gtk4.git
10 years agowayland: Make display closing work
Matthias Clasen [Fri, 28 Aug 2015 18:16:21 +0000 (14:16 -0400)]
wayland: Make display closing work

The code in _gdk_wayland_window_dispose was not safe against
being called twice - it would call g_hash_table_destroy twice
on the known_globals hash table, the second time operating on
freed memory. It was also leaking the list of async_roundtrips.

After fixing both of these issues, the displayclose testcase
now works on Wayland.

10 years agowayland: Be more careful when destroying windows
Matthias Clasen [Fri, 28 Aug 2015 17:55:23 +0000 (13:55 -0400)]
wayland: Be more careful when destroying windows

We call gdk_wayland_window_hide_surface when the window gets
destroyed, and in this case, the frame clock might not exist
anymore.

This was showing up in the displayclose testcase.

10 years agoTrivial formatting fix
Matthias Clasen [Fri, 28 Aug 2015 17:51:11 +0000 (13:51 -0400)]
Trivial formatting fix

10 years agowayland: Don't assert non-recursing in gdk_window_destroy
Matthias Clasen [Fri, 28 Aug 2015 17:48:46 +0000 (13:48 -0400)]
wayland: Don't assert non-recursing in gdk_window_destroy

While we do not have subwindows in Wayland, we do create an
artificial root window. When the display is closed, the root
window gets destroyed, causing recursing to be true for the
toplevel windows.

10 years agoTrivial formatting fix
Matthias Clasen [Fri, 28 Aug 2015 17:48:22 +0000 (13:48 -0400)]
Trivial formatting fix

10 years agoUpdated Greek translation
Tom Tryfonidis [Fri, 28 Aug 2015 18:37:44 +0000 (18:37 +0000)]
Updated Greek translation

10 years agoUpdated Greek translation
Tom Tryfonidis [Fri, 28 Aug 2015 18:37:03 +0000 (18:37 +0000)]
Updated Greek translation

10 years agoAdwaita: remove .list-row.activatable reset.
Lapo Calamandrei [Fri, 28 Aug 2015 17:18:35 +0000 (19:18 +0200)]
Adwaita: remove .list-row.activatable reset.

Since being 'activatable' istead of 'button' now that reset
is not needed anymore, the patch is pretty noisy since sass
interpreter changes, those look innocuous though.

10 years agoWork towards succeeding make check under Wayland
Matthias Clasen [Fri, 28 Aug 2015 16:54:32 +0000 (12:54 -0400)]
Work towards succeeding make check under Wayland

Skip tests that don't work under Wayland, such as tests for GtkPlug
and GtkSocket.

10 years agoplug/socket: Warn if used outside X11
Matthias Clasen [Fri, 28 Aug 2015 16:48:02 +0000 (12:48 -0400)]
plug/socket: Warn if used outside X11

Give a clear warning when either of these are realized on a non-X11
screen.

10 years agoMake display test not fail under Wayland
Matthias Clasen [Fri, 28 Aug 2015 16:26:25 +0000 (12:26 -0400)]
Make display test not fail under Wayland

This test really only works under X, since it relies on modifying
the DISPLAY env var.

10 years agoplaces view: Don't spew needlessly
Matthias Clasen [Fri, 28 Aug 2015 16:15:39 +0000 (12:15 -0400)]
places view: Don't spew needlessly

The check here was obviously meant to filter out cancellations,
not warn for them.

10 years agoUpdated Indonesian translation
Andika Triwidada [Fri, 28 Aug 2015 13:22:19 +0000 (13:22 +0000)]
Updated Indonesian translation

10 years agoUpdated German translation
Christian Kirbach [Thu, 27 Aug 2015 22:01:51 +0000 (22:01 +0000)]
Updated German translation

10 years agoUpdated Hebrew translation
Yosef Or Boczko [Thu, 27 Aug 2015 20:19:12 +0000 (23:19 +0300)]
Updated Hebrew translation

10 years agoAdwaita: Make selector more specific
Benjamin Otte [Thu, 27 Aug 2015 18:18:40 +0000 (20:18 +0200)]
Adwaita: Make selector more specific

A * selector applies to all widgets, so even GtkBox or GtkGrid - and
most importantly GtkListBoxRow - need to recompute their style because
of the * selector.

