gtk4.git
13 years agoUpdated galician translations
Fran Diéguez [Sun, 17 Mar 2013 22:22:29 +0000 (23:22 +0100)]
Updated galician translations

13 years agoGtkFontButton: default show-preview-entry to TRUE
Matthias Clasen [Sun, 17 Mar 2013 20:24:34 +0000 (16:24 -0400)]
GtkFontButton: default show-preview-entry to TRUE

This is what the property declaration says, after all.
https://bugzilla.gnome.org/show_bug.cgi?id=695948

13 years agoGtkEntryAccessible: also handle entry icon tooltip NULL in notify_gtk.
Alban Browaeys [Thu, 7 Mar 2013 14:54:04 +0000 (15:54 +0100)]
GtkEntryAccessible: also handle entry icon tooltip NULL in notify_gtk.

To clear the tooltip one is to set the tooltip to NULL. Though
the GtkEntryAccessible expect this tooltip to not be NULL in
gtk_entry_accessible_notify_gtk (already handling this case
 in its _init).

Fixes:
** (epiphany:23914): CRITICAL **: atk_object_set_description: assertion
`description != NULL' failed
when epiphany g_object_set the entry icon tooltip to NULL (clear the
tooltip) in its find bar.

https://bugzilla.gnome.org/show_bug.cgi?id=695375

13 years agoGtkLabel: fix mnemonic-keyval when use-markup is true
Sébastien Wilmet [Tue, 24 Apr 2012 20:22:22 +0000 (22:22 +0200)]
GtkLabel: fix mnemonic-keyval when use-markup is true

To extract the mnemonic key value, the string must contain the
underscore. But when the "gtk-auto-mnemonics" setting is true and when
the Alt key is not pressed, the underscore must not be displayed. The
problem was that the 'new_str' variable was used for both purposes:
extract the text to display, and extract the accelerator character.

When the underscore must not be visible, the underscores were removed
from the 'new_str' variable before extracting the accelerator character.

Now there are two strings, one for each purpose.

https://bugzilla.gnome.org/show_bug.cgi?id=674759

13 years agoUpdated POTFILES.in
Piotr Drąg [Sun, 17 Mar 2013 16:47:32 +0000 (17:47 +0100)]
Updated POTFILES.in

13 years agocss: Add style entries for client side decorations to default CSS
Rob Bradford [Tue, 19 Feb 2013 12:55:07 +0000 (12:55 +0000)]
css: Add style entries for client side decorations to default CSS

13 years agowindow: Render inner and outer border separately
Matthias Clasen [Sun, 10 Mar 2013 17:46:59 +0000 (13:46 -0400)]
window: Render inner and outer border separately

This lets themes render a window frame in the inner border,
and a shadow or nothing in the outer border.

13 years agowindow: Split the border into inner and outer
Matthias Clasen [Sun, 17 Mar 2013 01:37:51 +0000 (21:37 -0400)]
window: Split the border into inner and outer

For now, nothing changes, we're using the sum of inner and
outer border everywhere.

In the future, we will make the inner border the visible
window frame, and the outer border the shadow/resize border.

13 years agowindow: Hide the border when maximized
Matthias Clasen [Sun, 10 Mar 2013 07:30:26 +0000 (03:30 -0400)]
window: Hide the border when maximized

Dragging windows from the border is not useful when maximized.

13 years agowindow: Add a window menu to the titlebar
Matthias Clasen [Sun, 10 Mar 2013 06:39:19 +0000 (01:39 -0500)]
window: Add a window menu to the titlebar

The menu is triggered by right-click on the titlebar
or with the menu key.

13 years agowindow: Allow resizing by clicking on the border
Matthias Clasen [Sun, 10 Mar 2013 06:11:18 +0000 (01:11 -0500)]
window: Allow resizing by clicking on the border

We also change cursors to indicate what interactions
are possible in various window regions.
Double clicking on the titlebar maximizes the window.

13 years agowindow: Allow theme control over window buttons
Matthias Clasen [Sun, 10 Mar 2013 04:03:38 +0000 (23:03 -0500)]
window: Allow theme control over window buttons

Add a style property to control the presence and order of
window buttons. We allow buttons at the left and right side,
they can be specified like this: icon,close:minimize,maximize.

