gtk4.git
9 years agowayland: Don't traverse transient-ofs when faking root coordinate space
Jonas Ådahl [Thu, 4 Aug 2016 06:07:47 +0000 (14:07 +0800)]
wayland: Don't traverse transient-ofs when faking root coordinate space

The position of each transient-of will be in fake-root coordinate
space; thus we should not accumulate all the positions making it an
offset; each window is already in fake root coordinate space.

https://bugzilla.gnome.org/show_bug.cgi?id=769402

9 years agowayland: Use effective toplevel as popup parent
Jonas Ådahl [Thu, 4 Aug 2016 06:05:06 +0000 (14:05 +0800)]
wayland: Use effective toplevel as popup parent

When using the set transient-for as a popup parent, fetch the effective
toplevel instead, otherwise we will position against the wrong
coordinate.

https://bugzilla.gnome.org/show_bug.cgi?id=769402

9 years agoUse gtk_popover_popdown/popup where appropriate
Timm Bäder [Wed, 10 Aug 2016 17:18:27 +0000 (19:18 +0200)]
Use gtk_popover_popdown/popup where appropriate

https://bugzilla.gnome.org/show_bug.cgi?id=769706

9 years agoGtkPopover: Deprecate transitions-enabled
Timm Bäder [Wed, 10 Aug 2016 16:46:35 +0000 (18:46 +0200)]
GtkPopover: Deprecate transitions-enabled

The effect of transitions-enabled=true can now be
achieved using gtk_popover_popup/popdown and the effect
of transitions-enabled=false can be achieved using
gtk_widget_show/hide.

https://bugzilla.gnome.org/show_bug.cgi?id=769706

9 years agoGtkPopover: Add gtk_popover_popdown/popup
Timm Bäder [Wed, 10 Aug 2016 16:28:14 +0000 (18:28 +0200)]
GtkPopover: Add gtk_popover_popdown/popup

Since not chaining up in gtk_widget_show/gtk_widget_hide is not allowed,
we can't just implicitly delay the hiding in GtkPopover's hide
implementation. Fix this by introducing gtk_popover_popup() and
gtk_popover_popdown() to show or hide a popover with transition and
revert GtkPopover's show/hide implementation to apply their effect
without the transition.

https://bugzilla.gnome.org/show_bug.cgi?id=769706

9 years agoUpdated Indonesian translation
Andika Triwidada [Mon, 15 Aug 2016 06:49:41 +0000 (06:49 +0000)]
Updated Indonesian translation

9 years agoUpdated Indonesian translation
Andika Triwidada [Mon, 15 Aug 2016 06:30:29 +0000 (06:30 +0000)]
Updated Indonesian translation

9 years agomir: group DND, tooltips, and notifications with menu-type windows
William Hua [Fri, 12 Aug 2016 15:35:38 +0000 (11:35 -0400)]
mir: group DND, tooltips, and notifications with menu-type windows

9 years agoRemove outdated comments
Andreas Pokorny [Tue, 19 Jul 2016 08:46:06 +0000 (10:46 +0200)]
Remove outdated comments

https://bugzilla.gnome.org/show_bug.cgi?id=768138

9 years agoUse Menus to implement tooltips
Andreas Pokorny [Tue, 19 Jul 2016 08:45:33 +0000 (10:45 +0200)]
Use Menus to implement tooltips

The order in which tooltips are created, drawn, shown and then positioned,
always requires repositioning the surface. The tooltip window type only has
limited capability to do so. An alternative could be to use bufferstreams.

https://bugzilla.gnome.org/show_bug.cgi?id=768138

9 years agoFix execution of dialog
Andreas Pokorny [Tue, 19 Jul 2016 08:36:45 +0000 (10:36 +0200)]
Fix execution of dialog

When a dialog is created, the mir event source is already executed on the
call stack. So without the recurse flag it will not be run in the main loop
used for the dialog.

https://bugzilla.gnome.org/show_bug.cgi?id=768138

9 years agofile chooser: align date formatting with nautilus
Matthias Clasen [Thu, 11 Aug 2016 16:22:01 +0000 (12:22 -0400)]
file chooser: align date formatting with nautilus

We were treating 'yesterday' not quite the same way.