By using a more specific one, these common cases aren't affected
anymore.

Fixes slowdowns in gtk3-demo's listbox demo and in gnome-software.

10 years agocombobox: Add a special class for the button in the combobox
Benjamin Otte [Thu, 27 Aug 2015 17:36:18 +0000 (19:36 +0200)]
combobox: Add a special class for the button in the combobox

That way, the GTK engine doesn't think that the general .button CSS
might potentially apply to it.
And because combobox button is overly complex and stupid, it cannot be
cached.

So buttons thought they cannot ever cache anything because they might
suddenly end up inside a combobox without noticing and then they'd need
to round their corners differently. Of course they're just regular
"Remove" buttons like all the other 100s of "Remove" buttons in
gnome-software. But hey, better not cache anything for them and
recompute their CSS every time the :hover state changes on one of the
rows.

10 years agocss: Add gtk_css_change_to_string()
Benjamin Otte [Thu, 27 Aug 2015 12:07:10 +0000 (14:07 +0200)]
css: Add gtk_css_change_to_string()

Nobody is able to look up those hex values.

10 years agocss: Fix GtkCssChange enum
Benjamin Otte [Thu, 27 Aug 2015 12:06:13 +0000 (14:06 +0200)]
css: Fix GtkCssChange enum

Get rid of unused values (POSITION_ENUM) and use different values for
NAME and ID.

10 years agocssnode: Loosen requirements for style caching
Benjamin Otte [Wed, 26 Aug 2015 16:01:39 +0000 (18:01 +0200)]
cssnode: Loosen requirements for style caching

We can actually share :first-child/:last-child related things now,
because we special case them. So the only positions we cannot cache are
nth-child/nth-last-child.

This should take care of a lot of Adwaita's styling.

10 years agoUpdated Hungarian translation
Gábor Kelemen [Thu, 27 Aug 2015 15:06:37 +0000 (15:06 +0000)]
Updated Hungarian translation

10 years agoHighContrast: GtkListBox fixes
Jakub Steiner [Thu, 27 Aug 2015 09:08:23 +0000 (11:08 +0200)]
HighContrast: GtkListBox fixes

Make flat buttons in list-rows work.

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Thu, 27 Aug 2015 09:09:20 +0000 (09:09 +0000)]
Updated Portuguese translation

10 years agoAdwaita: Theme calendar day names differently
Matthias Clasen [Thu, 27 Aug 2015 02:59:55 +0000 (22:59 -0400)]
Adwaita: Theme calendar day names differently

Just to show how its done. This will need some tweaks.

10 years agoflow box: Avoid a critical in keynav
Matthias Clasen [Wed, 26 Aug 2015 23:57:15 +0000 (19:57 -0400)]
flow box: Avoid a critical in keynav

We were not taking into account that gtk_flow_box_get_next_focusable
can return NULL, in all callers. Fix that.

10 years agotestflowbox: Add an example with buttons
Matthias Clasen [Wed, 26 Aug 2015 23:48:29 +0000 (19:48 -0400)]
testflowbox: Add an example with buttons

This example populates a flow box with buttons, and makes the
flow box children unfocusable, with the intention that the focus
moves directly between the buttons. Currently, keynav does not
work at all in this case.

10 years agogdkgears: Add a popover
Matthias Clasen [Wed, 26 Aug 2015 22:26:56 +0000 (18:26 -0400)]
gdkgears: Add a popover

As pointed out in bug 754143, this reveals a crash in the Wayland
backend.

10 years agoUpdated Polish translation
Piotr Drąg [Wed, 26 Aug 2015 15:09:31 +0000 (17:09 +0200)]
Updated Polish translation

10 years agoplaces view row: Use proper gettext macros
Piotr Drąg [Wed, 26 Aug 2015 14:45:33 +0000 (16:45 +0200)]
places view row: Use proper gettext macros

P_() is meant only for property nicks and blurbs. Everything
else should use _().

