gtk4.git
3 years agoMerge branch 'sumibi-yakitori/fix-maximize-window-macos' into 'main'
Christian Hergert [Sun, 3 Apr 2022 22:14:52 +0000 (22:14 +0000)]
Merge branch 'sumibi-yakitori/fix-maximize-window-macos' into 'main'

macos: prohibit fullscreen transition if in transtion

This prevents performing additional fullscreen transitions while
a transition is already in progress.

Closes #4808

See merge request GNOME/gtk!4612

3 years agomacos: prohibit fullscreen transition if in transtion
sumibi-yakitori [Sun, 3 Apr 2022 22:14:52 +0000 (22:14 +0000)]
macos: prohibit fullscreen transition if in transtion

This prevents performing additional fullscreen transitions while
a transition is already in progress.

3 years agoMerge branch 'matthiasc/for-main' into 'main'
Matthias Clasen [Sun, 3 Apr 2022 20:11:29 +0000 (20:11 +0000)]
Merge branch 'matthiasc/for-main' into 'main'

inspector: Avoid a crash

See merge request GNOME/gtk!4617

3 years agoinspector: Avoid another crash
Matthias Clasen [Sun, 3 Apr 2022 19:53:52 +0000 (15:53 -0400)]
inspector: Avoid another crash

We need to handle all event types here.
This was tripping over GDK_TOUCHPAD_HOLD events.

3 years agoinspector: Avoid a crash
Matthias Clasen [Sun, 3 Apr 2022 19:53:34 +0000 (15:53 -0400)]
inspector: Avoid a crash

Attribute lists can be NULL, it turns out.

3 years agoMerge branch 'matthiasc/for-main' into 'main'
Matthias Clasen [Sun, 3 Apr 2022 14:21:13 +0000 (14:21 +0000)]
Merge branch 'matthiasc/for-main' into 'main'

gsk: Plug a memory leak

See merge request GNOME/gtk!4616

3 years agoMerge branch 'meson_fixes' into 'main'
Matthias Clasen [Sun, 3 Apr 2022 12:41:14 +0000 (12:41 +0000)]
Merge branch 'meson_fixes' into 'main'

meson: Use proper type for bools

See merge request GNOME/gtk!4615

3 years agocss: Plug a memory leak
Matthias Clasen [Sun, 3 Apr 2022 12:04:35 +0000 (08:04 -0400)]
css: Plug a memory leak

We were leaking the terms of calc values. Oops.

3 years agomeson: Use proper type for bools
illiliti [Sun, 3 Apr 2022 11:55:33 +0000 (14:55 +0300)]
meson: Use proper type for bools

Fix invalid usage of bools which violates official meson specification and thus
breaks muon, an implementation of meson written in C.

3 years agogsk: Plug a memory leak
Matthias Clasen [Sun, 3 Apr 2022 11:34:09 +0000 (07:34 -0400)]
gsk: Plug a memory leak

This was introduced in 9defc7fc647.

3 years agoMerge branch 'matthiasc/for-main' into 'main'
Matthias Clasen [Sat, 2 Apr 2022 19:47:03 +0000 (19:47 +0000)]
Merge branch 'matthiasc/for-main' into 'main'

imcontextsimple: Plug a memory leak

See merge request GNOME/gtk!4614

3 years agoimcontextsimple: Plug a memory leak
Matthias Clasen [Sat, 2 Apr 2022 19:22:12 +0000 (15:22 -0400)]
imcontextsimple: Plug a memory leak

3 years agoMerge branch 'selection-model-docs' into 'main'
Emmanuele Bassi [Fri, 1 Apr 2022 11:20:11 +0000 (11:20 +0000)]
Merge branch 'selection-model-docs' into 'main'

docs: Fix links in selection models

See merge request GNOME/gtk!4609

3 years agodocs: Fix links in selection models
Maximiliano Sandoval R [Fri, 1 Apr 2022 11:04:14 +0000 (13:04 +0200)]
docs: Fix links in selection models

3 years agoUpdate Indonesian translation
Andika Triwidada [Fri, 1 Apr 2022 03:56:36 +0000 (03:56 +0000)]
Update Indonesian translation

3 years agoMerge branch 'clipboard-seg' into 'main'
Matthias Clasen [Fri, 1 Apr 2022 00:56:32 +0000 (00:56 +0000)]
Merge branch 'clipboard-seg' into 'main'

x11: Check return of gdk_x11_get_xatom_name_for_display

See merge request GNOME/gtk!4607

3 years agoMerge branch 'fix-win32-empty-clipboard' into 'main'
Matthias Clasen [Fri, 1 Apr 2022 00:56:02 +0000 (00:56 +0000)]
Merge branch 'fix-win32-empty-clipboard' into 'main'

gdkclipboard-win32.c: Fix call to gdk_clipboard_claim_remote() (check GdkContentFormat is not NULL