https://bugzilla.gnome.org/show_bug.cgi?id=769568

9 years agodocs: Fix a dangling link
Matthias Clasen [Thu, 11 Aug 2016 13:28:58 +0000 (09:28 -0400)]
docs: Fix a dangling link

https://bugzilla.gnome.org/show_bug.cgi?id=769601

9 years agoReally fix unlinkage for shm_open
Matthias Clasen [Thu, 11 Aug 2016 11:07:05 +0000 (07:07 -0400)]
Really fix unlinkage for shm_open

Reported in https://bugzilla.gnome.org/show_bug.cgi?id=769603

9 years agoOnly apply type hint if it would map to a different mir surface type
Andreas Pokorny [Tue, 12 Jul 2016 14:25:01 +0000 (16:25 +0200)]
Only apply type hint if it would map to a different mir surface type

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoApply geometry hints to mir surface
Andreas Pokorny [Tue, 12 Jul 2016 14:11:46 +0000 (16:11 +0200)]
Apply geometry hints to mir surface

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoApply and forward title changes of gdk windows
Andreas Pokorny [Tue, 12 Jul 2016 10:44:34 +0000 (12:44 +0200)]
Apply and forward title changes of gdk windows

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoOnly update surface spec when there is no spec change pending
Andreas Pokorny [Tue, 19 Jul 2016 12:53:22 +0000 (14:53 +0200)]
Only update surface spec when there is no spec change pending

9 years agoRework window construction - only recreate surface when necessary
Andreas Pokorny [Tue, 19 Jul 2016 12:48:59 +0000 (14:48 +0200)]
Rework window construction - only recreate surface when necessary

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoForward repeated key presses as further down keys
Andreas Pokorny [Tue, 19 Jul 2016 12:44:42 +0000 (14:44 +0200)]
Forward repeated key presses as further down keys

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoFix warning on newer version so mir 0.22 and newer
Andreas Pokorny [Tue, 28 Jun 2016 14:37:06 +0000 (16:37 +0200)]
Fix warning on newer version so mir 0.22 and newer

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agoFix gcc warning on potentially uninitialized gdk_event.
Andreas Pokorny [Tue, 28 Jun 2016 14:35:36 +0000 (16:35 +0200)]
Fix gcc warning on potentially uninitialized gdk_event.

9 years agoUse the surface output event to keep track of the suggested scale value
Andreas Pokorny [Tue, 19 Jul 2016 12:56:34 +0000 (14:56 +0200)]
Use the surface output event to keep track of the suggested scale value

The event will be sent when the surface becomes visible on an output. With this change the GdkMirWindowImpl keeps track of the scale value and sends a configure event on change.

Signed-off-by: Andreas Pokorny <andreas.pokorny@canonical.com>
9 years agowidget-factory: Don't initially show popover
Timm Bäder [Wed, 10 Aug 2016 13:23:13 +0000 (15:23 +0200)]
widget-factory: Don't initially show popover

It'll  be set as the popover of a GtkMenuButton later anyway which will
hide it again.

9 years agopopover: Make some public API NULL-safe
Timm Bäder [Tue, 9 Aug 2016 17:29:41 +0000 (19:29 +0200)]
popover: Make some public API NULL-safe

Instad of potentially dereferencing a NULL-pointer, use the NULL-safe
gtk_popover_get_instance_private before the g_return_if_fail
precondition check.

9 years agolabel: Remove some useless casts
Timm Bäder [Tue, 9 Aug 2016 13:13:01 +0000 (15:13 +0200)]
label: Remove some useless casts

These were casts of a const GdkEvent* to GdkEvent* even though the
function they are passed to takes a const GdkEvent*.

9 years agoUse g_snprintf instead of snprintf.
Ignacio Casal Quinteiro [Wed, 10 Aug 2016 10:29:01 +0000 (12:29 +0200)]
Use g_snprintf instead of snprintf.

snprintf is not compatible with msvc.

9 years agogit.mk: Add config.rpath to the list of ignored files
Emmanuele Bassi [Tue, 9 Aug 2016 16:20:40 +0000 (17:20 +0100)]
git.mk: Add config.rpath to the list of ignored files

This finally cleans up the `git status` output.

9 years agoRemove config.rpath from EXTRA_DIST
Emmanuele Bassi [Tue, 9 Aug 2016 16:16:17 +0000 (17:16 +0100)]
Remove config.rpath from EXTRA_DIST

The config.rpath script was added by running gettextize, but it's not
really needed, as config.rpath is generated by autotools and it's
included in the dist tarball automatically.

