gtk4.git
10 years agoAdwaita: Remove non-working selector
Jasper St. Pierre [Sun, 2 Aug 2015 19:55:10 +0000 (12:55 -0700)]
Adwaita: Remove non-working selector

This was suppsed to be ".list-row.button", I'd imagine, but the missing
dot meant it never worked in the first place. Just clean it up.

10 years agoDon't trigger extraneous redraws on font / text-attr changes
Jasper St. Pierre [Sun, 2 Aug 2015 19:51:21 +0000 (12:51 -0700)]
Don't trigger extraneous redraws on font / text-attr changes

Any time that these change, we shouldn't need to force a redraw on
the widget either.

10 years agowidget: Don't queue redraws for properties that don't affect anything
Jasper St. Pierre [Sun, 2 Aug 2015 19:37:03 +0000 (12:37 -0700)]
widget: Don't queue redraws for properties that don't affect anything

Properties like transition-property might change when hovering over
something, even if the property itself does not change. These properties
don't affect drawing, so don't queue redraws for them.

10 years agolistbox: Don't queue draws on the entire widget
Jasper St. Pierre [Sun, 2 Aug 2015 19:36:28 +0000 (12:36 -0700)]
listbox: Don't queue draws on the entire widget

The set_state_flags code should already queue redraws when needed.

10 years agocssimage: Don't start a transition if the start and end are the same
Jasper St. Pierre [Sun, 2 Aug 2015 19:35:45 +0000 (12:35 -0700)]
cssimage: Don't start a transition if the start and end are the same

Since a lot of Adwaita sets transition: all, it's easy to end up in a
state where we're making dummy transitions for all of the icons, most of
which we'll never be showing.

10 years agoInspector: Preserve selection when rescanning
Matthias Clasen [Sun, 2 Aug 2015 19:20:17 +0000 (15:20 -0400)]
Inspector: Preserve selection when rescanning

We are semiregularly repopulating the object tree. Whenever
we do, we loose the selection, needlessly. Do the extra work
to preserve it.

http://bugzilla.gnome.org/show_bug.cgi?id=753130

10 years agocleanup: do not set members to null in finalize
Paolo Borelli [Sat, 1 Aug 2015 19:31:49 +0000 (21:31 +0200)]
cleanup: do not set members to null in finalize

10 years agocleanup: no need to set members to null when freeing the struct
Paolo Borelli [Sat, 1 Aug 2015 19:30:53 +0000 (21:30 +0200)]
cleanup: no need to set members to null when freeing the struct

10 years agocleanup: one more g_slist_free_full
Paolo Borelli [Sat, 1 Aug 2015 19:30:16 +0000 (21:30 +0200)]
cleanup: one more g_slist_free_full

10 years agoUpdated POTFILES.skip
Piotr Drąg [Sat, 1 Aug 2015 17:25:16 +0000 (19:25 +0200)]
Updated POTFILES.skip

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Sat, 1 Aug 2015 10:02:32 +0000 (10:02 +0000)]
Updated Portuguese translation

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Sat, 1 Aug 2015 09:55:44 +0000 (09:55 +0000)]
Updated Portuguese translation

10 years agosearchenginetracker: Ignore NULL directories
Benjamin Otte [Sat, 1 Aug 2015 03:49:54 +0000 (05:49 +0200)]
searchenginetracker: Ignore NULL directories

10 years agogtkoverlay: Optimize size allocations
Jasper St. Pierre [Sat, 1 Aug 2015 03:21:20 +0000 (20:21 -0700)]
gtkoverlay: Optimize size allocations

Doing raises / lowers here will invalidate the entire child widget
almost every time, since raising each widget causes it to invalidate
that area.

10 years agoCode cleanup
Matthias Clasen [Fri, 31 Jul 2015 21:57:04 +0000 (17:57 -0400)]
Code cleanup

Use g_slist_free_full more consistently. This commit just converts
the obvious cases where g_slist_forall is directly followed by
g_slist_free.

