gtk4.git
11 years agowayland: Only bind to supported gtk-shell versions
Jonas Ådahl [Fri, 6 Mar 2015 02:51:47 +0000 (10:51 +0800)]
wayland: Only bind to supported gtk-shell versions

The gtk-shell Wayland protocol extension is not meant to be backward
compatible right now, so avoid binding to any version that is not the
one supported.

https://bugzilla.gnome.org/show_bug.cgi?id=745721

11 years agobuild: Fix generating introspection on MacOS
Emmanuele Bassi [Tue, 10 Mar 2015 22:12:32 +0000 (22:12 +0000)]
build: Fix generating introspection on MacOS

We need to filter out platform-specific sources from the files that we
use with the introspection scanner.

In this specific case I could have moved the gtkclipboard-quartz.c and
the gtkdnd-quartz.c files out of the $(gtk_base_c_sources) variable, but
doing a filter-out on the variable itself is more resilient in case we
eventually add files and we forget about the result.

11 years agoinspector: Add some margin to css editor
Matthias Clasen [Tue, 10 Mar 2015 18:07:27 +0000 (14:07 -0400)]
inspector: Add some margin to css editor

Requested in
https://bugzilla.gnome.org/show_bug.cgi?id=745918

11 years agowindow: reset both press/drag gestures when move/resize drag starts
Carlos Garnacho [Tue, 10 Mar 2015 15:38:17 +0000 (16:38 +0100)]
window: reset both press/drag gestures when move/resize drag starts

Now that this is split in two separate gestures, both must be reset
when the WM grabs the pointer. Also, do on resize drags like on move
drags, and claim the gesture before resetting, so the ownership is
properly transferred across any other widgets.

https://bugzilla.gnome.org/show_bug.cgi?id=745969

11 years agoUpdated Swedish translation
Anders Jonsson [Tue, 10 Mar 2015 12:28:37 +0000 (12:28 +0000)]
Updated Swedish translation

11 years agoUpdated Swedish translation
Anders Jonsson [Tue, 10 Mar 2015 12:27:18 +0000 (12:27 +0000)]
Updated Swedish translation

11 years agonotebook: Fix boolean check
Benjamin Otte [Tue, 10 Mar 2015 12:23:06 +0000 (13:23 +0100)]
notebook: Fix boolean check

Commit c352093bdeb2be5390607de13eaf7a21e93272ef swapped the condition
when turning it into a return_if_fail().

11 years agoentrycompletion: Ensure transient_for is set before showing window
Carlos Garnacho [Mon, 9 Mar 2015 14:15:23 +0000 (15:15 +0100)]
entrycompletion: Ensure transient_for is set before showing window

gtk_entry_set_completion() (and _gtk_entry_completion_connect underneath)
may happen when the entry has no toplevel yet. This window acts as the
"parent" for subsurface windows on wayland, and is indispensable to create
its backing wl_subsurface, so this resulted in non-visible popups there.
Fix this by ensuring transient_for is set before showing the popup.

https://bugzilla.gnome.org/show_bug.cgi?id=745463

11 years agogtknotebook: Fix potential use of uninitialised variables
Philip Withnall [Wed, 20 Nov 2013 17:32:10 +0000 (17:32 +0000)]
gtknotebook: Fix potential use of uninitialised variables

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkprintunixdialog: Fix a potential g_object_unref(NULL) call
Philip Withnall [Tue, 26 Nov 2013 15:08:08 +0000 (15:08 +0000)]
gtkprintunixdialog: Fix a potential g_object_unref(NULL) call

The code above checks whether (printer == NULL), so we’d better do so
here as well.

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agoRevert "csd: Drop the GTK_FRAME_EXTENTS requirement"
Matthias Clasen [Tue, 10 Mar 2015 00:48:47 +0000 (20:48 -0400)]
Revert "csd: Drop the GTK_FRAME_EXTENTS requirement"

This reverts commit fb9a6bb6d8d6b60b25c9b9853decbcea938f7863.

In a recent test, I've found that Xfce and Mate now support
this, so they will not be affected by this requirement. And
adding the check back will solve the 'client-side shadow'
problem in KDE.

11 years agostylecontext: Store the right style in the cache
Benjamin Otte [Mon, 9 Mar 2015 22:53:43 +0000 (23:53 +0100)]
stylecontext: Store the right style in the cache

Store the newly computed style, not the old and wrong one.