Follow-up to 2d83dd62173fc5e651e7df95f331ad5cc46a29cf

http://bugzilla.gnome.org/show_bug.cgi?id=754086

10 years agoinspector: Handle object destruction without row references
Benjamin Otte [Wed, 26 Aug 2015 14:33:52 +0000 (16:33 +0200)]
inspector: Handle object destruction without row references

This way, we can live without row references.

A side effect is that opening the inspector on the gtk-demo list box
example now only takes 0.5s instead of the previous 3 minutes.

10 years agoinspector: Implement find_object() without GtkTreeRowReference
Benjamin Otte [Wed, 26 Aug 2015 13:44:55 +0000 (15:44 +0200)]
inspector: Implement find_object() without GtkTreeRowReference

Tree row references are slow to update. And we update the widget list
quite a bit, so when it grows it can get REALLY slow.

10 years agoinspector: Refactor object tree construction
Benjamin Otte [Wed, 26 Aug 2015 13:28:22 +0000 (15:28 +0200)]
inspector: Refactor object tree construction

Instead of having lots of if statements, create a vtable. This will be
useful for the following patches.

10 years agowindow: Only update the inspector once
Benjamin Otte [Tue, 25 Aug 2015 17:46:09 +0000 (19:46 +0200)]
window: Only update the inspector once

Instead of queueing a new idle handler every time we call
gtk_window_update_debugging(), only queue one if none is queued that.

Saves a lot of work, in particular when templates create context menus
for every row in a large listbox as in the gtk-demo listbox example.

10 years agolistbox: Mark activatable rows with a .activatable style class
Benjamin Otte [Tue, 25 Aug 2015 00:08:29 +0000 (02:08 +0200)]
listbox: Mark activatable rows with a .activatable style class

Do not use .button anymore.

This is for 2 reasons:
1. The styling is seperate in our themes, so it doesn't make sense to
   share the style class.
2. Due to the shared styling of .buton, listbox rows inherit all the
   special case styles that exist for buttons - such as linked buttons,
   header buttons, entry buttons, spinbutton buttons, etc. This means
   that the code has to check all these special cases all the time and
   for listbox rows, this is very slow.

10 years agoUpdated Hungarian translation
Gábor Kelemen [Wed, 26 Aug 2015 12:55:27 +0000 (12:55 +0000)]
Updated Hungarian translation

10 years agoInspector: Protect against childless windows
Timm Bäder [Wed, 26 Aug 2015 09:54:01 +0000 (11:54 +0200)]
Inspector: Protect against childless windows

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Wed, 26 Aug 2015 05:33:35 +0000 (05:33 +0000)]
Updated Portuguese translation

10 years agoplaces view: Use proper gettext macros
Matthias Clasen [Wed, 26 Aug 2015 03:28:56 +0000 (23:28 -0400)]
places view: Use proper gettext macros

P_() is meant only for property nicks and blurbs. Everything
else should use _().

http://bugzilla.gnome.org/show_bug.cgi?id=754086

10 years agoUpdated Chinese (Taiwan) translation
Chao-Hsiung Liao [Tue, 25 Aug 2015 22:41:58 +0000 (22:41 +0000)]
Updated Chinese (Taiwan) translation

10 years agoUpdated Chinese (Taiwan) translation
Chao-Hsiung Liao [Tue, 25 Aug 2015 22:40:15 +0000 (22:40 +0000)]
Updated Chinese (Taiwan) translation

10 years agoflow box: Document incompatible configurations
Matthias Clasen [Tue, 25 Aug 2015 17:58:40 +0000 (13:58 -0400)]
flow box: Document incompatible configurations

Explain in the docs that using a model is incompatible with
using filter or sort functions.

10 years agoflow box: Warn for nonworking configurations
Matthias Clasen [Tue, 25 Aug 2015 17:28:49 +0000 (13:28 -0400)]
flow box: Warn for nonworking configurations

GtkFlowBox sort and filter functionality does not work with
models. Issue a warning in this case.

