b20da543312fa9fdf1dcef39704b37

https://bugzilla.gnome.org/show_bug.cgi?id=742824

11 years agoibusManager: Disable IBus input sources on password entries
Takao Fujiwara [Thu, 27 Nov 2014 08:23:01 +0000 (17:23 +0900)]
ibusManager: Disable IBus input sources on password entries

Input method preedit text needs to be disabled on password entries
for security and usability reasons.
IBus 1.5.7 provides the signal set-content-type so that panel UIs can
handle these special purpose input entries:
https://github.com/ibus/ibus/commit/6ca5ddb302c9

Unfortunately IBus versions older than 1.5.10 have a bug which causes
spurious set-content-type emissions when switching input focus that
temporarily lose purpose and hints defeating its intended semantics
and confusing users. We thus don't use it in that case.

https://bugzilla.gnome.org/show_bug.cgi?id=730628

11 years agoUpdated Spanish translation
Daniel Mustieles [Mon, 12 Jan 2015 11:19:02 +0000 (12:19 +0100)]
Updated Spanish translation

11 years agoRequire gsettings-desktop-schemas >= 3.13.1
Adel Gadllah [Sun, 11 Jan 2015 14:50:08 +0000 (15:50 +0100)]
Require gsettings-desktop-schemas >= 3.13.1

https://bugzilla.gnome.org/show_bug.cgi?id=742748

11 years agorecorder: Cleanups
Adel Gadllah [Sun, 11 Jan 2015 14:47:52 +0000 (15:47 +0100)]
recorder: Cleanups

11 years agorecorder: Don't compute the buffer size twice
Adel Gadllah [Sun, 11 Jan 2015 14:42:18 +0000 (15:42 +0100)]
recorder: Don't compute the buffer size twice

11 years agorecorder: Switch to vp9
Adel Gadllah [Sun, 11 Jan 2015 11:16:40 +0000 (12:16 +0100)]
recorder: Switch to vp9

Currently we have been using the vp8 codec because it was the best unencumbered codec at that time. With vp9 we now have a successor that leads to smaller
files at at the same video quality and has been supported by current browsers
for a while.

With the raise of hidpi and 4K displays we need a better codec that handles
those resolutions better, so switch to vp9.

https://bugzilla.gnome.org/show_bug.cgi?id=742744

11 years agorecorder: Stop using deprecated cogl_read_pixels
Adel Gadllah [Sun, 11 Jan 2015 11:49:44 +0000 (12:49 +0100)]
recorder: Stop using deprecated cogl_read_pixels

11 years agocalendar: update current day highlight on day change
Sebastian Keller [Tue, 6 Jan 2015 23:35:13 +0000 (00:35 +0100)]
calendar: update current day highlight on day change

https://bugzilla.gnome.org/show_bug.cgi?id=742492

11 years agoslider: Don't hardcode scroll interval for smooth scrolling
Adel Gadllah [Sat, 10 Jan 2015 12:12:48 +0000 (13:12 +0100)]
slider: Don't hardcode scroll interval for smooth scrolling

Use the SLIDER_SCROLL_STEP constant instead for consistency.

https://bugzilla.gnome.org/show_bug.cgi?id=742648

11 years agoUpdated German translation
Bernd Homuth [Thu, 8 Jan 2015 22:40:32 +0000 (22:40 +0000)]
Updated German translation

11 years agoUpdated Brazilian Portuguese translation
Rafael Ferreira [Thu, 8 Jan 2015 18:44:49 +0000 (18:44 +0000)]
Updated Brazilian Portuguese translation

11 years agosearch: remove redundant check
Cosimo Cecchi [Wed, 7 Jan 2015 10:16:04 +0000 (18:16 +0800)]
search: remove redundant check

We check for (metasNeeded.length == 0) at the beginning of the function,
which is only ever called when when a non-zero number of results is
received back from the provider. Effectively, this means that
(metas.length != metasNeeded.length) will also catch (metas.length == 0)
and print a nicer message to the log.

11 years agosearch: improve log message when metas length mismatch
Cosimo Cecchi [Wed, 7 Jan 2015 10:15:30 +0000 (18:15 +0800)]
search: improve log message when metas length mismatch

Log also the name of the provider and the expected/received metas.

11 years agosearch: don't forget to notify caller when failing to ensure actors
Cosimo Cecchi [Wed, 7 Jan 2015 09:53:30 +0000 (17:53 +0800)]
search: don't forget to notify caller when failing to ensure actors

The updateSearch() function is called in SearchResults every time new
search hits are available from a search provider; SearchResults will
wait for updateSearch() to complete in a callaback, to update the
overall progress of the search operation.

updateSearch() will call _ensureResultActors(), which will in turn call
getResultMetas() on the search provider, which is an operation that can
fail arbitrarily or return inconsistent data, as it's entirely in the
hands of the search provider.