Fixes HighContrast menuitem color sometimes being black-on-black for
selected items.

11 years agogtkclipboard: Fix a potential g_signal_handler_disconnect(NULL) call
Philip Withnall [Thu, 28 Nov 2013 20:43:16 +0000 (20:43 +0000)]
gtkclipboard: Fix a potential g_signal_handler_disconnect(NULL) call

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkprintcontext: Fix several potential g_object_[un]ref(NULL) calls
Philip Withnall [Tue, 26 Nov 2013 15:04:45 +0000 (15:04 +0000)]
gtkprintcontext: Fix several potential g_object_[un]ref(NULL) calls

The page_setup of a GtkPrintContext or GtkPrintUnixDialog is nullable,
so all reference count changes to it have to be guarded against NULL
values.

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtktreeselection: Fix a potential use of uninitialised variables
Philip Withnall [Wed, 20 Nov 2013 17:39:02 +0000 (17:39 +0000)]
gtktreeselection: Fix a potential use of uninitialised variables

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtktreeselection: Fix potential NULL pointer dereferences
Philip Withnall [Wed, 20 Nov 2013 17:38:01 +0000 (17:38 +0000)]
gtktreeselection: Fix potential NULL pointer dereferences

_gtk_rbtree_first() can potentially return NULL if the RB tree is empty,
which would result in NULL pointer dereferences in the GtkTreeSelection
code. Gracefully handle them.

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtktextview: Fix a definite use of an uninitialised variable
Philip Withnall [Wed, 20 Nov 2013 17:33:59 +0000 (17:33 +0000)]
gtktextview: Fix a definite use of an uninitialised variable

This seems to have been a typo in the original code, and allowed access
to virtual_cursor_y when it was uninitialised.

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtknotebook: Remove dead variable assignments
Philip Withnall [Wed, 20 Nov 2013 17:32:47 +0000 (17:32 +0000)]
gtknotebook: Remove dead variable assignments

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtknotebook: Fix a potential NULL pointer dereference
Philip Withnall [Wed, 20 Nov 2013 17:32:32 +0000 (17:32 +0000)]
gtknotebook: Fix a potential NULL pointer dereference

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkentry: Return early from gtk_entry_clear() if no icon info exists
Philip Withnall [Wed, 20 Nov 2013 17:30:22 +0000 (17:30 +0000)]
gtkentry: Return early from gtk_entry_clear() if no icon info exists

This helps scan-build avoid some false positive potential NULL pointer
dereference warnings.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkentry: Remove dead assignments
Philip Withnall [Wed, 20 Nov 2013 17:30:07 +0000 (17:30 +0000)]
gtkentry: Remove dead assignments

Found by scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkentry: Call get_*_size() vfuncs unconditionally
Philip Withnall [Mon, 2 Dec 2013 11:03:28 +0000 (11:03 +0000)]
gtkentry: Call get_*_size() vfuncs unconditionally

These vfuncs cannot be NULL: implementations are provided by GtkEntry,
and subclasses should not set them to NULL. Instead of conditionalising
the calls to the vfuncs, assert that they’re set and call them
unconditionally.

This prevents the possibility of a subclass setting the vfunc to NULL
and then a gtk_entry_get_*_size() call returning undefined values in its
out variables.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogtkentry: Fix potential use of uninitialised variables
Philip Withnall [Mon, 2 Dec 2013 11:03:07 +0000 (11:03 +0000)]
gtkentry: Fix potential use of uninitialised variables

Found with scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agogdkwindow: Fix potential NULL pointer dereference
Philip Withnall [Wed, 20 Nov 2013 17:18:16 +0000 (17:18 +0000)]
gdkwindow: Fix potential NULL pointer dereference

gdk_window_ensure_native() can end up with a NULL parent pointer, which
it passes to find_native_parent_above()…but that expects a non-NULL
parent.