10 years agolist box: Document incompatible configurations
Matthias Clasen [Tue, 25 Aug 2015 17:20:53 +0000 (13:20 -0400)]
list box: Document incompatible configurations

Explain in the docs that using a model is incompatible with
using filter or sort functions.

http://bugzilla.gnome.org/show_bug.cgi?id=752615

10 years agolist box: Warn for nonworking configurations
Matthias Clasen [Tue, 25 Aug 2015 17:16:24 +0000 (13:16 -0400)]
list box: Warn for nonworking configurations

GtkListBox sort and filter functionality does not work with
models. Issue a warning in this case.

http://bugzilla.gnome.org/show_bug.cgi?id=752615

10 years agoFix a startup ordering problem
Matthias Clasen [Fri, 9 Jan 2015 04:16:39 +0000 (23:16 -0500)]
Fix a startup ordering problem

Defer a11y initialization until we have a display. A11y initialization
causes widget classes to be initalized, which in turn needs some
backend-specific information about modifier masks that can't be
obtained before we have a display.

https://bugzilla.gnome.org/show_bug.cgi?id=736125

10 years agoBug 736125 - The default gtk accelerator mod mask does not include all default backen...
John Ralls [Thu, 25 Dec 2014 00:10:12 +0000 (16:10 -0800)]
Bug 736125 - The default gtk accelerator mod mask does not include all default backend modifiers

10 years agoInspector: Make monitor rows unactivatable
Timm Bäder [Mon, 24 Aug 2015 14:06:08 +0000 (16:06 +0200)]
Inspector: Make monitor rows unactivatable

10 years agoFix grammar in "No networks locations found"
Piotr Drąg [Tue, 25 Aug 2015 15:11:17 +0000 (17:11 +0200)]
Fix grammar in "No networks locations found"

10 years agoUpdate Galician translations
Fran Dieguez [Tue, 25 Aug 2015 14:43:42 +0000 (16:43 +0200)]
Update Galician translations

10 years agoUpdated Galician translations
Fran Dieguez [Tue, 25 Aug 2015 14:29:19 +0000 (16:29 +0200)]
Updated Galician translations

10 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 25 Aug 2015 09:07:47 +0000 (11:07 +0200)]
Updated Spanish translation

10 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 25 Aug 2015 09:07:23 +0000 (11:07 +0200)]
Updated Spanish translation

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Tue, 25 Aug 2015 06:13:59 +0000 (06:13 +0000)]
Updated Portuguese translation

10 years agoUpdated Portuguese translation
Pedro Albuquerque [Tue, 25 Aug 2015 06:10:53 +0000 (06:10 +0000)]
Updated Portuguese translation

10 years agogtk-demo: Update listbox example with longer list
Benjamin Otte [Mon, 24 Aug 2015 19:23:49 +0000 (21:23 +0200)]
gtk-demo: Update listbox example with longer list

Add all 388 tweets of the @GTKtoolkit account. This shows the
performance behavior of the listbox (not good with that many rows) and
allows us to quickly notice when things get worse (or better).

And just so I have a place where I can dump how I generated this file:

First, I got Timm Bäder to download me the json for the twitter feed
into a file gtk.json, then I ran the jq tool on it like this:

jq ".[] | if .retweeted_status then .retweeted_status.user.name + \"|\"
+ .retweeted_status.user.screen_name else .user.name + \"|\" +
.user.screen_name end + \"|\" + .text" gtk.json | cat -n | sed
"s/\\s*\([0-9]*\)\t\"\(.*\)\"/\\1|\\2/" > messages.start

jq ".[] | .created_at" gtk.json | sed "s/\"\(.*\)\"/\1/" | while read
in; do date +%s -d "$in"; done > dates

jq ".[] | \"0|\" + if .retweeted_status then .user.screen_name else \"\"
end + \"|\" + (.favorite_count | tostring) + \"|\" + (.retweet_count |
tostring)" gtk.json | sed "s/\"\(.*\)\"/\\1/" > messages.end

paste -d\| messages.start dates messages.end > messages.txt

This whole machinery of going through 3 intermediate files was only
necessary to onvert the dates from ISO format to unix timestamps,
otherwise this could have been a single line.