Upstream gettextize has been fixed so it does not happen any more:
http://savannah.gnu.org/bugs/index.php?48729

9 years agoUpdated Czech translation
Marek Černocký [Tue, 9 Aug 2016 10:19:03 +0000 (12:19 +0200)]
Updated Czech translation

9 years agoUpdated Slovak translation
Dušan Kazik [Mon, 8 Aug 2016 07:38:30 +0000 (07:38 +0000)]
Updated Slovak translation

9 years agowayland: Link against librt
Matthias Clasen [Sun, 7 Aug 2016 20:11:55 +0000 (16:11 -0400)]
wayland: Link against librt

This is required for shm_open.

https://bugzilla.gnome.org/show_bug.cgi?id=769603

9 years agoheaderbar: Fix size allocation
Matthias Clasen [Fri, 5 Aug 2016 05:08:44 +0000 (01:08 -0400)]
headerbar: Fix size allocation

The GtkHeaderBar gadget implementation was subtly broken: it called
gtk_widget_set_allocation both in gtk_header_bar_size_allocate (with
the actual allocation) and in gtk_header_bar_allocate_contents (with
the content allocation of the main gadget). Dropping the second call
fixes the render node conversion for GtkHeaderBar.

9 years agoUpdated Hebrew translation
Yosef Or Boczko [Sun, 7 Aug 2016 13:05:42 +0000 (16:05 +0300)]
Updated Hebrew translation

9 years agoUpdated Hebrew translation
Yosef Or Boczko [Sun, 7 Aug 2016 12:56:25 +0000 (15:56 +0300)]
Updated Hebrew translation

9 years agogdkwindow: Fix potential NULL pointer dereferences in event code
Philip Withnall [Wed, 20 Nov 2013 17:21:28 +0000 (17:21 +0000)]
gdkwindow: Fix potential NULL pointer dereferences in event code

The event code could potentially dereference pointer_info if the
invariant that ENTER_NOTIFY and LEAVE_NOTIFY events are only emitted on
devices which have pointers is violated elsewhere.

Found with scan-build.

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

9 years agogtkicontheme: Don’t emit warning about fallback theme when in unit tests
Philip Withnall [Wed, 3 Aug 2016 20:12:27 +0000 (21:12 +0100)]
gtkicontheme: Don’t emit warning about fallback theme when in unit tests

In order to eliminate g_test_expect_message() (which doesn’t work with
G_LOG_USE_STRUCTURED), make the warning about the fallback theme not
existing be conditional on the icon theme search path containing a
system path. Any application code which modifies the search path does so
through appends and prepends, so this should not affect whether the
warning is emitted in production.

https://bugzilla.gnome.org/show_bug.cgi?id=769485

9 years agotests: Avoid g_test_expect_message
Matthias Clasen [Sat, 6 Aug 2016 21:10:43 +0000 (17:10 -0400)]
tests: Avoid g_test_expect_message

Instead of using g_test_expect_message, just ignore the warning.

9 years agoGtkMenuButton: use :toggled instead of :clicked
Colomban Wendling [Thu, 28 Jul 2016 22:39:30 +0000 (00:39 +0200)]
GtkMenuButton: use :toggled instead of :clicked

:toggled is triggered on :clicked, so using :toggled lead to the menu
to be popped up at the same time, while allowing to use the toggle state
and avoiding any need to a hack to prevent recursion, which somehow
wasn't enough for double emission of GtkMenuToolButton:show-popup.

https://bugzilla.gnome.org/show_bug.cgi?id=769287

9 years agoUpdated Polish translation
Piotr Drąg [Thu, 4 Aug 2016 19:09:45 +0000 (21:09 +0200)]
Updated Polish translation

9 years agocsslookup: Remove useless cast
Timm Bäder [Thu, 4 Aug 2016 15:32:05 +0000 (17:32 +0200)]
csslookup: Remove useless cast

The style passed to _gtk_css_lookup_resolve is already a
GtkCssStaticStyle.