Found with scan-build.

https://bugzilla.gnome.org/show_bug.cgi?id=712760

11 years agoPrevent empty password request popup for kerberos
Andre Heinecke [Thu, 5 Feb 2015 10:14:01 +0000 (11:14 +0100)]
Prevent empty password request popup for kerberos

    If auth_info_required is only set to "negotiate" the
    request password dialog is shown anyway for each print
    job without any input fields. As the request_password
    dialog only handles username, password and domain.

    https://bugzilla.gnome.org/show_bug.cgi?id=669008

11 years agoUpdate hello-world.c to GtkApplication, document
Bastian Ilsø [Tue, 10 Feb 2015 15:55:30 +0000 (16:55 +0100)]
Update hello-world.c to GtkApplication, document

11 years agoPrint dialog: Fix printer icons in HighContrast
Matthias Clasen [Mon, 9 Mar 2015 02:05:37 +0000 (22:05 -0400)]
Print dialog: Fix printer icons in HighContrast

We need to make the cell renderer follow the state,
for recoloring to work.

11 years agoHighContrast: Some formatting fixes
Matthias Clasen [Mon, 9 Mar 2015 01:32:38 +0000 (21:32 -0400)]
HighContrast: Some formatting fixes

11 years agoAdd a missing word in the docs
Matthias Clasen [Sun, 8 Mar 2015 22:14:18 +0000 (18:14 -0400)]
Add a missing word in the docs

11 years agoUpdated Lithuanian translation
Aurimas Černius [Sun, 8 Mar 2015 18:22:54 +0000 (20:22 +0200)]
Updated Lithuanian translation

11 years agodocs: Reword the deprecation notice for gtk_widget_override_*
Emmanuele Bassi [Sun, 8 Mar 2015 14:39:18 +0000 (14:39 +0000)]
docs: Reword the deprecation notice for gtk_widget_override_*

They are a bit terse, and they should point to the preferred way of
overriding the theme, i.e. using a GtkStyleProvider for your
application, and custom style classes.

11 years agodocs: Clarify replacement for gtk_style_context_get_background_color()
Emmanuele Bassi [Sun, 8 Mar 2015 14:29:04 +0000 (14:29 +0000)]
docs: Clarify replacement for gtk_style_context_get_background_color()

It seems odd that a getter function for the style context state is
replaced by a rendering call, but it makes sense in context.

11 years agoAdd a test for builder-constructed file filters
Matthias Clasen [Sun, 8 Mar 2015 13:11:51 +0000 (09:11 -0400)]
Add a test for builder-constructed file filters

11 years agoGtkFileFilter: Set name from builder id
Matthias Clasen [Sun, 8 Mar 2015 13:10:58 +0000 (09:10 -0400)]
GtkFileFilter: Set name from builder id

Suggested in https://bugzilla.gnome.org/show_bug.cgi?id=577862
as a way to make file filters constructed with GtkBuilder more
useful.

11 years agoUpdated Kazakh translation
Baurzhan Muftakhidinov [Sun, 8 Mar 2015 11:17:09 +0000 (11:17 +0000)]
Updated Kazakh translation

11 years agodocs: fix reference to GTK+ 2 index of deprecated symbols
Frédéric Péters [Sun, 8 Mar 2015 08:43:04 +0000 (09:43 +0100)]
docs: fix reference to GTK+ 2 index of deprecated symbols

11 years agoFix undefined reference to get_xkb()
Eric Le Bihan [Thu, 23 Oct 2014 11:00:54 +0000 (13:00 +0200)]
Fix undefined reference to get_xkb()

When configuring Gtk+ with --disable-xkb, the build fails because of an
undefined reference to get_xkb().

This patch fixes this issue.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
https://bugzilla.gnome.org/show_bug.cgi?id=739070

11 years agoModernize the xReply check
Matthias Clasen [Sat, 7 Mar 2015 23:25:57 +0000 (18:25 -0500)]
Modernize the xReply check

Using AC_CHECK_TYPE instead of AC_TRY_COMPILE avoids compiler
warnings from -Werror.

https://bugzilla.gnome.org/show_bug.cgi?id=535929

11 years agoAvoid calling g_object_ref on NULL
Matthias Clasen [Sat, 7 Mar 2015 22:48:11 +0000 (17:48 -0500)]
Avoid calling g_object_ref on NULL

gtk_widget_set_visual() allows NULL for the visual, so be
careful ot to ref it unconditionally.

11 years agoscrolled window: Fix size requisition with overlay scrolling
Matthias Clasen [Sat, 7 Mar 2015 19:18:13 +0000 (14:18 -0500)]
scrolled window: Fix size requisition with overlay scrolling

When the scrollbars are overlayed, the size requisition of the
scrolled window should not depend on whether the scrollbars are
visible or not. This was not quite the case, because we forgot
one case where scrollbar size was still added to the requisition.