10 years agogtk-demo: Order tweets newest first in listbox example
Benjamin Otte [Mon, 24 Aug 2015 19:23:29 +0000 (21:23 +0200)]
gtk-demo: Order tweets newest first in listbox example

10 years agoUpdated Slovak translation
Dušan Kazik [Mon, 24 Aug 2015 16:30:15 +0000 (16:30 +0000)]
Updated Slovak translation

10 years agoUpdated Slovak translation
Dušan Kazik [Mon, 24 Aug 2015 16:08:34 +0000 (16:08 +0000)]
Updated Slovak translation

10 years agofilechooser: check if we can access the parent folder
Paolo Borelli [Sat, 22 Aug 2015 16:20:17 +0000 (18:20 +0200)]
filechooser: check if we can access the parent folder

If we manually enter an unaccessible path in the entry, e.g
"/root/foo.txt", we should receive an error saying that the
folder is not accessible instead of showing the replace
confirmation dialog.

https://bugzilla.gnome.org/show_bug.cgi?id=753969

10 years agoRevert "gtkoverlay: Optimize size allocations"
Matthias Clasen [Sat, 22 Aug 2015 15:22:27 +0000 (11:22 -0400)]
Revert "gtkoverlay: Optimize size allocations"

This reverts commit 93171cf4f346c0a5e71f274d012c3cacfe3880ba.

This broke overlay controls in page 3 of widget-factory, gnome-photos,
gnome-documents, etc.

10 years agolist box: Make focusable headers possible
Matthias Clasen [Sat, 22 Aug 2015 00:51:03 +0000 (20:51 -0400)]
list box: Make focusable headers possible

Previously we were assuming that only list box rows could occur
as focus children of a list box, and would crash if that wasn't
the case. This commit handles this case, and integrates focusable
headers into directional keynav and the focus chain.

The typical case of using separators as headers is not affected
by this change.

https://bugzilla.gnome.org/show_bug.cgi?id=753694

10 years agoGtkOverlay: Update the description
Matthias Clasen [Fri, 21 Aug 2015 22:19:26 +0000 (18:19 -0400)]
GtkOverlay: Update the description

State that the overlays are placed wrt to the GtkOverlay, not
with respect to the main widget. This makes a difference for
small main widgets which are not configured to fill the entire
GtkOverlay.

10 years agoRevert "popover: Optimize the transition"
Matthias Clasen [Fri, 21 Aug 2015 22:08:30 +0000 (18:08 -0400)]
Revert "popover: Optimize the transition"

This reverts commit 84bd32a09a25bee24ae7740e30bd75727695d744.

It turned out that this did break the transition in some places.

10 years agogtkplacesview: bail out if operation cancelled
Carlos Soriano [Fri, 21 Aug 2015 17:57:07 +0000 (19:57 +0200)]
gtkplacesview: bail out if operation cancelled

When an operation is cancelled it's never safe to access
the object itself or the private struct, since it could be
called (and probably is) during finalize.

In case the operation is cancelled, just bail out to fix
the crashes.

10 years agoUpload Catalan translator
Jordi Mas [Fri, 21 Aug 2015 16:21:06 +0000 (18:21 +0200)]
Upload Catalan translator

10 years agogtkplacessidebar: order mounts by label
Carlos Soriano [Fri, 21 Aug 2015 13:00:07 +0000 (15:00 +0200)]
gtkplacessidebar: order mounts by label

We were not ordering them at all, and therefore they were
jumping around

https://bugzilla.gnome.org/show_bug.cgi?id=753925

10 years agogtkplacesview: use g_utf8_collate for sorting
Carlos Soriano [Fri, 21 Aug 2015 12:49:11 +0000 (14:49 +0200)]
gtkplacesview: use g_utf8_collate for sorting

Also rename the variable, since we usually use location
for GFiles variables.