9 years agoUpdated German translation
Bernd Homuth [Thu, 4 Aug 2016 18:38:09 +0000 (18:38 +0000)]
Updated German translation
(cherry picked from commit 070c4f74868c31cc9316bbad38e2ad42adc43443)

9 years agoW32: Prefer the deadkey combinations that the OS uses
Руслан Ижбулатов [Sat, 16 Jul 2016 09:23:22 +0000 (09:23 +0000)]
W32: Prefer the deadkey combinations that the OS uses

Pick the W32 API for possible deadkey+<something> combinations
and prefer these to other sources of deadkey combos.
Specifically, if W32 API supports at least one combo for a particular
deadkey, only use that data and do not attempt to do other, unsupported
combinations, even if they make sense otherwise.
This is needed to, for example, correctly support US-International
keyboard layout, which produces a combined character for <' + a>
combo, but not for <' + s>, for example.

This is achieved by stashing all the deadkeys that we find in
an array, then doing extra loop through all virtual key codes and
trying to combine them with each of these deadkeys. Any combinations
that produce a single character are cached for later use.

In GTK Simple IM context, call a new GDK W32 function to do a lookup
on that cached combination table early on, among the "special cases"
(which are now partially obsolete).

A limitation of this code is that combinations with more than
one deadkey are not supported, except for combinations that consist
entirely of 2 known deadkeys. The upshot is that lookups should
be relatively fast, as deadkey array stays small and the combination
tree stays shallow.

Note that the use of ToUnicodeEx() seems suboptimal, as it should
be possible to just load a keyboard library (KBD*.DLL) manually
and obtain and use its key table directly. However, that is much more
complicated and would result in a significant rewrite of gdkkeys-win32.
The code from this commit, though hacky, is a direct addition to
existing code and should cover vast majority of the use-cases.

https://bugzilla.gnome.org/show_bug.cgi?id=569581

9 years agoGDK W32: Cache multiple keyboard layouts simultaneously
Руслан Ижбулатов [Wed, 13 Jul 2016 11:41:35 +0000 (11:41 +0000)]
GDK W32: Cache multiple keyboard layouts simultaneously

This changes the group/level semantic.
Previously W32 backend used "group 0/1" to denote "AltGr OFF/ON"
and "level 0/1" to denote "Shift is OFF/ON".
Now "group" means "keyboard layout" and there can be up to 255 groups,
while AltGr and Shift are combined into a single level enum that
takes values between 0 and 4.
Unlike X, W32 doesn't do effective group overriding, meaning that
it will never tell the caller that a different group was actually
used (even for universal keys, such as Enter), because key symbol
table is completely fabricated and there's no point in trying to
save a few of kilobytes of RAM by not duplicating universal key
records for all groups.

Also contains many whitespace changes (tab elimination, fixed
indentation) and cleanup (axed a few global variables, these are
now accessed via the default keymap).

https://bugzilla.gnome.org/show_bug.cgi?id=768722

9 years agoshortcutlabel: Make parameter names match
Timm Bäder [Thu, 4 Aug 2016 11:33:07 +0000 (13:33 +0200)]
shortcutlabel: Make parameter names match

9 years agowidget: remove some unneeded function prototypes
Timm Bäder [Sat, 9 Jul 2016 16:22:27 +0000 (18:22 +0200)]
widget: remove some unneeded function prototypes

9 years agofilechooserwidget: Remove unused enum
Timm Bäder [Fri, 1 Jul 2016 08:51:06 +0000 (10:51 +0200)]
filechooserwidget: Remove unused enum

9 years agolabel: Destroy the popup menu on unmap
Timm Bäder [Tue, 19 Jul 2016 12:40:52 +0000 (14:40 +0200)]
label: Destroy the popup menu on unmap

9 years agoUpdated Spanish translation
Daniel Mustieles [Wed, 3 Aug 2016 12:28:56 +0000 (12:28 +0000)]
Updated Spanish translation

9 years agoModify git.mk to handle our split NLS directories
Emmanuele Bassi [Wed, 3 Aug 2016 10:52:08 +0000 (11:52 +0100)]
Modify git.mk to handle our split NLS directories

GTK+ has two directories for translations: the default 'po' and the
additional 'po-properties' for the GObject properties translation
domain.

Since the content of the translations directories are filled by
autoreconf and gettext, and are duplicated between the two gettext
domains we use, we should handle this ad hoc inside git.mk, instead of
trying to catch up by adding po-properties files in GITIGNOREFILES.

9 years agoAdd ignore file for macro dir
Emmanuele Bassi [Wed, 3 Aug 2016 10:49:15 +0000 (11:49 +0100)]
Add ignore file for macro dir

The contents of the macro directory are too complicated for git.mk to
handle: the contents are filled by autotools and may change between
systems or autotools releases; and we additionally ship our own m4
macros.