11 years agoAdd an example for accelerators with manual image menuitems
Matthias Clasen [Sat, 7 Mar 2015 01:33:38 +0000 (20:33 -0500)]
Add an example for accelerators with manual image menuitems

This is somewhat tricky to work out, so put some example
code here for future reference. The tricky part is that
GtkAccelLabel tries to be smart about hiding the accel
if there's not enough space, so we have to make sure to
pack the label with expand=TRUE and set align=GTK_ALIGN_FILL,
or things won't work.

11 years agotest/gtkgears.c: Fix on C89 Compilers
Chun-wei Fan [Fri, 6 Mar 2015 03:01:30 +0000 (11:01 +0800)]
test/gtkgears.c: Fix on C89 Compilers

Make sure that variables are declared at the top of the block.

Break up one of the sincos() calls into individual calls to sin() and cos()
so that we do not have to complicate the initialization of the following
GLfloat array.

11 years agoUpdated Korean translation
Changwoo Ryu [Fri, 6 Mar 2015 13:14:03 +0000 (22:14 +0900)]
Updated Korean translation

11 years agogl: Warn the user if they request a GL context version less than 3.2
Niels Nesse [Wed, 11 Feb 2015 04:35:01 +0000 (20:35 -0800)]
gl: Warn the user if they request a GL context version less than 3.2

If the user requests a version less than 3.2 the version is forced to 3.2.
Previous checking code have an inconsistent behavior depending on which
minor version number was specified. This is avoided now by temporarily
converting the major/minor pair into a single integer.

https://bugzilla.gnome.org/show_bug.cgi?id=744288

11 years agoUpdate config.h.win32.in for sincos()
Chun-wei Fan [Fri, 6 Mar 2015 02:49:25 +0000 (10:49 +0800)]
Update config.h.win32.in for sincos()

There was an added check in configure.ac for sincos() for gtkgears.c, where
that function is actually a GCC- and glibc-specific extension, which is
not available on Windows.  Reflect on this in config.h.win32.in for the
Visual Studio/Windows builds.

11 years agoMSVC Builds: "Install" .pdb Files
Chun-wei Fan [Fri, 6 Mar 2015 02:40:46 +0000 (10:40 +0800)]
MSVC Builds: "Install" .pdb Files

"Install" the .pdb files that were generated along with the build, to make
it easier for people to develop and debug GTK+, especially when used as a
stack, and as the .pdb files are already generated with all builds.

Also make the copy of the .dll, .lib and .exe files more selective so that
we do not accidently copy files that are not meant to be "installed", or
end up making extra copies of plugin DLLs in the wrong places, when the
projects here are included as a part of a grand solution that is used to
build the entire GTK+ stack, for example.

11 years agoMSVC 2010+ Builds: Fix .pdb Generation
Chun-wei Fan [Fri, 6 Mar 2015 02:25:18 +0000 (10:25 +0800)]
MSVC 2010+ Builds: Fix .pdb Generation

For Visual Studio 2010 and later, the .pdb filename needs to be explicitly
specified so that it will match the filename of the target, if the project
name differs from the filename of the target.  Make sure that is the case
for the 3 DLLs that we build for Visual Studio 2010 and later.

11 years agoUpdated Polish translation
Piotr Drąg [Thu, 5 Mar 2015 22:47:51 +0000 (23:47 +0100)]
Updated Polish translation

11 years agoImprove some GtkStock deprecation notes
Matthias Clasen [Thu, 5 Mar 2015 21:06:22 +0000 (16:06 -0500)]
Improve some GtkStock deprecation notes

Point at edit-delete and edit-paste as available icons.

11 years agoUpdated Slovak translation
Dušan Kazik [Thu, 5 Mar 2015 13:17:37 +0000 (13:17 +0000)]
Updated Slovak translation

11 years agoUpdated Slovak translation
Dušan Kazik [Thu, 5 Mar 2015 12:14:18 +0000 (12:14 +0000)]
Updated Slovak translation

11 years agowindow: Add comment documenting the target widget checks during window drag
Carlos Garnacho [Tue, 3 Mar 2015 21:17:58 +0000 (22:17 +0100)]
window: Add comment documenting the target widget checks during window drag

Was suggested during review of
https://bugzilla.gnome.org/show_bug.cgi?id=745562#c2

11 years agoUpdated French translation
Claude Paroz [Thu, 5 Mar 2015 07:49:25 +0000 (08:49 +0100)]
Updated French translation