10 years agogtkplacesview: add a loading property
Carlos Soriano [Wed, 19 Aug 2015 12:46:34 +0000 (14:46 +0200)]
gtkplacesview: add a loading property

So clients of the view can know if the view is busy.

https://bugzilla.gnome.org/show_bug.cgi?id=753786

10 years agogtkplacesview: improve networks feedback
Carlos Soriano [Tue, 18 Aug 2015 21:26:07 +0000 (23:26 +0200)]
gtkplacesview: improve networks feedback

Add a spinner when networks are being fetched and make
the network section permanent and show a placeholder with
a message that no networks were found in case there are no
networks. In this way users from previous versions won't be
confused with the fact that no networks are shown.

https://bugzilla.gnome.org/show_bug.cgi?id=753786

10 years agogtkplacesview: add networks in network:///
Carlos Soriano [Tue, 18 Aug 2015 15:58:41 +0000 (17:58 +0200)]
gtkplacesview: add networks in network:///

Previously we had a network item in the sidebar, which now
is replaced by the network section on other-locations view.
However we were not exposing the networks in network:///.

Fetch them and add them in the network section of other-locations
view.

https://bugzilla.gnome.org/show_bug.cgi?id=753786

10 years agogtkplacesview: style fix
Carlos Soriano [Tue, 18 Aug 2015 07:43:48 +0000 (09:43 +0200)]
gtkplacesview: style fix

https://bugzilla.gnome.org/show_bug.cgi?id=753786

10 years agoUpdated Slovak translation
Dušan Kazik [Fri, 21 Aug 2015 08:55:24 +0000 (08:55 +0000)]
Updated Slovak translation

10 years agowayland: Only bind supported unstable pointer gestures global
Jonas Ådahl [Thu, 20 Aug 2015 03:38:43 +0000 (11:38 +0800)]
wayland: Only bind supported unstable pointer gestures global

As the protocol is still considered unstable (meaning not backward
compatible), we should, as stated in the protocol, only bind the version
advertised is the version we implement.

https://bugzilla.gnome.org/show_bug.cgi?id=753856

10 years agooverlay: Always align widgets relative to the overlay
Benjamin Otte [Thu, 20 Aug 2015 22:46:35 +0000 (00:46 +0200)]
overlay: Always align widgets relative to the overlay

... not to the contained widget, which might not occupy the whole
overlay when an align mode is set.

10 years agooverlay: Use actual size of the overlay when positioning
Benjamin Otte [Thu, 20 Aug 2015 22:44:17 +0000 (00:44 +0200)]
overlay: Use actual size of the overlay when positioning

Previously we were using the natural requested size which is unrelated.

10 years agoUpdated Lithuanian translation
Aurimas Černius [Thu, 20 Aug 2015 20:01:41 +0000 (23:01 +0300)]
Updated Lithuanian translation

10 years agotreeview: check for pixel cache before setting property
Christian Hergert [Thu, 20 Aug 2015 18:13:30 +0000 (11:13 -0700)]
treeview: check for pixel cache before setting property

Code exists in the wild that calls this function after the widget has
been destroyed (and the pixel cache released). Simply check that the
pixel cache exists to preserve the existing state.

10 years agoUpdate Catalan translation
Jordi Mas [Thu, 20 Aug 2015 17:04:10 +0000 (19:04 +0200)]
Update Catalan translation

10 years agoimmodule: Remove unused parameter
Benjamin Otte [Tue, 18 Aug 2015 18:45:34 +0000 (20:45 +0200)]
immodule: Remove unused parameter

10 years agoUpdated POTFILES.skip
Piotr Drąg [Thu, 20 Aug 2015 11:01:14 +0000 (13:01 +0200)]
Updated POTFILES.skip

10 years agogtk-demo: Add a model button demo
Matthias Clasen [Thu, 20 Aug 2015 03:17:01 +0000 (23:17 -0400)]
gtk-demo: Add a model button demo

This shows how to hook model buttons up to various action types,
or how to create them manually.