To avoid unnecessary noise on the `git status` output we should simply
add an ignore file specifically for the macro directory.

9 years agoSkip GtkShortcutLabel::accelerator in notify test
Matthias Clasen [Wed, 3 Aug 2016 03:59:34 +0000 (23:59 -0400)]
Skip GtkShortcutLabel::accelerator in notify test

This property is not freely settable, since the string
gets parsed.

9 years agomenu: Fix up new properties
Matthias Clasen [Wed, 3 Aug 2016 03:42:11 +0000 (23:42 -0400)]
menu: Fix up new properties

The new positioning-related properties had some quality of
implementation issues, such as incorrect initial values and
excessive change notification. This broke the notify test.

9 years agoMake icontheme test work with structured logging
Matthias Clasen [Wed, 3 Aug 2016 03:23:43 +0000 (23:23 -0400)]
Make icontheme test work with structured logging

Since we've opted in to G_LOG_USE_STRUCTURED, this test
now needs to do its thing using a log writer function.

9 years agoSkip GdkDrawingContext in default value tests
Matthias Clasen [Wed, 3 Aug 2016 02:32:25 +0000 (22:32 -0400)]
Skip GdkDrawingContext in default value tests

GdkDrawingContext objects can't be freely constructed.

9 years agoRequire a new version of GLib
Emmanuele Bassi [Tue, 2 Aug 2016 21:13:54 +0000 (22:13 +0100)]
Require a new version of GLib

We need GLib 2.49.4 because of the new API requirements introduced by
commit 936c1649.

https://bugzilla.gnome.org/show_bug.cgi?id=769451

9 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 2 Aug 2016 14:52:29 +0000 (16:52 +0200)]
Updated Spanish translation

9 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 2 Aug 2016 14:52:19 +0000 (16:52 +0200)]
Updated Spanish translation

9 years agoUpdate the docs for gtk_window_get_position()
Emmanuele Bassi [Tue, 2 Aug 2016 11:28:00 +0000 (12:28 +0100)]
Update the docs for gtk_window_get_position()

Drop mentions of GnomeClient, and add mentions of different windowing
systems instead of just assuming that we're using X11.

9 years agoAdd a release not about GtkTable expand flags
Matthias Clasen [Mon, 1 Aug 2016 13:09:19 +0000 (09:09 -0400)]
Add a release not about GtkTable expand flags

9 years agogtkbox: remove unnecessary queue_compute_expand()
Christoph Reiter [Tue, 26 Jul 2016 11:09:21 +0000 (13:09 +0200)]
gtkbox: remove unnecessary queue_compute_expand()

The expand child property does not have any effect on the
expand state of the GtkBox, so queuing a compute_expand
when changing it is not needed.

https://bugzilla.gnome.org/show_bug.cgi?id=769162

9 years agogtktable: don't try to propagate expand related child props in compute_expand()
Christoph Reiter [Tue, 26 Jul 2016 10:37:55 +0000 (12:37 +0200)]
gtktable: don't try to propagate expand related child props in compute_expand()

It tried to set the expand state if either xexpand/yexpand where true.
Due to a missing queue_compute_expand when adding a child it actually
only computed the expand state in case a child queued after being added
or in case a child had the expand property set (see optimization in
gtk_widget_set_parent)

In my case this broke layouts as a child of GtkCombBox started setting
an exand flag with 3.20 which queued a compute_expand, which in turn
propagated an expand child props set for a cell in the same table up
and overrode the expand child prop of a parent GtkBox.

This removes the custom compute_expand implementation to match the
behaviour of GtkBox (don't propagate child prop expand flags
but let child expand flags override the child props) and not get random
expand behaviour depending on whether and when child widgets set their
expand state.

https://bugzilla.gnome.org/show_bug.cgi?id=769162

9 years agoUpdated Lithuanian translation
Aurimas Černius [Sun, 31 Jul 2016 16:58:56 +0000 (19:58 +0300)]
Updated Lithuanian translation