In case _ensureResultActors() returns a failure, updateSearch() is
currently failing to notify the passed-in callback, which might leave
SearchResults in an inconsistent state: make sure the asynchronous flow
always ends up with a notification to the updateSearch() callback.

11 years agoUpdated Czech translation
Marek Černocký [Wed, 7 Jan 2015 07:53:43 +0000 (08:53 +0100)]
Updated Czech translation

11 years agowindowManager: Make sure to reset the resizePopup after using it
Jasper St. Pierre [Fri, 2 Jan 2015 17:27:52 +0000 (09:27 -0800)]
windowManager: Make sure to reset the resizePopup after using it

So that we'll recreate it the next time we want to show it. Otherwise,
we'll try to call things on a half-destroyed ResizePopup and end up
causing errors instead of showing the user their resize popup.

11 years agoUpdated Hungarian translation
Balázs Úr [Wed, 31 Dec 2014 10:21:00 +0000 (10:21 +0000)]
Updated Hungarian translation

11 years agoUpdated Norwegian bokmål translation.
Kjartan Maraas [Tue, 30 Dec 2014 23:46:24 +0000 (00:46 +0100)]
Updated Norwegian bokmål translation.

11 years agoUpdated Russian translation
Stas Solovey [Tue, 30 Dec 2014 23:42:17 +0000 (23:42 +0000)]
Updated Russian translation

11 years agoUpdated Basque language
Inaki Larranaga Murgoitio [Tue, 30 Dec 2014 14:14:01 +0000 (15:14 +0100)]
Updated Basque language

11 years agoUpdated Bengali (India) translation
Saibal Ray [Tue, 30 Dec 2014 11:20:57 +0000 (11:20 +0000)]
Updated Bengali (India) translation

11 years agowindowManager: Implement the resize popup here
Jasper St. Pierre [Tue, 30 Dec 2014 01:30:38 +0000 (17:30 -0800)]
windowManager: Implement the resize popup here

mutter recently removed its implementation, so add a simple one here.

11 years agowindow-tracker: remove now unused launched_pid_to_app
Sebastian Keller [Mon, 8 Dec 2014 19:12:03 +0000 (20:12 +0100)]
window-tracker: remove now unused launched_pid_to_app

https://bugzilla.gnome.org/show_bug.cgi?id=736527

11 years agowindow-tracker: Fix pid based window/app association
Sebastian Keller [Mon, 8 Dec 2014 18:53:39 +0000 (19:53 +0100)]
window-tracker: Fix pid based window/app association

The lookup table used by get_app_from_window_pid contained only pids of
apps launched by gnome-shell itself, but not pids of apps running before
gnome-shell was (re-)started. Also the pids in that table might not even
be the pid of the process that is actually showing the window if wrapper
scripts are used.

Instead use shell_window_tracker_get_app_from_pid which uses the pids
from the windows themselves.

This removes the only use of launched_pid_to_app which will be removed
in the following commit.

This fixes: https://bugzilla.gnome.org/show_bug.cgi?id=736527

11 years agoUpdated Slovenian translation
Matej Urbančič [Thu, 25 Dec 2014 18:18:45 +0000 (19:18 +0100)]
Updated Slovenian translation

11 years agoUpdated Indonesian translation
Andika Triwidada [Wed, 24 Dec 2014 04:51:40 +0000 (04:51 +0000)]
Updated Indonesian translation

11 years agoshellDBus: Add mode parameter to AcceleratorActivated signal
Florian Müllner [Wed, 12 Nov 2014 19:49:24 +0000 (20:49 +0100)]
shellDBus: Add mode parameter to AcceleratorActivated signal

This will allow g-s-d to handle actions differently based on the
current mode - namely, allow the power button when locked, but
make sure to never show any dialogs in that case.

https://bugzilla.gnome.org/show_bug.cgi?id=711682

11 years agoshellDBus: Change AcceleratorActivated signature
Florian Müllner [Wed, 12 Nov 2014 20:21:44 +0000 (21:21 +0100)]
shellDBus: Change AcceleratorActivated signature

Adding new parameters to the signal currently will break keybindings
until gnome-settings-daemon is updated to the new API as well.
Put additional parameters into a dictionary instead to make future
extensions easier.

https://bugzilla.gnome.org/show_bug.cgi?id=711682

11 years agoBump version to 3.15.3
Florian Müllner [Fri, 19 Dec 2014 10:47:50 +0000 (11:47 +0100)]
Bump version to 3.15.3

Update NEWS.