Also, change the default button layout back to have just a close
button on the right, use icons in buttons, and set style classes
on the buttons, to allow better theming.

13 years agoAdd icons for frame buttons
Matthias Clasen [Sat, 16 Mar 2013 23:11:26 +0000 (19:11 -0400)]
Add icons for frame buttons

Add window-minimize, window-maximize, window-restore and window-delete
icons to the builtin icon theme. These will be used for icons in
the window buttons, and the expectation is that the icon theme
will provide icons matching the desired decoration style.

window-delete is used instead of window-close, since window-close
is also used for GTK_STOCK_CLOSE, and the two may require different
styles when used inside the application vs in the window frame.

13 years agowindow: Hide frame buttons when appropriate
Matthias Clasen [Sun, 10 Mar 2013 03:28:39 +0000 (22:28 -0500)]
window: Hide frame buttons when appropriate

Hide the close button if the window is not deletable, hide the
maximize button if the window is not resizable, and hide all
buttons if the window is a dialog.

Update buttons when the window state changes.

13 years agowindow: Use same title fallback mechanism as X backend
Matthias Clasen [Sun, 17 Mar 2013 16:08:17 +0000 (12:08 -0400)]
window: Use same title fallback mechanism as X backend

This looks at the application name or program name and uses that
as the title if gtk_window_set_title has not been called.

13 years agowindow: Add support for enabling client decorations on non-Wayland
Matthias Clasen [Sun, 17 Mar 2013 16:06:26 +0000 (12:06 -0400)]
window: Add support for enabling client decorations on non-Wayland

Client side decorations can be enabled on non-Wayland platforms by
setting the GTK_CSD="1" environment variable.

We must ensure we have a GdkVisual that has an alpha channel since
the decorations rely on transparency. If we cannot get a visual with
an alpha channel then we do not enable client side decorations.

13 years agowindow: Hide the decorations on fullscreen
Rob Bradford [Thu, 31 Jan 2013 17:24:31 +0000 (17:24 +0000)]
window: Hide the decorations on fullscreen

This is expected behaviour for fullscreen windows.

13 years agowindow: Don't set a background pattern/colour when client decorated
Rob Bradford [Tue, 19 Feb 2013 19:26:19 +0000 (19:26 +0000)]
window: Don't set a background pattern/colour when client decorated

Otherwise we'll potentially get some background sticking through our rounded
corners in our decorations. The actual background will get drawn as part of
the decoration drawing.

13 years agowindow: Add initial support for client-side decorations under Wayland
Kristian Høgsberg [Wed, 11 Apr 2012 16:55:54 +0000 (17:55 +0100)]
window: Add initial support for client-side decorations under Wayland

This change comprises four main parts:
 * the creation of the widgets that form the decorations,
 * implementation of get_preferred_height/width, and the for_width/for_height
   variants,
 * taking the decorations into account when allocating,
 * and drawing the decorations themselves.

Kristian did the bulk of the original work on this but any bugs are almost
certainly mine through the many refactorings and rebasings.

13 years agoAdd GtkHeaderBar
Matthias Clasen [Sun, 10 Mar 2013 19:15:45 +0000 (15:15 -0400)]
Add GtkHeaderBar

This widget has been developed as GdHeaderBar in libgd.
The copy here has been renamed and changed to use GTK+
internals.

13 years agowindow: Allow _gtk_window_set_allocation to return a modified allocation
Rob Bradford [Tue, 26 Feb 2013 19:33:04 +0000 (19:33 +0000)]
window: Allow _gtk_window_set_allocation to return a modified allocation

Update the documentation and users of this function to handle
the future case that that we have some internal decorations to the window and
useable allocation is thus smaller.

By having a separate out parameter there is no need to have an in/out function
and allows for greater robustness.

The current implementation simply returns the allocation provided.

13 years agoAdd an index for 3.10 api additions
Matthias Clasen [Sun, 17 Mar 2013 07:18:22 +0000 (03:18 -0400)]
Add an index for 3.10 api additions

13 years agoDefine 3.10 version macros
Matthias Clasen [Sun, 17 Mar 2013 07:07:57 +0000 (03:07 -0400)]
Define 3.10 version macros

13 years agoBump version to 3.9.0
Matthias Clasen [Sun, 17 Mar 2013 07:07:25 +0000 (03:07 -0400)]
Bump version to 3.9.0

Needed so that GDK_AVAILABLE_IN_3_10 works.