10 years agoHC: shade spinbuttons as entries
Jakub Steiner [Sat, 1 Aug 2015 00:21:30 +0000 (02:21 +0200)]
HC: shade spinbuttons as entries

https://bugzilla.gnome.org/show_bug.cgi?id=746491

10 years agogtk-demo: Add a scale example
Matthias Clasen [Fri, 31 Jul 2015 20:36:02 +0000 (16:36 -0400)]
gtk-demo: Add a scale example

10 years agofile chooser: Allow activating without double-click
Matthias Clasen [Fri, 31 Jul 2015 11:03:25 +0000 (07:03 -0400)]
file chooser: Allow activating without double-click

Interpret a unmodified primary click on the selection like a double
click. This makes it possible to activate a file or open a folder
without using double-click.

10 years agogesture single: Use gtk_simulate_touchscreen
Matthias Clasen [Fri, 31 Jul 2015 17:45:17 +0000 (13:45 -0400)]
gesture single: Use gtk_simulate_touchscreen

Instead of manual tests.

10 years agotext view: Use gtk_simulate_touchscreen
Matthias Clasen [Fri, 31 Jul 2015 17:44:56 +0000 (13:44 -0400)]
text view: Use gtk_simulate_touchscreen

Instead of repeated manual tests.

10 years agoentry: Use gtk_simulate_touchscreen
Matthias Clasen [Fri, 31 Jul 2015 17:44:11 +0000 (13:44 -0400)]
entry: Use gtk_simulate_touchscreen

Instead of repeated manual checks.

10 years agoAdd a private function to determine if we simulate
Matthias Clasen [Fri, 31 Jul 2015 17:30:05 +0000 (13:30 -0400)]
Add a private function to determine if we simulate

These same two or three tests are currently duplicated
in several places, with separate static variables. Lets
do this in one place.

10 years agoHC: custom color button fixes
Jakub Steiner [Fri, 31 Jul 2015 13:42:34 +0000 (15:42 +0200)]
HC: custom color button fixes

https://bugzilla.gnome.org/show_bug.cgi?id=753051

10 years agoentry completion: Clarify index in documentation
Christian Dywan [Fri, 4 Sep 2009 12:32:00 +0000 (12:32 +0000)]
entry completion: Clarify index in documentation

https://bugzilla.gnome.org/show_bug.cgi?id=579466

10 years agotestfilechooser: Say Save
Matthias Clasen [Fri, 31 Jul 2015 11:12:35 +0000 (07:12 -0400)]
testfilechooser: Say Save

Thats a more typical affirmative action in save mode.

10 years agoUse pango_fc_font_map_config_changed
Matthias Clasen [Fri, 31 Jul 2015 04:50:47 +0000 (00:50 -0400)]
Use pango_fc_font_map_config_changed

Use pango_fc_font_map_confit_changed() instead of
pango_fc_font_map_cache_clear().

https://bugzilla.gnome.org/show_bug.cgi?id=748772

10 years agoGtkMenuButton: explicitly protect against recursion
Carlos Garnacho [Thu, 23 Jul 2015 12:26:25 +0000 (14:26 +0200)]
GtkMenuButton: explicitly protect against recursion

The visibility toggling happening on ::click() relied implicitly
on the popover animation, but breaks on disabled animations. The
recursion happening within gtk_toggle_button_set_active() (which
triggers ::clicked when changing state) makes this vfunc to run
again, inverting the visibility of the popover in result.

Fix this by explicitly checking about recursion, we want the
button to be toggled to the right state, but we don't want the
callback running again.

https://bugzilla.gnome.org/show_bug.cgi?id=752577

10 years agocombobox: Refactor cell view treatment
Benjamin Otte [Tue, 28 Jul 2015 23:12:09 +0000 (01:12 +0200)]
combobox: Refactor cell view treatment

Instead of putting it inside the button manually, put it there properly
by adding a box containing the cell view and the arrow.

Do the same thing in list mode, instead of creating an event box that
tries to behave as a button.

10 years agocombobox: Handle NULLing the cellview on remove simpler
Benjamin Otte [Wed, 29 Jul 2015 20:25:52 +0000 (22:25 +0200)]
combobox: Handle NULLing the cellview on remove simpler

10 years agocombobox: Remove unused variables
Benjamin Otte [Tue, 28 Jul 2015 17:21:15 +0000 (19:21 +0200)]
combobox: Remove unused variables

10 years agocombobox: Create cell view identically
Benjamin Otte [Tue, 28 Jul 2015 16:57:48 +0000 (18:57 +0200)]
combobox: Create cell view identically

... in the two places it can be created.