11 years agowindowManager: Update stacking during workspace switches
Florian Müllner [Wed, 17 Dec 2014 21:57:44 +0000 (22:57 +0100)]
windowManager: Update stacking during workspace switches

When animating workspace switches, windows on the old and new workspaces
are temporarily reparented. If windows are restacked, those windows will
thus be ignored by mutter until meta_switch_workspace_completed() resyncs
the stacking at the end of the animation.
As a result, activating a window on another workspace that is not on top
of the stack is very noticeably a two-step operation of switching workspace
and raising the window. There is a technical reason for that order[0], but
we can avoid the visible disruption by manually syncing the stack during
the switch operation.

[0] https://git.gnome.org/browse/mutter/tree/src/core/workspace.c#n590

https://bugzilla.gnome.org/show_bug.cgi?id=741680

11 years agoRename KeyBindingMode to ActionMode
Florian Müllner [Thu, 11 Dec 2014 14:35:40 +0000 (15:35 +0100)]
Rename KeyBindingMode to ActionMode

The keybinding mode is no longer used exclusively for actions triggered
by keybindings, so reflect this by a more generic name.

https://bugzilla.gnome.org/show_bug.cgi?id=740237

11 years agogestures: Restrict actions based on keybindingMode
Florian Müllner [Thu, 27 Nov 2014 19:46:50 +0000 (20:46 +0100)]
gestures: Restrict actions based on keybindingMode

Just like keybindings and the message tray pointer barrier, gestures
don't always make sense - for instance, swiping up the screen shield
should not trigger the message tray just as the SelectArea action around
the left edge should not open the overview.
To avoid this, restrict gestures based on the current keybinding mode.

https://bugzilla.gnome.org/show_bug.cgi?id=740237

11 years agologinDialog: use two column view if banner message long
Ray Strode [Tue, 11 Nov 2014 14:11:01 +0000 (09:11 -0500)]
loginDialog: use two column view if banner message long

Frequently banner messages are longer than can reasonable
fit in a one column view, which leads to a smooshed layout.

This commit changes the layout to a two column view, with the
banner on the left and the prompt on the right, if the banner
message is long enough that it can't fit well above the prompt.
If there isn't enough space for two columns then we keep the
one column layout but add scrollbars.

https://bugzilla.gnome.org/show_bug.cgi?id=703972

11 years agologinDialog: display banner message when disable-user-list=true
Ray Strode [Tue, 11 Nov 2014 14:11:01 +0000 (09:11 -0500)]
loginDialog: display banner message when disable-user-list=true

The login screen supports showing a banner message which admins
can use to mention login rules or disclaimers.

This message only shows up currently if the user list is enabled.
Most people who want to show a banner message also want to disable
the user list.

This commit moves the banner message to display when the user is
prompted for login credentials instead of when showing the user
list. It also adds a scrollbar if the message is too long.

https://bugzilla.gnome.org/show_bug.cgi?id=703972

11 years agologinDialog: allocate children manually
Ray Strode [Mon, 10 Nov 2014 19:36:07 +0000 (14:36 -0500)]
loginDialog: allocate children manually

The login screen is pretty custom full screen container and the standard
layout managers aren't really a good fit for the kind of layout that's
happening. This will be even more problematic with upcoming changes
to login banners, so we need to switch techniques.

This commit moves login dialog over to using a custom allocate handler
that has specific domain knowledge of the parts of the login screen
and where they go.

https://bugzilla.gnome.org/show_bug.cgi?id=703972

11 years agoUpdated Kazakh translation
Baurzhan Muftakhidinov [Thu, 18 Dec 2014 00:40:51 +0000 (00:40 +0000)]
Updated Kazakh translation

11 years agoUpdated Basque language
Inaki Larranaga Murgoitio [Wed, 17 Dec 2014 15:31:51 +0000 (16:31 +0100)]
Updated Basque language

11 years agoCalendar: ignore timeouts starting the calendar-server
Giovanni Campagna [Thu, 11 Dec 2014 00:41:48 +0000 (16:41 -0800)]
Calendar: ignore timeouts starting the calendar-server

In certain cases the timeout for starting the calendar helper can
be reached but the calendar helper still loads fine. If so, just
ignore the timeout and wait until we get a notification from
dbus of the successful start.

https://bugzilla.gnome.org/show_bug.cgi?id=735308

11 years agocalendar-server: activate evolution-source-registry manually at startup
Giovanni Campagna [Thu, 11 Dec 2014 00:32:17 +0000 (16:32 -0800)]
calendar-server: activate evolution-source-registry manually at startup

g_dbus_proxy_new() (and library calls that wrap it) has an hardcoded
timeout of 25 seconds, which is insufficient for starting up e-s-r
in certain setups. Avoid a timeout error by starting the service
manually with a longer timeout before hand.
Also demote the error to a warning + exit failure instead of
a crash, to avoid triggering abrt reports.

https://bugzilla.gnome.org/show_bug.cgi?id=735308

11 years agoUpdated Kazakh translation
Baurzhan Muftakhidinov [Sun, 14 Dec 2014 14:08:39 +0000 (14:08 +0000)]
Updated Kazakh translation