13 years agoUpdated Belarusian translation.
Ihar Hrachyshka [Sun, 17 Mar 2013 12:52:46 +0000 (15:52 +0300)]
Updated Belarusian translation.

13 years agogdk_{pointer,keyboard}_grab are deprecated
Guido Günther [Fri, 15 Mar 2013 17:24:27 +0000 (18:24 +0100)]
gdk_{pointer,keyboard}_grab are deprecated

so mention the non deprecated gdk_device_grab as well

13 years agoUpdated Danish translation of the properties
Ask H. Larsen [Sun, 17 Mar 2013 07:37:56 +0000 (08:37 +0100)]
Updated Danish translation of the properties

13 years agoUpdated Danish translation
Ask H. Larsen [Sun, 17 Mar 2013 07:14:27 +0000 (08:14 +0100)]
Updated Danish translation

13 years agoSmall API doc fixes
Sébastien Wilmet [Sat, 16 Mar 2013 14:12:35 +0000 (15:12 +0100)]
Small API doc fixes

For gtk_text_iter_get_char(), due to the "Returns" at the beginning of
the description, the description was not visible. So the first sentence
has been reworded.

13 years agoGtkGrid: Tweak size allocation
Matthias Clasen [Sat, 16 Mar 2013 03:10:39 +0000 (23:10 -0400)]
GtkGrid: Tweak size allocation

If there are rows that contain only spanning children,
our algorithm was unnecessarily distributing extra space
to the other rows, even if they contain only non-expanding
children.

We improve the behaviour by treating rows containing only
spanning children as expanding.

13 years agoAdd another grid sizing reftest
Matthias Clasen [Sat, 16 Mar 2013 03:31:05 +0000 (23:31 -0400)]
Add another grid sizing reftest

This test exercises spanning children such that there are
rows which contain only spanning cells.

13 years agostylecontext: Do invalidation on first resize container
Benjamin Otte [Fri, 15 Mar 2013 23:05:03 +0000 (00:05 +0100)]
stylecontext: Do invalidation on first resize container

... instead of taking the last one we find. This is necessary as
attached widgets (mostly menus) can be attached to an invisible widget,
but we still want to invalidate styles for them.

https://bugzilla.gnome.org/show_bug.cgi?id=695772

13 years agoUpdated po-properties French translation
Claude Paroz [Fri, 15 Mar 2013 22:29:06 +0000 (23:29 +0100)]
Updated po-properties French translation

Forward-port from gtk-3-8 branch.

13 years agodocs: Typo fix
Christian Persch [Fri, 15 Mar 2013 17:49:56 +0000 (18:49 +0100)]
docs: Typo fix

13 years agoreftests: Add reftest for separator fixes
Benjamin Otte [Fri, 15 Mar 2013 17:17:40 +0000 (18:17 +0100)]
reftests: Add reftest for separator fixes

Reftest for 61c4425161be3146c0c753a766f35442549bfe72.

Includes a workaround for a grid bug which should be removed once
GtkGrid is fixed.

13 years agoseparator: Don't use padding and borders wrongly
Benjamin Otte [Fri, 15 Mar 2013 13:52:49 +0000 (14:52 +0100)]
separator: Don't use padding and borders wrongly

Borders for size requests and padding for drawing? Really?

Removed them, as that is the least invasive fix shortly before a
release.

13 years agowayland: Rename GdkDeviceManagerCore to GdkWaylandDeviceManager
Matthias Clasen [Fri, 15 Mar 2013 11:27:35 +0000 (07:27 -0400)]
wayland: Rename GdkDeviceManagerCore to GdkWaylandDeviceManager

The 'core' naming is just an artifact of having several generations
of input extensions in X11, no need to propagate that into other
backends.