10 years agofile chooser: Don't crawl recent://
Matthias Clasen [Thu, 30 Jul 2015 18:03:17 +0000 (14:03 -0400)]
file chooser: Don't crawl recent://

This is unnecessary - all the recent files are in the model already.
It also leads to duplicates, since our duplicate filtering is based
on g_file_equal, which does not consider recent:///blabla with
target-uri=/my/example to be the same as file:///my/example.

10 years agowidget-factory: Test busy state
Matthias Clasen [Thu, 30 Jul 2015 14:32:44 +0000 (10:32 -0400)]
widget-factory: Test busy state

Makes it easy to test insensitive state of widgets.i
Also exposes brokenness in the shell busy indicator...

10 years agoHC: no need for green in scales
Jakub Steiner [Thu, 30 Jul 2015 12:01:32 +0000 (14:01 +0200)]
HC: no need for green in scales

10 years agoHC: fix typo
Jakub Steiner [Thu, 30 Jul 2015 11:54:29 +0000 (13:54 +0200)]
HC: fix typo

10 years agoExpand builder menu documentation
Matthias Clasen [Thu, 30 Jul 2015 03:17:43 +0000 (23:17 -0400)]
Expand builder menu documentation

List the supported attributes.

10 years agogtk-demo: Move the transparent example under Overlay
Matthias Clasen [Wed, 29 Jul 2015 23:15:08 +0000 (19:15 -0400)]
gtk-demo: Move the transparent example under Overlay

10 years agogtk-demo: Rename textview demos to "Text View"
Matthias Clasen [Wed, 29 Jul 2015 23:07:56 +0000 (19:07 -0400)]
gtk-demo: Rename textview demos to "Text View"

Fits better with "Tree View".

10 years agogtk-demo: Tweak the markup demo
Matthias Clasen [Wed, 29 Jul 2015 23:05:25 +0000 (19:05 -0400)]
gtk-demo: Tweak the markup demo

10 years agogtk-demo: Add font features to markup demo
Matthias Clasen [Wed, 29 Jul 2015 22:35:56 +0000 (18:35 -0400)]
gtk-demo: Add font features to markup demo

10 years agoGtkTextView: Support font features
Matthias Clasen [Wed, 29 Jul 2015 22:28:56 +0000 (18:28 -0400)]
GtkTextView: Support font features

Add a ::font-features attribute to GtkTextTag, and support
font features when inserting Pango markup into a text buffer.

10 years agoplacesview: don't show mount for Computer row
Georges Basile Stavracas Neto [Wed, 29 Jul 2015 22:37:19 +0000 (19:37 -0300)]
placesview: don't show mount for Computer row

The "Computer" row from places view is an abstract volume
that represents the root of the current partition. As such,
it cannot be mounted or unmounted.

Remove the related item on context menu for Computer row.

10 years agofontchooserwidget: use appropriate model for GtkTreeIter
Christian Hergert [Wed, 29 Jul 2015 19:57:22 +0000 (12:57 -0700)]
fontchooserwidget: use appropriate model for GtkTreeIter

We were previously mixing the model used when filtering with an iter that
has been resolved to the backing model.

This results in both an invalid row index as well as an invalid
iter->stamp.

10 years agoFix oversights in templates
Matthias Clasen [Wed, 29 Jul 2015 19:15:10 +0000 (15:15 -0400)]
Fix oversights in templates

Some of our templates were not setting the domain attribute
on the <interface> element.

10 years agogtk3-demo: Add a markup demo
Matthias Clasen [Wed, 29 Jul 2015 17:03:56 +0000 (13:03 -0400)]
gtk3-demo: Add a markup demo

10 years agoHC: make progressbar labels more visible
Jakub Steiner [Wed, 29 Jul 2015 11:32:34 +0000 (13:32 +0200)]
HC: make progressbar labels more visible

- use the same colro as regular labels, no
need to be more subtle

https://bugzilla.gnome.org/show_bug.cgi?id=752978

10 years agoHC: make pushed columns visible
Jakub Steiner [Wed, 29 Jul 2015 11:26:42 +0000 (13:26 +0200)]
HC: make pushed columns visible

https://bugzilla.gnome.org/show_bug.cgi?id=752977

10 years agoHC: provide explicit treeview styling
Jakub Steiner [Wed, 29 Jul 2015 11:24:11 +0000 (13:24 +0200)]
HC: provide explicit treeview styling

deals with some details like:

- DND stuff
- progressbar
- expanders
- separators

https://bugzilla.gnome.org/show_bug.cgi?id=752975

10 years agoHC: :checked:hover color fix for menuitems
Jakub Steiner [Wed, 29 Jul 2015 10:40:16 +0000 (12:40 +0200)]
HC: :checked:hover color fix for menuitems

https://bugzilla.gnome.org/show_bug.cgi?id=752969

10 years agoHC: make link buttons distinct
Jakub Steiner [Wed, 29 Jul 2015 10:34:44 +0000 (12:34 +0200)]
HC: make link buttons distinct

https://bugzilla.gnome.org/show_bug.cgi?id=752965

10 years agoHC: whitespace for sidebar
Jakub Steiner [Wed, 29 Jul 2015 10:20:28 +0000 (12:20 +0200)]
HC: whitespace for sidebar

- make some room around the sidebar icons

https://bugzilla.gnome.org/show_bug.cgi?id=752960

10 years agogtk3-demo: Add a scrollbar context menu
Matthias Clasen [Wed, 29 Jul 2015 03:48:10 +0000 (23:48 -0400)]
gtk3-demo: Add a scrollbar context menu

Just to demonstrate the capability, not because this is
excellent UI.

10 years agogtk3-demo: Add another overlay demo
Matthias Clasen [Wed, 29 Jul 2015 02:47:07 +0000 (22:47 -0400)]
gtk3-demo: Add another overlay demo

10 years agoplacesview: simplify code
Georges Basile Stavracas Neto [Tue, 28 Jul 2015 14:46:25 +0000 (11:46 -0300)]
placesview: simplify code

Since we started adding a persistent Computer item,
all the code related to the view modes became obsolete,
since the view is never empty anymore. So, drop this
dead code and use the plain stack to manage the empty
search results view.

Also, this patch fixed a very annoying keyboard navigation
issue where we couldn't go from the On This Computer to Networks
lists, because they were two separate widgets. Merge the two
lists into a single one, and update headers accordingly.

10 years agoplacesviewrow: hook up finalize
Georges Basile Stavracas Neto [Tue, 28 Jul 2015 10:41:32 +0000 (07:41 -0300)]
placesviewrow: hook up finalize

It is needed to hook up the finalize method, since
it's overriden manually.

10 years agoplacessidebar: fix ellipsizing glitch
Georges Basile Stavracas Neto [Tue, 28 Jul 2015 10:08:03 +0000 (07:08 -0300)]
placessidebar: fix ellipsizing glitch

The places sidebar rows' labels were missing a proper
xalign, which caused the labels to keep shaking while
the sidebar was resized.

Fix that by setting the label's xalign to 0.

10 years agoplacesview: plug some memory leaks
Georges Basile Stavracas Neto [Tue, 28 Jul 2015 09:51:14 +0000 (06:51 -0300)]
placesview: plug some memory leaks

The GFile containing the Computer item was not
properly dereferenced, so plug that leak by both
dereferencing it and adding some reference management
on GtkPlacesViewRow.

10 years agoplacesview: show Computer item
Georges Basile Stavracas Neto [Mon, 27 Jul 2015 19:52:37 +0000 (16:52 -0300)]
placesview: show Computer item

GtkPlacesView widget manages persistent locations,
factoring out GtkPlacesSidebar functionality.

It, however, does not completely shows all sidebar
locations, since Computer is still missing.

Add a Computer item, adjusting some internal behavior
to make that possible.

10 years agoplacesview: don't show network addresses
Georges Basile Stavracas Neto [Mon, 27 Jul 2015 19:51:23 +0000 (16:51 -0300)]
placesview: don't show network addresses

It is distracting, not relevant and too space
consuming.

Remove the network addresses label.

10 years agoplacesview: fix documentation
Georges Basile Stavracas Neto [Mon, 27 Jul 2015 19:49:18 +0000 (16:49 -0300)]
placesview: fix documentation

The doc for gtk_places_view_get_search_query was being
set twice, while gtk_places_view_set_search_query was
never set.