11 years agoUpdated Turkish translation
Muhammet Kara [Sat, 13 Dec 2014 17:46:23 +0000 (17:46 +0000)]
Updated Turkish translation

11 years agoUpdated Swedish translation
Josef Andersson [Thu, 11 Dec 2014 19:06:46 +0000 (19:06 +0000)]
Updated Swedish translation

11 years agoFix typo
Florian Müllner [Sat, 29 Nov 2014 18:05:11 +0000 (19:05 +0100)]
Fix typo

11 years agoUpdated Hungarian translation
Balázs Úr [Sun, 30 Nov 2014 15:31:24 +0000 (15:31 +0000)]
Updated Hungarian translation

11 years agotheme: Add high-contrast variant
Florian Müllner [Sat, 29 Nov 2014 15:04:14 +0000 (15:04 +0000)]
theme: Add high-contrast variant

For now, simply enforce symbolic icons.

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agoapp: Respect icon-style for faded icon texture
Florian Müllner [Sat, 29 Nov 2014 15:52:37 +0000 (15:52 +0000)]
app: Respect icon-style for faded icon texture

Just like regular application icons, the faded icon texture used
in the app menu should follow the theme's icon style setting.

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agoapp: Use StIcon as icon_texture when loading from a GIcon
Florian Müllner [Sat, 29 Nov 2014 00:25:04 +0000 (00:25 +0000)]
app: Use StIcon as icon_texture when loading from a GIcon

Themes - namely the HighContrast one - may now request symbolic
icons rather than fullcolor ones. In order to have recoloring
work as expected in that case, we will need a theme node to pick
up colors from - using an StIcon instead of manually loading a
texture from the cache gives us that for free, so do that.

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agost-theme-node: Add support for -st-icon-style property
Florian Müllner [Sat, 29 Nov 2014 14:53:36 +0000 (14:53 +0000)]
st-theme-node: Add support for -st-icon-style property

GTK+ added support for a -gtk-icon-style property in themes to
enforce a particular icon style. Do the same for shell themes
with an -st-icon-style property, with the same set of possible
values as the GTK+ variant:
  'requested' - use symbolic or fullcolor icon depending on the
                icon name (default)
  'regular'   - enforce fullcolor icons
  'symbolic'  - enforce symbolic icons

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agost-texture-cache: Remove load_gicon_with_colors()
Florian Müllner [Sat, 29 Nov 2014 14:43:50 +0000 (14:43 +0000)]
st-texture-cache: Remove load_gicon_with_colors()

The split between st_texture_cache_load_gicon() and load_gicon_with_colors()
no longer makes any sense, so just move the code into the public method.

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agomain: Add support for -high-contrast theme variants
Florian Müllner [Sat, 29 Nov 2014 13:37:19 +0000 (13:37 +0000)]
main: Add support for -high-contrast theme variants

While the default Shell style is fairly decent with regard to
accessibility requirements, having the ability to tweak certain
aspects where the regular style works less well is still useful.
For this purpose, try to load a -high-contrast theme variant of
the default stylesheet when a high-contrast theme is requested
(as determined by the GTK+ theme name).

https://bugzilla.gnome.org/show_bug.cgi?id=740447

11 years agooverview: Set _shown before calling _animateNotVisible
Jasper St. Pierre [Thu, 27 Nov 2014 02:05:36 +0000 (18:05 -0800)]
overview: Set _shown before calling _animateNotVisible

_hideDone checks _shown to determine if anything has shown the overview
while we hid it, and if so, shows the overview forward just in case.

In a local patch that called _hideDone immediately inside _hide for
testing, this broke. While we don't actually depend on this anywhere,
it doesn't hurt so that the next person to hack this up (perhaps me!)
doesn't get stuck debugging it for 20 minutes.

11 years agoUpdate .gitignore
Florian Müllner [Thu, 27 Nov 2014 20:00:36 +0000 (21:00 +0100)]
Update .gitignore

11 years agomain: Fix unsetting of custom stylesheets
Florian Müllner [Thu, 27 Nov 2014 14:24:49 +0000 (14:24 +0000)]
main: Fix unsetting of custom stylesheets

Since moving to a GFile based API in commit 642bf2b7782819,
setThemeStylesheet() no longer accepts %null to revert to
the default theme. We should have some way to revert to the
default and the least intrusive option is to return to the
old behavior, so do that.

11 years agomain: Update doc comment
Florian Müllner [Thu, 27 Nov 2014 14:43:03 +0000 (14:43 +0000)]
main: Update doc comment

11 years agoBump version to 3.15.2
Florian Müllner [Thu, 27 Nov 2014 13:56:26 +0000 (13:56 +0000)]
Bump version to 3.15.2

Update NEWS.