13 years agoInclude config.h in gdkenumtypes.c.template
Matthias Clasen [Fri, 15 Mar 2013 12:25:14 +0000 (08:25 -0400)]
Include config.h in gdkenumtypes.c.template

At the same time, drop the GDK_ENABLE_BROKEN define.
Nothing broken in GDK broken anymore :-)

13 years agoUpdated Galician translations
Fran Diéguez [Wed, 13 Mar 2013 09:26:18 +0000 (10:26 +0100)]
Updated Galician translations

Conflicts:
po/gl.po

13 years agoAssamese translation updated for gnome 3.8
Nilamdyuti Goswami [Fri, 15 Mar 2013 09:42:09 +0000 (15:12 +0530)]
Assamese translation updated for gnome 3.8

13 years agoAssamese translation updated for gnome 3.8
Nilamdyuti Goswami [Fri, 15 Mar 2013 09:31:04 +0000 (15:01 +0530)]
Assamese translation updated for gnome 3.8

13 years agoMinor doc cleanup
Matthias Clasen [Fri, 15 Mar 2013 00:24:19 +0000 (20:24 -0400)]
Minor doc cleanup

Remove widget references from interface docs.

https://bugzilla.gnome.org/show_bug.cgi?id=695874

13 years agoa11y: Emit text-changed signals directly
Benjamin Otte [Thu, 14 Mar 2013 23:54:26 +0000 (00:54 +0100)]
a11y: Emit text-changed signals directly

When setting new text on the label, the text-changed::delete signal
needs to be emitted before deleting the text (so that atk-bridge can
query the old text) while the text-changed::insert event needs to happen
afterwards (for the same reason). The old code using the notify signal
was only emitted after changing the text.

13 years agoGtkApplication: Allow passing windows on non-X11 targets
Bastien Nocera [Wed, 13 Mar 2013 14:32:46 +0000 (15:32 +0100)]
GtkApplication: Allow passing windows on non-X11 targets

It's not the application's fault that we don't do anything
with the passed GtkWindow on non-X11, so don't try to get an XID
on non-X11.

https://bugzilla.gnome.org/show_bug.cgi?id=695783

13 years agoUpdated Indonesian translation
Andika Triwidada [Thu, 14 Mar 2013 11:22:04 +0000 (18:22 +0700)]
Updated Indonesian translation

13 years agoUpdated Korean translation
Changwoo Ryu [Tue, 12 Mar 2013 20:14:03 +0000 (05:14 +0900)]
Updated Korean translation

13 years agoprinting: Use DBus calls instead of Avahi API
Marek Kasik [Tue, 12 Mar 2013 16:18:16 +0000 (17:18 +0100)]
printing: Use DBus calls instead of Avahi API

Converts usage of Avahi API to DBus calls. This change allows
us to remove dependency on avahi-gobject and avoids of possible
circular dependency.
Lists printers if Gtk+ is compiled with CUPS 1.6 or newer.

13 years agoprinting: Remove usage of GError in connection test
Marek Kasik [Wed, 13 Mar 2013 14:24:50 +0000 (15:24 +0100)]
printing: Remove usage of GError in connection test

Really remove usage of error in previous commit.

Related to
https://bugzilla.gnome.org/show_bug.cgi?id=695714

13 years agoprinting: Don't hang when getting info for Avahi printer
Marek Kasik [Wed, 13 Mar 2013 13:57:55 +0000 (14:57 +0100)]
printing: Don't hang when getting info for Avahi printer

Asynchronously create testing connection to address returned
by GaServiceResolver. Don't request info for such printer if
the connection fails.

https://bugzilla.gnome.org/show_bug.cgi?id=695714

13 years agoAdd a missing runtime check
Matthias Clasen [Wed, 13 Mar 2013 10:01:57 +0000 (06:01 -0400)]
Add a missing runtime check

The cursor theme and size settting code was ifdefed to only
be compiled with the X11 backend, but it didn't check for
running under X at runtime. Fix that.
http://bugzilla.gnome.org/695495

13 years agoprinting: Show Avahi printers on IPv6 network
Marek Kasik [Tue, 12 Mar 2013 15:26:37 +0000 (16:26 +0100)]
printing: Show Avahi printers on IPv6 network

Create correct URIs for printers on IPv6 networks.