Closes #4796

See merge request GNOME/gtk!4604

3 years agoMerge branch 'nls' into 'main'
Matthias Clasen [Thu, 31 Mar 2022 17:07:37 +0000 (17:07 +0000)]
Merge branch 'nls' into 'main'

Remove #ifdef ENABLE_NLS

See merge request GNOME/gtk!4606

3 years agoMerge branch 'wip/carlosg/immulticontext-display-switch' into 'main'
Matthias Clasen [Thu, 31 Mar 2022 16:29:38 +0000 (16:29 +0000)]
Merge branch 'wip/carlosg/immulticontext-display-switch' into 'main'

gtkimmulticontext: Handle switches between displays

Closes #4805

See merge request GNOME/gtk!4605

3 years agoMerge branch 'msys2-ci-cleanup' into 'main'
Christoph Reiter [Thu, 31 Mar 2022 15:46:54 +0000 (15:46 +0000)]
Merge branch 'msys2-ci-cleanup' into 'main'

CI: clean up MSYS2 build dependencies

See merge request GNOME/gtk!4600

3 years agoRemove #ifdef ENABLE_NLS
Xavier Claessens [Thu, 31 Mar 2022 15:41:34 +0000 (11:41 -0400)]
Remove #ifdef ENABLE_NLS

libintl API is guaranteed to always be available, glib will fallback to
proxy-libintl in case gettext is not found.

3 years agox11: Check return of gdk_x11_get_xatom_name_for_display
Dr. David Alan Gilbert [Thu, 31 Mar 2022 14:56:09 +0000 (15:56 +0100)]
x11: Check return of gdk_x11_get_xatom_name_for_display

When given an invalid atom, gdk_x11_get_xatom_name_for_display can
return NULL and trigger a seg in gdk_x11_clipboard_formats_from_atoms.
Check for NULL.

Why I'm seeing a bad atom there is probably a separate question.
https://bugzilla.redhat.com/show_bug.cgi?id=2037786

3 years agogtkimmulticontext: Handle switches between displays
Carlos Garnacho [Thu, 31 Mar 2022 14:29:34 +0000 (16:29 +0200)]
gtkimmulticontext: Handle switches between displays

Currently the GtkIMMultiContext may stick to a delegate GtkIMContext
that no longer applies after the multicontext is dissociated from
any widget.

Handle set_client_widget() so that it can handle changes between
widgets from 2 different display, but also so the delegate is made
NULL whenever the context has a NULL widget.

Doing so, any new client widget results in a new delegate IM context
lookup from the right GdkDisplay and GtkSettings, which avoids any
mix up.

Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4805
3 years agogdkclipboard-win32.c: Fix call to gdk_clipboard_claim_remote()
Chun-wei Fan [Thu, 31 Mar 2022 02:36:45 +0000 (10:36 +0800)]
gdkclipboard-win32.c: Fix call to gdk_clipboard_claim_remote()

The call to gdk_win32_clipboard_request_contentformats() can return NULL even
without an error condition being hit (such as when the system clipboard is
empty), so check whether the returned GdkContentFormat pointer is not NULL
before calling gdk_clipboard_claim_remote(), which expects it to be not NULL,
otherwise we face a warning from that funtion and the subsequent
g_object_unref().

This at least partially fixes issue #4796.

3 years agoMerge branch 'wip/exalm/scroll-hold' into 'main'
Carlos Garnacho [Mon, 28 Mar 2022 14:27:55 +0000 (14:27 +0000)]
Merge branch 'wip/exalm/scroll-hold' into 'main'

eventcontrollerscroll: Always propagate hold events

See merge request GNOME/gtk!4599

3 years agoeventcontrollerscroll: Always propagate hold events
Alexander Mikhaylenko [Sun, 27 Mar 2022 21:05:23 +0000 (01:05 +0400)]
eventcontrollerscroll: Always propagate hold events

Otherwise a stray scroll controller may prevent others from getting hold
events, even if it always propagates scroll events and does absolutely
nothing.

3 years agoUpdate Latvian translation
Rūdolfs Mazurs [Sun, 27 Mar 2022 20:19:10 +0000 (20:19 +0000)]
Update Latvian translation

(cherry picked from commit 3b50f2e8b996f92f49d2974ccff2b160905e8e73)

3 years agoUpdate Latvian translation
Rūdolfs Mazurs [Sun, 27 Mar 2022 19:53:51 +0000 (19:53 +0000)]
Update Latvian translation

(cherry picked from commit d4dd7969d64c85984d6dd1e3535347b1669ecfff)

3 years agoCI: clean up MSYS2 build dependencies
Christoph Reiter [Sun, 27 Mar 2022 17:08:36 +0000 (19:08 +0200)]
CI: clean up MSYS2 build dependencies

We only need a C compiler and not the whole toolchain,
and gst-plugins-bad was split into libraries and plugins.
pkg-config -> pkgconf.