Fix that by correcting the wrong getter documentation.

10 years agoplacessidebar: fix documentation
Georges Basile Stavracas Neto [Mon, 27 Jul 2015 19:47:40 +0000 (16:47 -0300)]
placessidebar: fix documentation

Use the same explanation of why GtkPlacesSidebar::show-connect-server
was deprecated and its replacements on getter and setter.

Also, fix a mistakenly deprecated function.

10 years agoinspector: Show monitor information
Matthias Clasen [Tue, 28 Jul 2015 05:14:49 +0000 (01:14 -0400)]
inspector: Show monitor information

This is useful to verify the monitor information returned by
other backends.

10 years agoinspector: Show wayland display information
Matthias Clasen [Tue, 28 Jul 2015 04:46:34 +0000 (00:46 -0400)]
inspector: Show wayland display information

Don't call it "X display" when we show useful information under
Wayland as well.

10 years agowayland: Return a display name
Matthias Clasen [Tue, 28 Jul 2015 04:43:54 +0000 (00:43 -0400)]
wayland: Return a display name

Make gdk_screen_make_display_name() return a likely correct
name for the Wayland socket we are using.

10 years agox11: Make selection handling work across screens
Matthias Clasen [Tue, 28 Jul 2015 03:18:27 +0000 (23:18 -0400)]
x11: Make selection handling work across screens

When dealing with selection events, we might see windows from
other screens in the requestor field. The current x11 backend
code fails to wrap these in a foreign GdkWindow, since we
don't have the corresponding GdkScreen anymore. Work around
this by creating such 'foreign screens' on demand. We still
maintain the 1:1 relation between the display and the screen
returned by gdk_display_get_default_screen().

https://bugzilla.gnome.org/show_bug.cgi?id=721398

10 years agoUpdate Catalan translation
Jordi Mas [Mon, 27 Jul 2015 21:03:13 +0000 (23:03 +0200)]
Update Catalan translation

10 years agoPlug a memory leak
Matthias Clasen [Mon, 27 Jul 2015 18:54:13 +0000 (14:54 -0400)]
Plug a memory leak

GtkCssProvider was using the wrong unref function for GtkCssKeyframe
objects. Funnily, this didn't crash at all, the refcount was at the
same place in the struct - it just didn't free all the memory, causing
valgrind to complain.

10 years agoquartz app menu: add hidpi support for menu icons
Christoph Reiter [Fri, 24 Jul 2015 13:28:27 +0000 (15:28 +0200)]
quartz app menu: add hidpi support for menu icons

Use the new cairo to NSImage converter function
to set the device scale.

Remove the pixbuf converter function as this was
the last user.

10 years agoquartz dnd: add hidpi support for gtk_drag_set_icon_surface()
Christoph Reiter [Fri, 24 Jul 2015 13:09:48 +0000 (15:09 +0200)]
quartz dnd: add hidpi support for gtk_drag_set_icon_surface()

Instead of storing the dnd icon as a pixbuf store it
as a cairo surface and transfer the device scale
when converting it to a NSImage. In the pixbuf/stock/named
setters convert to a surface instead (no hidpi support yet)

10 years agoquartz dnd: fix drag icon position relative to cursor position
Christoph Reiter [Fri, 24 Jul 2015 11:57:35 +0000 (13:57 +0200)]
quartz dnd: fix drag icon position relative to cursor position

The y axis is reversed on osx

10 years agoquartz: disable system shadows if a GdkWindow shadow is set.
Christoph Reiter [Sun, 19 Jul 2015 18:13:22 +0000 (20:13 +0200)]
quartz: disable system shadows if a GdkWindow shadow is set.

The shadow will be drawn in the wrong place in those cases so all we can
do is disable it. This fixes double shadows drawn around menus, popups
and tooltips.

https://bugzilla.gnome.org/show_bug.cgi?id=734984

10 years agoplacessidebar: show Computer item when not in Other Locations
Georges Basile Stavracas Neto [Mon, 20 Jul 2015 01:09:29 +0000 (22:09 -0300)]
placessidebar: show Computer item when not in Other Locations

Previous commits removed from places sidebar the code related
to displaying the Computer item, which should be shown when
the sidebar is not displaying the Other Locations item.