11 years agoHighContrast: Make sure selections are visible in infobars
Matthias Clasen [Thu, 5 Mar 2015 02:56:02 +0000 (21:56 -0500)]
HighContrast: Make sure selections are visible in infobars

Pointed out in https://bugzilla.gnome.org/show_bug.cgi?id=745622

11 years agoAdwaita: Make sure selections are visible in infobars
Matthias Clasen [Thu, 5 Mar 2015 02:55:11 +0000 (21:55 -0500)]
Adwaita: Make sure selections are visible in infobars

Pointed out in https://bugzilla.gnome.org/show_bug.cgi?id=745622

11 years agowidget: Improve hack to ignore drags from widgets using motion events
Carlos Garnacho [Tue, 3 Mar 2015 21:17:58 +0000 (22:17 +0100)]
widget: Improve hack to ignore drags from widgets using motion events

Postpone until the last moment whether the target widget still
potentially uses updates from this sequence, or window dragging
actually applies because all gestures on the target went to denied
state.

This fixes window dragging on empty space in a headerbar that is
contained in a paned (as in e.g. gedit).

https://bugzilla.gnome.org/show_bug.cgi?id=745562

11 years agoEnsure print-related types are registered
Matthias Clasen [Thu, 5 Mar 2015 01:35:09 +0000 (20:35 -0500)]
Ensure print-related types are registered

These types are not covered by g_test_register_all_types,
and having the types registered when a print dialog is used
helps in cases where GTK+ is dlopened.

https://bugzilla.gnome.org/show_bug.cgi?id=745065

11 years agoplaces sidebar: compare bookmarks by index
Georges Basile Stavracas Neto [Wed, 4 Mar 2015 20:12:40 +0000 (17:12 -0300)]
places sidebar: compare bookmarks by index

GtkPlacesSidebar applies a sorting function on
the tree model that does not consider the case
of bookmarks, which are sorted by their indexes.

By adding the bookmarks corner case and comparing
then by their indexes, GtkPlacesSidebar can sort
the bookmarks properly in the order they're saved.

https://bugzilla.gnome.org/show_bug.cgi?id=744589

11 years agoCheck for existence of sincos
Matthias Clasen [Wed, 4 Mar 2015 11:30:13 +0000 (06:30 -0500)]
Check for existence of sincos

It apparently doesn't exist everywhere.

https://bugzilla.gnome.org/show_bug.cgi?id=745578

11 years agofile chooser: Really stop typeahead search popups
Matthias Clasen [Tue, 3 Mar 2015 23:02:09 +0000 (18:02 -0500)]
file chooser: Really stop typeahead search popups

GtkTreeView is a bad widget and constantly resets the search-column
behind our back. We need to re-unset it every time a model is set :-(

11 years agoUpdated Galician translations
Fran Dieguez [Tue, 3 Mar 2015 22:44:15 +0000 (23:44 +0100)]
Updated Galician translations

11 years agomake-pot: Document variables
Matthias Clasen [Tue, 3 Mar 2015 20:00:06 +0000 (15:00 -0500)]
make-pot: Document variables

Damn-lies will have to use one of these to override the name
of the generated file. Lets document them all, while I still
remember this.

11 years agofilechooser: Only do search_start_query() while we are in search mode
Carlos Garnacho [Tue, 3 Mar 2015 16:43:14 +0000 (17:43 +0100)]
filechooser: Only do search_start_query() while we are in search mode

This signal can be emitted by GtkSearchEntry after search has been
cancelled, and other operation mode is set. It doesn't make sense to
populate the search model in that state anymore, so just avoid doing it.

https://bugzilla.gnome.org/show_bug.cgi?id=745479

11 years agosearchenginetracker: Keep a reference on the search engine while querying
Carlos Garnacho [Tue, 3 Mar 2015 12:36:07 +0000 (13:36 +0100)]
searchenginetracker: Keep a reference on the search engine while querying

The object might be destroyed when mid operation, causing crashes as the
query callback still expects the object pointer to be valid. Also, remove
the gdk_threads_enter/leave pairs, the callback will be executed on the
caller (UI) thread, so this is not necessary.

https://bugzilla.gnome.org/show_bug.cgi?id=745479

11 years agofilechooser: Disconnect signal handlers from search engine before destroying
Carlos Garnacho [Tue, 3 Mar 2015 12:34:03 +0000 (13:34 +0100)]
filechooser: Disconnect signal handlers from search engine before destroying

The search engine might stay alive longer due to extra temporary refs, so
the signal handlers should be removed for the filechooser to ignore these
properly.

https://bugzilla.gnome.org/show_bug.cgi?id=745479

11 years agoMake a pot generating command available
Matthias Clasen [Tue, 3 Mar 2015 12:11:41 +0000 (07:11 -0500)]
Make a pot generating command available

Translators don't want to run autogen before generating pot,
so give them a script.