9 years agodemos: Fix build failure
Bastien Nocera [Wed, 27 Jul 2016 17:13:31 +0000 (19:13 +0200)]
demos: Fix build failure

gtk+/demos/gtk-demo/css_blendmodes.c: In function ‘update_css_for_blend_mode’:
gtk+/demos/gtk-demo/css_blendmodes.c:49:26: error: format not a string literal, argument types not checked [-Werror=format-nonliteral]
                          blend_mode);
                          ^~~~~~~~~~

https://bugzilla.gnome.org/show_bug.cgi?id=769236

9 years agodocs: trivial build fix
Sébastien Wilmet [Thu, 28 Jul 2016 18:51:05 +0000 (20:51 +0200)]
docs: trivial build fix

The error was:
gtk3-scan.c:193: undefined reference to `gtk_shortcuts_label_get_type'
collect2: error: ld returned 1 exit status

It's since commit 7543cd8ce419a4d660f0ff7614f0c6e1cac81804, which made
the GtkShortcutLabel class public.

9 years agoPort GtkPrintOperationPortal to gtk_window_export_handle
Matthias Clasen [Tue, 26 Jul 2016 19:45:39 +0000 (15:45 -0400)]
Port GtkPrintOperationPortal to gtk_window_export_handle

9 years agoPort gtk_show_uri_on_window to gtk_window_export_handle
Matthias Clasen [Tue, 26 Jul 2016 19:44:49 +0000 (15:44 -0400)]
Port gtk_show_uri_on_window to gtk_window_export_handle

As part of this, make it use the new
g_app_info_launch_default_for_uri_async, since we need a way
to unexport the window handle afterwards.

9 years agoPort GtkFileChooserNativePortal to gtk_window_export_handle
Matthias Clasen [Tue, 26 Jul 2016 19:48:18 +0000 (15:48 -0400)]
Port GtkFileChooserNativePortal to gtk_window_export_handle

9 years agoPut window exporting behind a display protocol agnostic API
Matthias Clasen [Tue, 26 Jul 2016 19:46:41 +0000 (15:46 -0400)]
Put window exporting behind a display protocol agnostic API

Introduce a private API meant for abstracting how to get a handle
of a window that can be shared with other processes. The API is
async, since some implementations will require that. Currently,
only X11 is supported, which doesn't.

Based on a patch by Jonas Adahl.

9 years agoGDK W32: Support UTF-16 surrogate pairs passed via VK_PACKET
Руслан Ижбулатов [Sun, 24 Jul 2016 14:26:06 +0000 (14:26 +0000)]
GDK W32: Support UTF-16 surrogate pairs passed via VK_PACKET

This is, essentially, a piece of g_utf16_to_ucs4() built into GDK
W32 keyboard message processing.

https://bugzilla.gnome.org/show_bug.cgi?id=769126

9 years agocss: Fix case where we didn't convert pt => px
Benjamin Otte [Thu, 28 Jul 2016 15:41:21 +0000 (11:41 -0400)]
css: Fix case where we didn't convert pt => px

Fixes gtk_widget_override_font() not behaving properly since commit
df08fc91bdc1d2e4c866122304fabe4dd298a7de.

https://bugzilla.gnome.org/show_bug.cgi?id=768902

9 years agocss: Rename functions
Benjamin Otte [Thu, 28 Jul 2016 15:31:30 +0000 (11:31 -0400)]
css: Rename functions

I don't want to use a generic function for font sizes as font sizes are
special. Plus, the function is only used in one place.

9 years agoProperly declare the font CSS property
Emilio Pozuelo Monfort [Thu, 28 Jul 2016 15:13:15 +0000 (17:13 +0200)]
Properly declare the font CSS property

https://bugzilla.gnome.org/show_bug.cgi?id=769004

9 years agoDon't use one-stop gradients
Emilio Pozuelo Monfort [Thu, 28 Jul 2016 15:12:25 +0000 (17:12 +0200)]
Don't use one-stop gradients

They are deprecated, which causes a warning that makes
the test fail.

https://bugzilla.gnome.org/show_bug.cgi?id=769004

9 years agoFix tiny typo.
Murray Cumming [Thu, 28 Jul 2016 09:12:54 +0000 (11:12 +0200)]
Fix tiny typo.

9 years agoAdwaita: use a transition to animate checks and radios
Lapo Calamandrei [Wed, 27 Jul 2016 20:54:20 +0000 (22:54 +0200)]
Adwaita: use a transition to animate checks and radios

see https://bugzilla.gnome.org/show_bug.cgi?id=762260

9 years agoshortcut-label: add 'disabled-text' property
Georges Basile Stavracas Neto [Tue, 26 Jul 2016 20:12:31 +0000 (17:12 -0300)]
shortcut-label: add 'disabled-text' property

When there's no useful shortcut accelerator set,
GtkShortcutLabel doesn't show any useful information.