Add back the item when the sidebar is not in Other Locations'
mode.

10 years agoquartz: Fix building
Patrick Griffis [Mon, 27 Jul 2015 13:33:32 +0000 (09:33 -0400)]
quartz: Fix building

Broken by 6d04183c5d6 and 6ac07a2e85ea

10 years agofile chooser: Note when searching remote locations
Matthias Clasen [Fri, 24 Jul 2015 20:26:32 +0000 (16:26 -0400)]
file chooser: Note when searching remote locations

Inform the user that we won't search recursively when searching
in remote locations.

10 years agoAvoid more GFile<>uri roundtrips
Matthias Clasen [Fri, 24 Jul 2015 04:40:59 +0000 (00:40 -0400)]
Avoid more GFile<>uri roundtrips

Make GtkSearchHit carry a GFile instead of an uri. Most of the
search engines already have the object around, and converting
to an uri and back is unnecessary extra work.

10 years agosearch engine: Don't crawl remote locations
Matthias Clasen [Fri, 24 Jul 2015 19:52:37 +0000 (15:52 -0400)]
search engine: Don't crawl remote locations

This is slow, and causes too much network I/O.
So skip locations that look like they are remote.

10 years agoAdd a helper function for remote locations
Matthias Clasen [Fri, 24 Jul 2015 20:01:25 +0000 (16:01 -0400)]
Add a helper function for remote locations

Add a helper function that says whether a location should be
considered remote. To determine this, we look at the filesystem
type reported by gvfs, and say 'remote' for sftp, webdav, ftp,
nfs and cifs.

10 years agoActually use search results from the model
Matthias Clasen [Fri, 24 Jul 2015 19:44:20 +0000 (15:44 -0400)]
Actually use search results from the model

We didn't connect to the signals, so we never picked up
any search results from the model engine. Good thing, since
it was returning the wrong thing.

10 years agoPort to GtkSearchHit
Matthias Clasen [Fri, 24 Jul 2015 19:41:45 +0000 (15:41 -0400)]
Port to GtkSearchHit

GtkSearchEngineModel was still trying to return uris as hits.
This is not working anymore, we are expected to return GtkSearchHit
structs.

10 years agoAlways pass a location into the query
Matthias Clasen [Fri, 24 Jul 2015 19:37:03 +0000 (15:37 -0400)]
Always pass a location into the query

Failure to do so was causing search to start from the home dir
when the current folder is not represented on the sidebar.

10 years agoStore locations as GFile
Matthias Clasen [Fri, 24 Jul 2015 19:29:51 +0000 (15:29 -0400)]
Store locations as GFile

It is a bit pointless to have the file chooser get a uri from an
existing GFile to put in the query, only to have some of the search
engines reconstruct a GFile from it.

10 years agoRemove a redundant assignment
Matthias Clasen [Mon, 27 Jul 2015 12:06:32 +0000 (08:06 -0400)]
Remove a redundant assignment

10 years agoUpdated Slovak translation
Dušan Kazik [Mon, 27 Jul 2015 08:21:37 +0000 (08:21 +0000)]
Updated Slovak translation

10 years agoUpdated Greek translation
Dimitris Spingos [Mon, 27 Jul 2015 09:36:37 +0000 (12:36 +0300)]
Updated Greek translation

10 years agocombobox: Use ui file to construct internal button child
Benjamin Otte [Sat, 25 Jul 2015 22:06:46 +0000 (00:06 +0200)]
combobox: Use ui file to construct internal button child

10 years agocombobox: Create button on init() and destroy in destroy()
Benjamin Otte [Sat, 25 Jul 2015 21:19:40 +0000 (23:19 +0200)]
combobox: Create button on init() and destroy in destroy()

The button is the same in both menu and list mode, so there's no need to
destroy and recreate it.

10 years agocombobox: Add scroll support in list mode
Benjamin Otte [Fri, 24 Jul 2015 15:31:25 +0000 (17:31 +0200)]
combobox: Add scroll support in list mode

10 years agocombobox: Remove unused argument
Benjamin Otte [Fri, 24 Jul 2015 15:20:03 +0000 (17:20 +0200)]
combobox: Remove unused argument

The visible check is always FALSE, so no need to keep the code around.