13 years agogdk/gdkwindow.c, gtk/gtkwidget.c: Include fallback-c89.c
Chun-wei Fan [Thu, 21 Feb 2013 07:54:18 +0000 (15:54 +0800)]
gdk/gdkwindow.c, gtk/gtkwidget.c: Include fallback-c89.c

Both of them started to make use of round(), a C99 function.  So, include
fallback-c89.c to provide a fallback implementation for round() for
compilers that don't have round()

https://bugzilla.gnome.org/show_bug.cgi?id=694339

13 years agogdk/fallback-c89.c: Add fallback for round()
Chun-wei Fan [Thu, 21 Feb 2013 07:50:54 +0000 (15:50 +0800)]
gdk/fallback-c89.c: Add fallback for round()

This is essentially done by simply copying from gtk/fallback-c89.c

https://bugzilla.gnome.org/show_bug.cgi?id=694339

13 years agotreeview: Don't emit cursor-changed in destruction
John Lindgren [Wed, 13 Mar 2013 01:03:45 +0000 (02:03 +0100)]
treeview: Don't emit cursor-changed in destruction

This is a workaround until we have a good idea on how we want to handle
signal disconnection before/during/after dispose and destroy vfunc
calls.

https://bugzilla.gnome.org/show_bug.cgi?id=671939

13 years agoreftests: Add animation loading test
Benjamin Otte [Tue, 12 Mar 2013 20:58:47 +0000 (21:58 +0100)]
reftests: Add animation loading test

XPM files weren't loaded properly.

https://bugzilla.gnome.org/show_bug.cgi?id=695228

13 years agoDrop the Motif DND protocol
Matthias Clasen [Sat, 9 Mar 2013 00:30:02 +0000 (19:30 -0500)]
Drop the Motif DND protocol

The implementation is not working and unused, since XDND is the
de-facto standard under X nowadays.

https://bugzilla.gnome.org/show_bug.cgi?id=695476

13 years agofilechooserbutton: Duh, remove all the timeouts after tests
Federico Mena Quintero [Tue, 12 Mar 2013 19:34:13 +0000 (13:34 -0600)]
filechooserbutton: Duh, remove all the timeouts after tests

Otherwise tests may start failing if they start after an old timeout triggers.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Add a description string to assertions about emitted signals
Federico Mena Quintero [Tue, 12 Mar 2013 18:36:14 +0000 (12:36 -0600)]
filechooserbutton: Add a description string to assertions about emitted signals

On failed tests, this lets us more easily see see which assertion failed.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Give proper names to the SELECT_FOLDER tests with cancelled dialog
Federico Mena Quintero [Tue, 12 Mar 2013 18:24:06 +0000 (12:24 -0600)]
filechooserbutton: Give proper names to the SELECT_FOLDER tests with cancelled dialog

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Don't duplicate tests for GTK_RESPONSE_DELETE_EVENT
Federico Mena Quintero [Tue, 12 Mar 2013 18:06:07 +0000 (12:06 -0600)]
filechooserbutton: Don't duplicate tests for GTK_RESPONSE_DELETE_EVENT

Instead, re-run tests made for GTK_RESPONSE_CANCEL with the delete-event response.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Do not propagate state from the dialog unless it is active
Federico Mena Quintero [Tue, 12 Mar 2013 17:21:35 +0000 (11:21 -0600)]
filechooserbutton: Do not propagate state from the dialog unless it is active

Change of plans to match the tests from the previous commit.

The state of the underlying dialog is never reflected by GtkFileChooserButton's API,
as the dialog is a transient thing.  The file chooser button only updates its state from the dialog,
and reflects the dialog's state, when the dialog has been confirmed and dismissed by the user.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: whitespace fixes
Federico Mena Quintero [Tue, 12 Mar 2013 17:02:38 +0000 (11:02 -0600)]
filechooserbutton: whitespace fixes

13 years agofilechooserbutton: For the open-dialog-cancel-* tests, add a real nomenclature and...
Federico Mena Quintero [Tue, 12 Mar 2013 16:59:43 +0000 (10:59 -0600)]
filechooserbutton: For the open-dialog-cancel-* tests, add a real nomenclature and test all the combinations

We used to have numeric names, which are a pain to maintain when new tests are added.
Now we have a real nomenclature (see the comment at the beginning of the open-dialog-cancel-* tests),
which lets us see easily if we have tested all the combinations.