11 years agoMSVC 2008 Builds: Speed Up Release Builds
Chun-wei Fan [Tue, 3 Mar 2015 10:40:32 +0000 (18:40 +0800)]
MSVC 2008 Builds: Speed Up Release Builds

Use the /MP compiler option, where the build time for release builds can
be cut down by quite a bit.  This will however cause a brief warning with
debug builds due to the use of /Gm, but the code will otherwise build
normally.  Unlike the Visual Studio 2010+ builds, we can't use /d2Zi+ as
Visual Studio 2008 does not support that, so we can't get a better
debugging experience for release builds here.

11 years agoFix "MSVC Builds: Update icon-browser Projects"
Chun-wei Fan [Tue, 3 Mar 2015 10:36:00 +0000 (18:36 +0800)]
Fix "MSVC Builds: Update icon-browser Projects"

The Visual Studio 2008 Project was not updated correctly, so fix that.

11 years agotooltip: remove unused variable and related g_object_set
Stefan Sauer [Tue, 3 Mar 2015 06:49:38 +0000 (07:49 +0100)]
tooltip: remove unused variable and related g_object_set

We never use the value of has_tooltip. This is a left-over from the refactoring
of gtk_tooltip_run_requery().

11 years agoMSVC Builds: Update icon-browser Projects
Chun-wei Fan [Tue, 3 Mar 2015 06:08:51 +0000 (14:08 +0800)]
MSVC Builds: Update icon-browser Projects

There is a new source file that needs to be built, so fix the projects...

11 years agoMSVC Builds: Improve Build Speed and Debugging
Chun-wei Fan [Tue, 3 Mar 2015 06:07:26 +0000 (14:07 +0800)]
MSVC Builds: Improve Build Speed and Debugging

Use Multiprocessor compilation which can cut down build times by quite a
bit and use the /d2Zi+ flag to have better debugging info being logged to
the .pdb for release builds.

These are only applicable for Visual Studio 2010/2012 and later.

11 years agogtkswitch.c: Fix Build on C89 Compilers
Chun-wei Fan [Tue, 3 Mar 2015 04:14:24 +0000 (12:14 +0800)]
gtkswitch.c: Fix Build on C89 Compilers

This file now uses round(), which was not available until C99, so include
fallback-c89.c instead of math.h, which includes math.h and does a fallback
implementation of round().

11 years ago3.15.10
Matthias Clasen [Tue, 3 Mar 2015 00:46:43 +0000 (19:46 -0500)]
3.15.10

11 years agoAvoid stray output from configure
Matthias Clasen [Tue, 3 Mar 2015 01:57:29 +0000 (20:57 -0500)]
Avoid stray output from configure

The new xgettext check was leaving a messages.po file behind.

11 years agoFix 7c4bf742e82d812ecc5b0c3280db86d2689eb093
Matthias Clasen [Tue, 3 Mar 2015 01:16:25 +0000 (20:16 -0500)]
Fix 7c4bf742e82d812ecc5b0c3280db86d2689eb093

The objects-finalize test constructs gestures with NULL as
a widget. The recent addition of weak references was not
ready for that.

11 years agogears: Update to slightly more modern OpenGL
Emmanuele Bassi [Tue, 3 Mar 2015 00:06:35 +0000 (00:06 +0000)]
gears: Update to slightly more modern OpenGL

Instead of using glxgears, which still uses OpenGL 2.1 and the fixed
pipeline, we use a slightly modified es2gears, OpenGL 3.2, and the
programmable pipeline.

11 years agofile chooser: Avoid animated scrolling for editing
Matthias Clasen [Mon, 2 Mar 2015 21:22:56 +0000 (16:22 -0500)]
file chooser: Avoid animated scrolling for editing

Avoid scrolling animation when scrolling to where we want
to insert the new folder, since adding the entry doesn't
do the right thing if the treeview is still scrolling.