10 years agoGtkTextView: fix margin/padding scroll behaviour
Sebastien Lafargue [Wed, 19 Aug 2015 10:33:06 +0000 (12:33 +0200)]
GtkTextView: fix margin/padding scroll behaviour

When you move line by line, only padding is
automaticly shown and you need to use Page key to show margin.

This commit also fix cursor going out of the screen bug.

10 years agogdkevent: Hold refs to device/source_device
Carlos Garnacho [Mon, 3 Aug 2015 15:22:29 +0000 (17:22 +0200)]
gdkevent: Hold refs to device/source_device

The extra reference will be held from GdkEventPrivate data, so there's
a common place to all events. Without this, events queued after devices/
capabilities disappear (eg. on TTY switch) might hold invalid pointers.
Windowing level operations on those devices (queries, grabs...) are
expected to fail at that time, but we should hold meaningful data for
the regular event handling paths.

https://bugzilla.gnome.org/show_bug.cgi?id=753185

10 years agogtkscrolledwindow: Fold kinetic deceleration handling into scroll_event()
Carlos Garnacho [Wed, 19 Aug 2015 16:45:50 +0000 (18:45 +0200)]
gtkscrolledwindow: Fold kinetic deceleration handling into scroll_event()

In order to play along with child widgets that use scroll events for anything
else than scrolling, it will be better to do this in the bubble phase, so
the child widget has an opportunity to GDK_EVENT_STOP the event before we
trigger kinetic scrolling.

This of course won't work for widgets that choose to reimplement scroll event
handling themselves, they should be smart at resorting to GtkScrolledWindow's
scroll event handling.

This fixes kinetic scrolling kicking in too pervasively on widgets that eg.
implement zoom on scroll events.

https://bugzilla.gnome.org/show_bug.cgi?id=753495

10 years agoUpdated Hungarian translation
Gábor Kelemen [Wed, 19 Aug 2015 10:07:21 +0000 (10:07 +0000)]
Updated Hungarian translation

10 years agoplacessidebar: set Other Locations uri
Georges Basile Stavracas Neto [Tue, 18 Aug 2015 22:34:21 +0000 (19:34 -0300)]
placessidebar: set Other Locations uri

By assigning an URI to Other Locations item, we
can programaticaly select it. Fixes a bug in Nautilus,
where the Other Locations item is unselected imediately
after being clicked.

10 years agomultipress: Only emit preedit signals if needed
Matthias Clasen [Tue, 18 Aug 2015 19:42:56 +0000 (15:42 -0400)]
multipress: Only emit preedit signals if needed

When clearing the compose buffer, don't emit preedit-changed
unconditionally. This triggers a recursion which leads to
a crash.

https://bugzilla.gnome.org/show_bug.cgi?id=753644

10 years agoprint operation: let subclasses use a custom paginate
Paolo Borelli [Mon, 17 Aug 2015 16:31:15 +0000 (18:31 +0200)]
print operation: let subclasses use a custom paginate

GtkPrintOperation was emitting paginate only if a signal was
connected, this meant that subclassing and overriding the
paginate vfunc lead to the unexpected result that paginate did
not run.
Instead we always emit the signal and use a custom accumulator:
if there is a signal we just run that and avoid the default
handler, otherwise we run the default handler which can be the
one by the subclass or the default handler that just skips
pagination.

Patch by Yevgen Muntyan, fixes #345345

10 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 18 Aug 2015 11:56:26 +0000 (13:56 +0200)]
Updated Spanish translation

10 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 18 Aug 2015 11:56:16 +0000 (13:56 +0200)]
Updated Spanish translation

10 years ago3.17.7
Matthias Clasen [Tue, 18 Aug 2015 02:09:03 +0000 (22:09 -0400)]
3.17.7

10 years agowayland: Use g_get_prgname() to get the xdg_surface.set_app_id string
Jonas Ådahl [Tue, 18 Aug 2015 00:38:07 +0000 (08:38 +0800)]
wayland: Use g_get_prgname() to get the xdg_surface.set_app_id string

Prior to this patch, the ID of the GtkApplication was always used for
clients which were GtkApplications. This would only be guaranteed to be
correct for D-Bus activatable programs. As a result, all
non-D-Bus-activatable applications would set the wrong ID making the
shell unable to find the corresponding .desktop file.