To work around that, add a new property to set the
text to be displayed when there's no accelerator
available.

https://bugzilla.gnome.org/show_bug.cgi?id=769205

9 years agoshortcut-label: make it public
Georges Basile Stavracas Neto [Tue, 26 Jul 2016 19:00:47 +0000 (16:00 -0300)]
shortcut-label: make it public

GtkShortcutLabel is a widget that displays a single
shortcut accelerator or gesture in the user interface,
and is currently used by the shortcuts window.

This widget, however, has public value as other applications
also may want to expose their own shortcuts. For instance,
it'll be useful for the Keyboard panel on Control Center and
the new shortcut editor in Pitivi, among others.

This patch exposes GtkShortcutLabel as a public widget,
and adds the necessary documentation.

https://bugzilla.gnome.org/show_bug.cgi?id=769205

9 years agoGtkScrolledWindow: reset scroll history if it contained no real history
Carlos Garnacho [Wed, 27 Jul 2016 18:23:35 +0000 (20:23 +0200)]
GtkScrolledWindow: reset scroll history if it contained no real history

Scroll history must refer to a timespan for the values to be valid, otherwise
we return FALSE, in this case the stored event(s) should be discarded anyway.

9 years agoGtkScrolledWindow: Always uninstall scroll cursor when starting deceleration
Carlos Garnacho [Wed, 27 Jul 2016 18:19:30 +0000 (20:19 +0200)]
GtkScrolledWindow: Always uninstall scroll cursor when starting deceleration

It could be the case that the last scroll event is received long after any
previous scroll event, in this case the last scroll event discards all "old"
scroll events, and scroll_history_finish() returns FALSE because there's no
time/offset deltas in the scroll history.

This is desired so we don't trigger the deceleration effect if there was no
effective velocity, we still must reset the installed scroll cursor, so take
it out of this if() condition.

9 years agoUpdated Portuguese translation
Tiago Santos [Wed, 27 Jul 2016 07:35:14 +0000 (07:35 +0000)]
Updated Portuguese translation

9 years agoStop using one-stop gradients in reftests
Matthias Clasen [Tue, 26 Jul 2016 11:25:01 +0000 (07:25 -0400)]
Stop using one-stop gradients in reftests

This is causing warnings now that make the tests fail.

9 years agoRevert a mistaken change
Matthias Clasen [Tue, 26 Jul 2016 04:21:48 +0000 (00:21 -0400)]
Revert a mistaken change

I thought I needed ot rearrange the ordering of the animation-direction
values for the parser, overlooking the fact that we already parse them
backwards to address this very problem.

9 years agodocs: GtkWidget::style-updated vs GtkStyleContext::changed
Sébastien Wilmet [Sun, 24 Jul 2016 15:20:29 +0000 (17:20 +0200)]
docs: GtkWidget::style-updated vs GtkStyleContext::changed

Explain the difference between those two signals.

Add "Since: 3.0" for GtkStyleContext::changed, since that signal has
been added in commit 9f84e101bf150a85f82e15c56130b1f98b8fdbf0, present
since 2.91.6.

https://bugzilla.gnome.org/show_bug.cgi?id=769047

9 years agodocs: add missing info for gtk_widget_get_style_context()
Sébastien Wilmet [Sun, 24 Jul 2016 11:16:31 +0000 (13:16 +0200)]
docs: add missing info for gtk_widget_get_style_context()

It is important to know whether the returned object can or cannot
change, for a certain widget. For example to connect to the
GtkStyleContext::changed signal.

https://bugzilla.gnome.org/show_bug.cgi?id=769047

9 years agoGtkApplication: some cleanups
Matthias Clasen [Mon, 25 Jul 2016 13:11:21 +0000 (09:11 -0400)]
GtkApplication: some cleanups

Always return an error if we fail to get a dbus proxy; the callers
are only looking whether error is set, not whether the return value
is NULL.

Use the same function for the inhibit proxy as well, and clean up
the sm_proxy in finalize.