This should speed the CI setup up a bit.

3 years agoMerge branch 'antoniof-main-patch-34986' into 'main'
Emmanuele Bassi [Sun, 27 Mar 2022 16:25:50 +0000 (16:25 +0000)]
Merge branch 'antoniof-main-patch-34986' into 'main'

overlaylayout: Set position style class on child

Closes nautilus#2099

See merge request GNOME/gtk!4597

3 years agooverlaylayout: Set position style class on child
António Fernandes [Sun, 27 Mar 2022 14:59:10 +0000 (14:59 +0000)]
overlaylayout: Set position style class on child

As documented:
> Overlay children whose alignments cause them to be positioned
> at an edge get the style classes “.left”, “.right”, “.top”,
> and/or “.bottom” according to their position.

Likely accidental regression in b7ee2cbc289b4ab8a950cd77bdcd69a6f13932a7

Fixes https://gitlab.gnome.org/GNOME/nautilus/-/issues/2099

3 years agoMerge branch 'inspector-attributes' into 'main'
Matthias Clasen [Fri, 25 Mar 2022 13:03:00 +0000 (13:03 +0000)]
Merge branch 'inspector-attributes' into 'main'

inspector: Allow viewing PangoAttrList properties

See merge request GNOME/gtk!4596

3 years agoinspector: Allow viewing PangoAttrList properties
Matthias Clasen [Fri, 25 Mar 2022 12:35:16 +0000 (08:35 -0400)]
inspector: Allow viewing PangoAttrList properties

We have pango_attr_list_to/from_string, so this is
easy. The editing UI isn't ideal, but it solves my
immediate need to see attributes.

3 years agoMerge branch 'main' into 'main'
Matthias Clasen [Fri, 25 Mar 2022 11:48:28 +0000 (11:48 +0000)]
Merge branch 'main' into 'main'

Check for 'rst2man' misses installed 'rst2man.py' (#4728)

See merge request GNOME/gtk!4586

3 years agoMerge branch 'wip/chergert/fix-gl-rgba' into 'main'
Benjamin Otte [Fri, 25 Mar 2022 01:34:15 +0000 (01:34 +0000)]
Merge branch 'wip/chergert/fix-gl-rgba' into 'main'

gdk/gl: handle GL_RGBA/GL_UNSIGNED_NORMALIZED

Closes #4783

See merge request GNOME/gtk!4594

3 years agogdk/gl: handle GL_RGBA/GL_UNSIGNED_NORMALIZED
Christian Hergert [Thu, 24 Mar 2022 03:20:02 +0000 (20:20 -0700)]
gdk/gl: handle GL_RGBA/GL_UNSIGNED_NORMALIZED

WebKit's GTK 4 port can give us textures with an internal format of
GL_RGBA with GL_UNSIGNED_NORMALIZED and a bit-depth of 8. This fixes
warnings for every GdkGLTexture created/delivered to the GskGLRenderer.

The format is essentially the same as GL_RGBA8 since it is normalized
between 0.0..1.0 for 8-bit components.

Fixes #4783

3 years agoMerge branch 'want_prepare_without_play_for_size' into 'main'
Matthias Clasen [Thu, 24 Mar 2022 19:19:01 +0000 (19:19 +0000)]
Merge branch 'want_prepare_without_play_for_size' into 'main'

GtkMediaStream with gstreamer backend isn't 'prepared' until media is played

See merge request GNOME/gtk!4560

3 years agoApply 1 suggestion(s) to 1 file(s)
Matthias Clasen [Thu, 24 Mar 2022 17:16:11 +0000 (17:16 +0000)]
Apply 1 suggestion(s) to 1 file(s)

3 years agoUpdate British English translation
Bruce Cowan [Thu, 24 Mar 2022 13:19:20 +0000 (13:19 +0000)]
Update British English translation

(cherry picked from commit dd198cfc065f616d0729e1e02bf5773085dcd014)

3 years agoUpdate Indonesian translation
Kukuh Syafaat [Wed, 23 Mar 2022 15:39:17 +0000 (15:39 +0000)]
Update Indonesian translation

3 years agoUpdate Indonesian translation
Kukuh Syafaat [Wed, 23 Mar 2022 13:46:25 +0000 (13:46 +0000)]
Update Indonesian translation

3 years agoMerge branch 'fix-popover-menu-rtl-position' into 'main'
Matthias Clasen [Wed, 23 Mar 2022 02:16:41 +0000 (02:16 +0000)]
Merge branch 'fix-popover-menu-rtl-position' into 'main'

popover-menu: Fix buttons' position in RTL

See merge request GNOME/gtk!4587

3 years agoUpdate Chinese (Taiwan) translation
Cheng-Chia Tseng [Tue, 22 Mar 2022 14:47:50 +0000 (14:47 +0000)]
Update Chinese (Taiwan) translation