Also, added all the combinations that were missing and removed redundant tests.
Not all the tests pass currently.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Update the button's state only on programmatic changes to the...
Federico Mena Quintero [Tue, 12 Mar 2013 16:05:22 +0000 (10:05 -0600)]
filechooserbutton: Update the button's state only on programmatic changes to the filechooser's state

The idea is that the button will only update its state of the selection and current folder
when changes to those are done either by the calling program (with the filechooser's API)
or when the user actually confirms and dismisses the underlying GtkFileChooserDialog.

If the user makes changes to the dialog but has not dismissed it yet, those changes
will not be reflected in the button (as one would expect).

This commit also makes sure the current-folder-changed and selection-changed signals
are emitted at the right times.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Emit 'selection-changed' when changing the selection programmatically
Federico Mena Quintero [Tue, 12 Mar 2013 03:54:22 +0000 (21:54 -0600)]
filechooserbutton: Emit 'selection-changed' when changing the selection programmatically

We only emitted that signal when the selection changed through the underlying GtkFileChooserDialog.

To do this when the dialog is not active and the selection is changed by the calling program
(instead of by the user), we need to wait until the GtkFileChooserButton's UI has been updated
via an async callback from GIO.  So, we keep track of whether an entry point into the
button's API caused a programmatic change in the selection.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: For tests, wait for signals instead of just sleeping
Federico Mena Quintero [Tue, 12 Mar 2013 03:17:34 +0000 (21:17 -0600)]
filechooserbutton: For tests, wait for signals instead of just sleeping

This should let tests complete faster.  Also, this will let us test
that the correct signals are actually being emitted.

The tests now fail, as the signals are not being emitted when they
should.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: whitespace fixes
Federico Mena Quintero [Tue, 12 Mar 2013 02:06:10 +0000 (20:06 -0600)]
filechooserbutton: whitespace fixes

13 years agoUpdated Korean translation
Changwoo Ryu [Tue, 12 Mar 2013 20:14:03 +0000 (05:14 +0900)]
Updated Korean translation

13 years agolabel: Fix memleak
Carl-Anton Ingmarsson [Tue, 12 Mar 2013 19:22:18 +0000 (20:22 +0100)]
label: Fix memleak

Caused by recent refactoring in e09957a47da9425cc26d1b33cb4e9cc3e92e9ac7

13 years ago[l10n] Updated Catalan translation
Gil Forcada [Tue, 12 Mar 2013 18:51:17 +0000 (19:51 +0100)]
[l10n] Updated Catalan translation

13 years agostylecontext: If we force invalidate, invalidate all properties
Benjamin Otte [Tue, 12 Mar 2013 11:44:03 +0000 (12:44 +0100)]
stylecontext: If we force invalidate, invalidate all properties

We only FORCE_INVALIDATE when something weird changes that the CSS
machinery can't detect. But now that our style_updated functions skip
recomputations when some properties don't change we want to make sure
these recomputations are still run. So we just claim all properties
changed.

http://bugzilla.gnome.org/show_bug.cgi?id=695482

13 years agolabel: refactor code
Benjamin Otte [Tue, 12 Mar 2013 11:43:14 +0000 (12:43 +0100)]
label: refactor code

If text didn't change, don't change it.

And if that sentence made no sense, that might explain why I refactored.

13 years agocombobox: popdown the combobox when on 'grab-broken-event'
Aleksander Morgado [Wed, 6 Mar 2013 12:16:12 +0000 (13:16 +0100)]
combobox: popdown the combobox when on 'grab-broken-event'

And also explicitly remove pointer/keyboard grabs from the display.

Whenever the grab is reported lost, we should popdown the combobox, so that the
GDK_WINDOW_TEMP window is hidden and removed from the toplevel, as done with
the menu for example.