This change makes it so that the GDK backend always uses the name
passed to g_set_prgname, or the default value if not explicitly set, as
this more often corresponds to the .desktop file.

This means that in order to make D-Bus activatable applications set the
correct application ID, they must, for now, manually call
g_set_prgname() with their application ID (basename of the .desktop
file).

If g_get_prgname() returns NULL, fallback to gdk_get_program_class()
even though it will most likely never be correct according to the
xdg_surface.set_app_id specification.

https://bugzilla.gnome.org/show_bug.cgi?id=746435

10 years agogtkplacesview: allow to cancel connection to server
Carlos Soriano [Mon, 17 Aug 2015 22:42:23 +0000 (00:42 +0200)]
gtkplacesview: allow to cancel connection to server

We were not allowing to cancel the operation at all, and at
most the operation was cancelled only when clicked connect again.

Also due to gvfs bug 753735 we actually weren't cancelling
at all, and therefore creating multiple dialogs.

10 years agoplaces-view: show a more verbose entry hint
Georges Basile Stavracas Neto [Mon, 17 Aug 2015 17:25:18 +0000 (14:25 -0300)]
places-view: show a more verbose entry hint

As pointed out by Allan, this is not very helpful. A
call to action label fits better in this case.

10 years agoflowbox: Sink the floating reference of model-based children
Emmanuele Bassi [Mon, 17 Aug 2015 12:32:35 +0000 (13:32 +0100)]
flowbox: Sink the floating reference of model-based children

We don't want to leak references if the widget created to represent the
item in the model does not have a floating reference — which is usually
what happens in bindings, as they automatically sink references when
creating new instances.

See commit 6e03e7e8 for the similar change in GtkListBox.

10 years agoMSVC Introspection Builds: Remove GCC Requirement
Chun-wei Fan [Mon, 17 Aug 2015 11:20:04 +0000 (19:20 +0800)]
MSVC Introspection Builds: Remove GCC Requirement

Sorry, the last commit added a generated file instead of the
template.

G-I has been updated to not require a Windows GCC installation
anymore to generate the .gir files, so update the NMake Makefiles
that are used for this purpose.

As a result, it is no longer necessary to define time_t for the .gir
generation as we are on the same compiler throughout the process.

10 years agoRevert "MSVC Introspection Builds: Remove GCC Requirement"
Chun-wei Fan [Mon, 17 Aug 2015 11:18:03 +0000 (19:18 +0800)]
Revert "MSVC Introspection Builds: Remove GCC Requirement"

This reverts commit aea7809f55ac520b9d2d5d9c1eea1345fabbd7a7.

10 years agoMSVC Introspection Builds: Remove GCC Requirement
Chun-wei Fan [Mon, 17 Aug 2015 11:11:53 +0000 (19:11 +0800)]
MSVC Introspection Builds: Remove GCC Requirement

G-I has been updated to not require a Windows GCC installation
anymore to generate the .gir files, so update the NMake Makefiles
that are used for this purpose.

As a result, it is no longer necessary to define time_t for the .gir
generation as we are on the same compiler throughout the process.

10 years agogtk-demo: Use top/bottom-margin for text views
Matthias Clasen [Mon, 17 Aug 2015 05:36:37 +0000 (01:36 -0400)]
gtk-demo: Use top/bottom-margin for text views

This lets us replace the hack of inserting tags in the first
and last paragraph to set a above/below-paragraph space.

10 years agotext view: add padding and margin
Sebastien Lafargue [Tue, 23 Dec 2014 20:30:24 +0000 (21:30 +0100)]
text view: add padding and margin

Add API to set top and bottom margins, and add the CSS padding
from the theme to the programmatically set margins.

https://bugzilla.gnome.org/show_bug.cgi?id=406159