An alternative would be to wait for the animated scrolling
to reach its target before starting the editing, but this
is easier.

https://bugzilla.gnome.org/show_bug.cgi?id=729366

11 years agofile chooser: Avoid triggering assertions
Matthias Clasen [Mon, 2 Mar 2015 20:53:35 +0000 (15:53 -0500)]
file chooser: Avoid triggering assertions

cairo surfaces are boxed types, so don't try to set them
with g_value_set_object().

11 years agofile chooser: Avoid warnings from the location column
Matthias Clasen [Mon, 2 Mar 2015 20:38:29 +0000 (15:38 -0500)]
file chooser: Avoid warnings from the location column

Avoid criticals that would come out of this code if file is / or NULL.

11 years agoUpdated French translation
Claude Paroz [Mon, 2 Mar 2015 18:32:07 +0000 (18:32 +0000)]
Updated French translation

11 years agox11: Store last axes from device
Carlos Garnacho [Mon, 2 Mar 2015 17:19:16 +0000 (18:19 +0100)]
x11: Store last axes from device

And use these for the missing axes if the valuator mask is incomplete.
This used to work fine on tablets because the Wacom driver ensures all
valuators are sent, which is not true if using the evdev driver.

https://bugzilla.gnome.org/show_bug.cgi?id=703610

11 years agoscrolledwindow: Cancel kinetic/overshoot animation on captured scroll events
Carlos Garnacho [Mon, 2 Mar 2015 16:01:21 +0000 (17:01 +0100)]
scrolledwindow: Cancel kinetic/overshoot animation on captured scroll events

This ensures the animation is cancelled if the child widget happens to
GDK_EVENT_STOP scroll events.

https://bugzilla.gnome.org/show_bug.cgi?id=745315

11 years agoUpdates
Matthias Clasen [Mon, 2 Mar 2015 16:11:24 +0000 (11:11 -0500)]
Updates

11 years agoscrolledwindow: Add motion mask to the widget window
Carlos Garnacho [Mon, 2 Mar 2015 14:31:11 +0000 (15:31 +0100)]
scrolledwindow: Add motion mask to the widget window

This is not specified specifically by the attached controllers, so let
the scrolledwindow set the mask, as motion events with no buttons pressed
are interesting to it.

https://bugzilla.gnome.org/show_bug.cgi?id=745344

11 years agoUpdated Hebrew translation
Yosef Or Boczko [Mon, 2 Mar 2015 13:39:40 +0000 (15:39 +0200)]
Updated Hebrew translation

11 years agowayland: Delay wl_subsurface interface creation until the window is shown
Carlos Garnacho [Wed, 25 Feb 2015 14:59:26 +0000 (15:59 +0100)]
wayland: Delay wl_subsurface interface creation until the window is shown

It isn't really necessary anytime before, so just make sure it's there only
when the window is visible.

https://bugzilla.gnome.org/show_bug.cgi?id=743427

11 years agogtkwindow: Do not show subsurface windows for hidden popover widgets
Carlos Garnacho [Wed, 25 Feb 2015 14:52:49 +0000 (15:52 +0100)]
gtkwindow: Do not show subsurface windows for hidden popover widgets

This is really just necessary when we have a visible widget, otherwise
the window would temporarily linger with odd size and surface.

https://bugzilla.gnome.org/show_bug.cgi?id=743427

11 years agowayland: Thaw the clock if we hide a GdkWindow mid-frame
Carlos Garnacho [Wed, 25 Feb 2015 13:46:54 +0000 (14:46 +0100)]
wayland: Thaw the clock if we hide a GdkWindow mid-frame

When a window is hidden, its surface and all its roles are destroyed,
if this happens when we already issued a wl_surface_commit and are
awaiting for a frame callback, the clock will remain frozen for the
next time the window is shown.

To avoid this, keep track of the wl_surface_frame() calls issued,
and ensure the clock is thawed after hiding. If we happen to receive
the frame callback, it is just ignored.

https://bugzilla.gnome.org/show_bug.cgi?id=743427

11 years agogtkwindow: Move window dragging to a standalone drag gesture
Carlos Garnacho [Wed, 25 Feb 2015 19:34:12 +0000 (20:34 +0100)]
gtkwindow: Move window dragging to a standalone drag gesture

The gesture is hooked to the capture phase, so it works for buttons in
header bars and whatnot. In order to be friendly to the widget it is
capturing events from, an ugly hack is in place to avoid capturing
events when the target widget has a gesture that would consume motion
events.