Leaving the GDK_WINDOW_TEMP window open when re-activating the application
triggers several issues in the win32 backend, due to restacking windows of the
non-toplevel group into the toplevel group:

https://bugzilla.gnome.org/show_bug.cgi?id=695200

13 years agodocs: Disable GtkPlug/GtkSocket docs on Wayland
Bastien Nocera [Tue, 12 Mar 2013 07:33:57 +0000 (08:33 +0100)]
docs: Disable GtkPlug/GtkSocket docs on Wayland

As those two types do not exist in the Wayland backend,
and the docs build would fail.

https://bugzilla.gnome.org/show_bug.cgi?id=695682

13 years agofilechooserbutton: During unselect_all(), update the widgets in all situations
Federico Mena Quintero [Tue, 12 Mar 2013 00:12:23 +0000 (18:12 -0600)]
filechooserbutton: During unselect_all(), update the widgets in all situations

Not just if we thought we had a selection.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: Add tests for unselect_all()
Federico Mena Quintero [Mon, 11 Mar 2013 23:32:28 +0000 (17:32 -0600)]
filechooserbutton: Add tests for unselect_all()

These all pass except one.

13 years agofilechooserbutton: HACK: Set a reasonable size for the dialog in the tests' second...
Federico Mena Quintero [Mon, 11 Mar 2013 23:27:44 +0000 (17:27 -0600)]
filechooserbutton: HACK: Set a reasonable size for the dialog in the tests' second iteration

Something is causing the GtkFileChooserDialog to be resized really small on the second time it is run
during each test for GtkFileChooserButton.  So as a temporary hack we set it to 500x500 pixels on
the second run, so the size allocation code doesn't bomb on us.

13 years agogtkfilechooserbutton: In tests, allow the possibility of doing unselect_all
Federico Mena Quintero [Mon, 11 Mar 2013 23:11:34 +0000 (17:11 -0600)]
gtkfilechooserbutton: In tests, allow the possibility of doing unselect_all

13 years agofilechooserbutton: When the combo box changes, set the *file*, not the current folder
Federico Mena Quintero [Mon, 11 Mar 2013 22:48:17 +0000 (16:48 -0600)]
filechooserbutton: When the combo box changes, set the *file*, not the current folder

The currently-selected file *is* the selection even in SELECT_FOLDER mode.  Do not confuse this
with the current folder.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agoPunjabi: Translation updated (aalam)
A S Alam [Mon, 11 Mar 2013 14:26:55 +0000 (19:56 +0530)]
Punjabi: Translation updated (aalam)

13 years agoUpdated Serbian translation
Мирослав Николић [Mon, 11 Mar 2013 10:34:40 +0000 (11:34 +0100)]
Updated Serbian translation

13 years ago[l10n] Updated German translation
Mario Blättermann [Mon, 11 Mar 2013 07:48:10 +0000 (08:48 +0100)]
[l10n] Updated German translation

13 years ago[l10n] Updated German translation
Mario Blättermann [Mon, 11 Mar 2013 07:46:58 +0000 (08:46 +0100)]
[l10n] Updated German translation

13 years agoUpdated Indonesian translation of po-properties
Andika Triwidada [Mon, 11 Mar 2013 05:47:31 +0000 (12:47 +0700)]
Updated Indonesian translation of po-properties

13 years agoUpdated Indonesian translation
Andika Triwidada [Mon, 11 Mar 2013 05:07:24 +0000 (12:07 +0700)]
Updated Indonesian translation

13 years agoFix typo reqired -> required
Sébastien Wilmet [Sun, 10 Mar 2013 16:22:26 +0000 (17:22 +0100)]
Fix typo reqired -> required

13 years agoUpdated Brazilian Portuguese translation
Enrico Nicoletto [Sat, 9 Mar 2013 23:52:10 +0000 (20:52 -0300)]
Updated Brazilian Portuguese translation

13 years agoUpdated Brazilian Portuguese translation
Enrico Nicoletto [Sat, 9 Mar 2013 23:12:05 +0000 (20:12 -0300)]
Updated Brazilian Portuguese translation

Conflicts:
po-properties/pt_BR.po

13 years agoUpdated Brazilian Portuguese translation
Enrico Nicoletto [Sat, 9 Mar 2013 22:20:22 +0000 (19:20 -0300)]
Updated Brazilian Portuguese translation