10 years agocombobox: Remove the box from the button
Benjamin Otte [Fri, 24 Jul 2015 15:09:36 +0000 (17:09 +0200)]
combobox: Remove the box from the button

... when in menu mode.

Now that we don't have a separator anymore, the box is not needed
anymore.

10 years agocombobox: Remove the separator
Benjamin Otte [Fri, 24 Jul 2015 15:02:48 +0000 (17:02 +0200)]
combobox: Remove the separator

It was a special case and it's not used by themes anymore.

Plus, in the CSS world, if a theme wants a separator, it can just
style the button's border to create one.

10 years agoRemove a duplicate string
Matthias Clasen [Sun, 26 Jul 2015 06:57:39 +0000 (02:57 -0400)]
Remove a duplicate string

dialog-information was twice in the list of accessible labels
for icon names.

10 years agolistbox: Fix shift-selection
Matthias Clasen [Sun, 26 Jul 2015 06:40:26 +0000 (02:40 -0400)]
listbox: Fix shift-selection

Selecting a range of rows with Shift-click was accidentally broken
in e6e511eb460d8fcd. Make it work again.

10 years agoquartz: Implement NO_TARGET drag result
Matthias Clasen [Sun, 26 Jul 2015 06:10:47 +0000 (02:10 -0400)]
quartz: Implement NO_TARGET drag result

This is necessary to make notebook tab DND work.

https://bugzilla.gnome.org/show_bug.cgi?id=752638

10 years agoquartz: Check if a drag ends inside a known window
Matthias Clasen [Sun, 26 Jul 2015 06:04:48 +0000 (02:04 -0400)]
quartz: Check if a drag ends inside a known window

If a drag ends inside a known window, set the dest_window field
in the drag context. This information is needed to implemented
notebook tab dragging.

https://bugzilla.gnome.org/show_bug.cgi?id=752638

10 years agoGtkMenu: Properly detach menu when the window is detached
Jonas Ådahl [Thu, 23 Jul 2015 06:34:12 +0000 (14:34 +0800)]
GtkMenu: Properly detach menu when the window is detached

Without properly cleaning up GtkMenu private attach state
(GtkMenuAttachData) when the attached widget is freed, we would end up
with an invalid pointer to a freed widget. Trying to detach from that
widget would cause a segmentation fault.

https://bugzilla.gnome.org/show_bug.cgi?id=752761

10 years agowayland: Fix initial fullscreen monitor support
Giovanni Campagna [Sun, 26 Jul 2015 00:22:41 +0000 (17:22 -0700)]
wayland: Fix initial fullscreen monitor support

-1 means that we have no specific preference for an initial
fullscreen monitor, and -1 is less than the number of monitors,
so we would end up accessing invalid memory. Prevent that.

https://bugzilla.gnome.org/show_bug.cgi?id=752875

10 years agoFix a warning
Matthias Clasen [Sun, 26 Jul 2015 05:49:19 +0000 (01:49 -0400)]
Fix a warning

10 years agoUse standard cursor names
Matthias Clasen [Sun, 10 May 2015 02:50:02 +0000 (22:50 -0400)]
Use standard cursor names

This changes GTK+ to use gdk_cursor_new_from_name() with the
'standard' css names, instead of GdkCursorType.

https://bugzilla.gnome.org/show_bug.cgi?id=652085

10 years agowin32: Support all css cursor names
Matthias Clasen [Sun, 26 Jul 2015 05:34:38 +0000 (01:34 -0400)]
win32: Support all css cursor names

Approximate some of the resize cursors with similar cursors.

10 years agobroadway: fix use-after-free on read errors
Peter Wu [Thu, 23 Jul 2015 15:13:54 +0000 (17:13 +0200)]
broadway: fix use-after-free on read errors

Call chain:
 - input_data_cb
   - broadway_server_read_all_input_nonblocking (input)
     - broadway_input_free (input)
       (now input is invalid)
     attempt to use input->active -> use-after-free

Make broadway_server_read_all_input_nonblocking return a boolean, TRUE
if the input was valid, FALSE otherwise. This allows input_data_cb to
detect whether the input was gone or not.

https://bugzilla.gnome.org/show_bug.cgi?id=741685