9 years agoDon't apply GDK_HINT_RESIZE_INC to GDK_WINDOW_STATE_TILED windows
Simon McVittie [Mon, 6 Jun 2016 08:48:36 +0000 (09:48 +0100)]
Don't apply GDK_HINT_RESIZE_INC to GDK_WINDOW_STATE_TILED windows

This matches the behaviour of Mutter, Metacity and traditional X11
window managers on the window manager side, and is what we want
for at least gnome-terminal. I can't think of any reason why we'd
want incremental resize in any other tiled window.

Signed-off-by: Simon McVittie <smcv@debian.org>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=760944

https://bugzilla.gnome.org/show_bug.cgi?id=755947

9 years agoreftests: override GSETTINGS_SCHEMA_DIR when running tests
Cosimo Cecchi [Sat, 16 Jul 2016 18:48:32 +0000 (11:48 -0700)]
reftests: override GSETTINGS_SCHEMA_DIR when running tests

The reftest depends on a schema we install ourselves. Instruct GIO to
look in the build directory to avoid a cyclic dep.

https://bugzilla.gnome.org/show_bug.cgi?id=768930

9 years agoAvoid a division by zero
Matthias Clasen [Mon, 25 Jul 2016 12:32:24 +0000 (08:32 -0400)]
Avoid a division by zero

It appears that xvfb returns nonsense in its xrandr resources.
Avoid a crash when that happens.

https://bugzilla.gnome.org/show_bug.cgi?id=768999

9 years agoUse g_clear_object in a few more places
Matthias Clasen [Mon, 25 Jul 2016 12:32:08 +0000 (08:32 -0400)]
Use g_clear_object in a few more places

9 years agoAdwaita: use margins for entry image spacing
Lapo Calamandrei [Sun, 24 Jul 2016 17:11:27 +0000 (19:11 +0200)]
Adwaita: use margins for entry image spacing

there used to be a padding, since it was the only option in the
past, that makes little sense now, hence use margins.

9 years agoHC: export unfocused_insensitive_color...
Lapo Calamandrei [Sun, 24 Jul 2016 16:58:45 +0000 (18:58 +0200)]
HC: export unfocused_insensitive_color...

...to reflect Adwaita changes.

9 years agoAdwaita: export yet another named color for insensitive bg/fg...
Lapo Calamandrei [Sun, 24 Jul 2016 16:04:51 +0000 (18:04 +0200)]
Adwaita: export yet another named color for insensitive bg/fg...

...which was needed, but not there, use responsibly.

9 years agoFix a few documentation parameter names
Timm Bäder [Sat, 23 Jul 2016 12:07:47 +0000 (14:07 +0200)]
Fix a few documentation parameter names

9 years agoOpt in to structured logging
Matthias Clasen [Sat, 23 Jul 2016 03:11:44 +0000 (23:11 -0400)]
Opt in to structured logging

Define G_LOG_USE_STRUCTURED, so that all our g_debug, g_warning,
etc calls directly use structured logging and provide source information.

9 years agowayland: Fix build
Carlos Garnacho [Fri, 22 Jul 2016 18:58:55 +0000 (20:58 +0200)]
wayland: Fix build

Missing include pointing to tablet-unstable-v1-client-protocol.h,
pays me for not testing on a clean checkout.

9 years agowayland: Use wl_fixed_t on wp_tablet_tool angle arguments
Carlos Garnacho [Fri, 13 May 2016 13:25:25 +0000 (15:25 +0200)]
wayland: Use wl_fixed_t on wp_tablet_tool angle arguments

This is an incompatible change in tablet protocol v2.

9 years agowayland: Update current tablet support to using v2
Carlos Garnacho [Fri, 13 May 2016 13:19:04 +0000 (15:19 +0200)]
wayland: Update current tablet support to using v2

Only update to using v2 headers/structs. The incompatible changes
to tool events are dealt with in the next commit. Pads aren't handled
in this commit either.

9 years agoAdwaita: dim calendar `uninteresting` days some more
Lapo Calamandrei [Fri, 22 Jul 2016 12:10:56 +0000 (14:10 +0200)]
Adwaita: dim calendar `uninteresting` days some more

see https://bugzilla.gnome.org/show_bug.cgi?id=769003

9 years agoAdwaita: clean calendar styling
Lapo Calamandrei [Wed, 20 Jul 2016 20:48:25 +0000 (22:48 +0200)]
Adwaita: clean calendar styling

remove some non working props and generally simplify and clean up
the code, in the process address:

https://bugzilla.gnome.org/show_bug.cgi?id=769003