13 years agotreeview: fix a critical warning
Cosimo Cecchi [Fri, 8 Mar 2013 22:56:08 +0000 (17:56 -0500)]
treeview: fix a critical warning

gtk_tree_view_column_unset_tree_view() resets column->priv->tree_view to
NULL.
The function is called when a column is removed, but later from the same
function we would call _gtk_tree_view_column_unrealize_button(), which
expects column->priv->tree_view to be != NULL, causing these critical
warnings

Gtk-CRITICAL **: gtk_widget_unregister_window: assertion
`GTK_IS_WIDGET (widget)' failed

This commit moves the call to unset the tree view after the button is
unrealized.

https://bugzilla.gnome.org/show_bug.cgi?id=695473

13 years agodocs: change id for Wayland section to not conflict with OS X
Evan Nemerson [Thu, 7 Mar 2013 19:32:11 +0000 (11:32 -0800)]
docs: change id for Wayland section to not conflict with OS X

https://bugzilla.gnome.org/show_bug.cgi?id=695391

13 years agoAvoid a warning under Wayland
Matthias Clasen [Sat, 9 Mar 2013 00:44:34 +0000 (19:44 -0500)]
Avoid a warning under Wayland

Just having an ifdef is not enough, we need to check the type
of the window at runtime.

13 years agoUpdated Polish translation
Piotr Drąg [Fri, 8 Mar 2013 19:38:32 +0000 (20:38 +0100)]
Updated Polish translation

13 years agoUpdated POTFILES.in
Piotr Drąg [Fri, 8 Mar 2013 19:32:48 +0000 (20:32 +0100)]
Updated POTFILES.in

13 years agofilechooserbutton: Update the combo box even after the dialog is cancelled
Federico Mena Quintero [Fri, 8 Mar 2013 18:43:10 +0000 (12:43 -0600)]
filechooserbutton: Update the combo box even after the dialog is cancelled

We assumed that we didn't have to update the combo box if the dialog got cancelled,
as it should simply retain its previous contents.  But this assumption doesn't work
as the dialog is brought up with the 'Other...' item - we don't want the
combo box to keep showing 'Other...' if the dialog is cancelled.

The test from the previous commit now passes.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: New test for opening the Other item in the combo box and then...
Federico Mena Quintero [Fri, 8 Mar 2013 18:37:20 +0000 (12:37 -0600)]
filechooserbutton: New test for opening the Other item in the combo box and then canceling the dialog

There is this bug:

1. Start with a file chooser button in SELECT_FOLDER mode, and select a folder from the combo box.

2. Click on the button's combo box, select 'Other...'

3. You get the file chooser dialog. Cancel the dialog.

4. The file chooser button's combo box still shows 'Other...' instead of
   showing the selection from (1).

This is a test to ensure that the original selection is restored.
The test fails right now.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agoFix the build
Matthias Clasen [Fri, 8 Mar 2013 11:26:15 +0000 (06:26 -0500)]
Fix the build

This was broken in commit 5bbbc47a4c306653e8347f7afb85a940a503f755

13 years agospinbutton: don't override initial text in non-numeric-only spin buttons
Aleksander Morgado [Wed, 6 Mar 2013 18:16:25 +0000 (19:16 +0100)]
spinbutton: don't override initial text in non-numeric-only spin buttons

https://bugzilla.gnome.org/show_bug.cgi?id=695312

13 years agofilechooserbutton: Don't show the (None) item when the combo box is popped up
Federico Mena Quintero [Thu, 7 Mar 2013 23:20:16 +0000 (17:20 -0600)]
filechooserbutton: Don't show the (None) item when the combo box is popped up

This is surprisingly tricky, since the (None) item *has* to be a visible item while
the combo box is *not* popped up, so that it can show its contents.  But the item
has to be *not* visible when the combo box is popped up.

Also, update the whole button's selection, not just the underlying dialog's, when
the combo box changes its selection - based on a patch by Paul Davis in
https://bugzilla.gnome.org/show_bug.cgi?id=691040#c20

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
13 years agofilechooserbutton: In SELECT_FOLDER, when the selection is empty, show (None) in...
Federico Mena Quintero [Thu, 7 Mar 2013 20:14:47 +0000 (14:14 -0600)]
filechooserbutton: In SELECT_FOLDER, when the selection is empty, show (None) in the combo box

We do this by adding a new row type, ROW_TYPE_EMPTY_SELECTION.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>