11 years agogtkmenubutton: Popup menu/popover on GtkButton:clicked
Carlos Garnacho [Wed, 25 Feb 2015 19:31:09 +0000 (20:31 +0100)]
gtkmenubutton: Popup menu/popover on GtkButton:clicked

This happens on button release, which is more convenient if the gesture
can be consumed by something else (eg. window dragging), and already behaves
correctly wrt cancelled gestures, broken grabs, etc.

This also allows us to unify pointer and keyboard behavior, popping up the
menu widget in a single place.

11 years agogesture: On cancel, remove touchpoint before checking the recognized state
Carlos Garnacho [Wed, 25 Feb 2015 19:30:11 +0000 (20:30 +0100)]
gesture: On cancel, remove touchpoint before checking the recognized state

Otherwise the touch being removed doesn't account in recognition.

11 years agobutton: Ensure the button is "left" when a gesture is cancelled
Carlos Garnacho [Wed, 25 Feb 2015 19:29:22 +0000 (20:29 +0100)]
button: Ensure the button is "left" when a gesture is cancelled

This prevents from stale widget states if the gesture happens to be
cancelled mid-press.

11 years agoeventcontroller: Chain up on constructed
Debarshi Ray [Sun, 1 Mar 2015 12:35:02 +0000 (13:35 +0100)]
eventcontroller: Chain up on constructed

https://bugzilla.gnome.org/show_bug.cgi?id=745225

11 years agoeventcontroller, widget: Don't crash if destroyed before the other
Debarshi Ray [Sun, 1 Mar 2015 12:28:21 +0000 (13:28 +0100)]
eventcontroller, widget: Don't crash if destroyed before the other

There are two scenarios. A widget sub-class owns a GtkEventController
and passes itself to it, or a controller owned by something else is
passed a widget.

In the second case, if the widget is destroyed before the controller,
we will have a crash when destructing the controller because we will
be accessing invalid memory. Adding a weak reference on the widget
addresses that problem.

This leads to a crash in the first case. When the widget is getting
destroyed, it will drop the reference to its own controller. The
controller will skip touching the widget because the weak reference
would have turned it to NULL. However, when the widget sub-class chains
up to GtkWidget it will try to free all the controllers in its list.
Unfortunately, all these controllers have already been destroyed. So
we need to guard against this too.

https://bugzilla.gnome.org/show_bug.cgi?id=745225

11 years agogtk_label_set_yalign(): Fix xalign/yalign typo in parameter name.
Murray Cumming [Mon, 2 Mar 2015 09:58:23 +0000 (10:58 +0100)]
gtk_label_set_yalign(): Fix xalign/yalign typo in parameter name.

11 years agoprinting: Set printer state correctly
Marek Kasik [Fri, 27 Feb 2015 15:48:40 +0000 (16:48 +0100)]
printing: Set printer state correctly

Consider NULL printer state message as empty too.
Printer state IPP_PRINTER_STOPPED can be considered as paused state.

https://bugzilla.gnome.org/show_bug.cgi?id=743323

11 years agogtk_misc_set_alignment(): Improve the deprecation documenation.
Murray Cumming [Mon, 2 Mar 2015 08:45:57 +0000 (09:45 +0100)]
gtk_misc_set_alignment(): Improve the deprecation documenation.

To link to GtkWidget.halign/valign and to mention
GtkLabel.xalign/yalign.

11 years agogdkglcontext-win32: fix variable names in GDK_NOTE
Ryan Lortie [Mon, 2 Mar 2015 02:35:53 +0000 (21:35 -0500)]
gdkglcontext-win32: fix variable names in GDK_NOTE

These probably weren't caught because GDK_NOTE was disabled when the
change was tested.

https://bugzilla.gnome.org/show_bug.cgi?id=745400

11 years agogdkglcontext-win32: remove unused variables
Ryan Lortie [Mon, 2 Mar 2015 02:35:34 +0000 (21:35 -0500)]
gdkglcontext-win32: remove unused variables

https://bugzilla.gnome.org/show_bug.cgi?id=745400

11 years agoUpdated Czech translation
Marek Černocký [Sun, 1 Mar 2015 17:54:56 +0000 (18:54 +0100)]
Updated Czech translation