3 years agoMerge branch 'wip/chergert/reduce-overlay-overhead' into 'main'
Matthias Clasen [Tue, 22 Mar 2022 13:27:54 +0000 (13:27 +0000)]
Merge branch 'wip/chergert/reduce-overlay-overhead' into 'main'

Default: avoid use of opacity for overlay scrollbars

See merge request GNOME/gtk!4590

3 years agoMerge branch 'fix-high-depth-switch' into 'main'
Matthias Clasen [Tue, 22 Mar 2022 12:58:07 +0000 (12:58 +0000)]
Merge branch 'fix-high-depth-switch' into 'main'

surface: Use correct display when destroying a surface for depth switch

Closes #4773

See merge request GNOME/gtk!4591

3 years agosurface: Use correct display when destroying a surface for depth switch
Sebastian Keller [Tue, 22 Mar 2022 07:11:18 +0000 (08:11 +0100)]
surface: Use correct display when destroying a surface for depth switch

When surface depth switches from non-high-depth to high-depth (or vice
versa) the current surface has to be destroyed before a new one can be
created for this window. eglDestroySurface however was getting passed a
GdkDisplay, rather than the EGLDisplay it expects. As a result the old
surface did not get destroyed and the new surface could not be created
causing rendering to freeze.

Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4773
3 years agoDefault: avoid use of opacity for overlay scrollbars
Christian Hergert [Tue, 22 Mar 2022 01:54:09 +0000 (18:54 -0700)]
Default: avoid use of opacity for overlay scrollbars

If using the opacity CSS property the renderer cannot optimize these
handles without the use of offscreens due to the use of both a border
and rgb render node.

Instead, we can apply the alpha to the color values and get the same
effect in a way that the GL renderer can optimize without the use of
offscreen textures for a sizeable reduction in runtime overhead.

3 years agoUpdated Czech translation
Marek Černocký [Mon, 21 Mar 2022 13:57:35 +0000 (14:57 +0100)]
Updated Czech translation

3 years agoUpdated Czech translation
Marek Černocký [Mon, 21 Mar 2022 11:14:36 +0000 (12:14 +0100)]
Updated Czech translation

3 years agopopover-menu: Fix buttons' position in RTL
Yosef Or Boczko [Sat, 19 Mar 2022 22:43:57 +0000 (00:43 +0200)]
popover-menu: Fix buttons' position in RTL

The radio/check/previous buttons shows in
the wrong place in RTL. Fix it.

#4641

Signed-off-by: Yosef Or Boczko <yoseforb@gmail.com>
3 years agoCheck for 'rst2man' misses installed 'rst2man.py' (#4728)
aneejit1 [Sat, 19 Mar 2022 10:41:32 +0000 (10:41 +0000)]
Check for 'rst2man' misses installed 'rst2man.py' (#4728)

By default, 'docutils' installs 'rst2man' as 'rst2man.py'. Amend the
check for 'rst2man' to look for 'rst2man.py' as well if 'rst2man' is
not found.

3 years agoUpdate German translation
Philipp Kiemle [Sat, 19 Mar 2022 12:32:20 +0000 (12:32 +0000)]
Update German translation

(cherry picked from commit d9c39f67954921eb2dab4cb5d044eacd4ee4297d)

3 years agoMerge branch 'wip/carlosg/fix-accumulated-velocity' into 'main'
Carlos Garnacho [Sat, 19 Mar 2022 12:02:56 +0000 (12:02 +0000)]
Merge branch 'wip/carlosg/fix-accumulated-velocity' into 'main'

kinetic scroll fixes

See merge request GNOME/gtk!4572

3 years agogtkkineticscrolling: Do not take distance based shortcuts
Carlos Garnacho [Tue, 15 Mar 2022 21:42:13 +0000 (22:42 +0100)]
gtkkineticscrolling: Do not take distance based shortcuts

The pixel distance could be small enough between tick() calls that
this kind of checks might potentially become a problem. Rely only on
the calculated velocity to trigger the STOPPED phase, and use a lower
threshold to avoid cutting the animation too early.

Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/4725

3 years agogtkscrolledwindow: Do not trigger kinetic helpers on 0 velocity
Carlos Garnacho [Tue, 15 Mar 2022 21:41:27 +0000 (22:41 +0100)]
gtkscrolledwindow: Do not trigger kinetic helpers on 0 velocity

Doing this is pointless, so it could be skipped.

3 years agogtkscrolledwindow: Change lifetime of kinetic scroll helpers
Carlos Garnacho [Tue, 15 Mar 2022 21:01:36 +0000 (22:01 +0100)]
gtkscrolledwindow: Change lifetime of kinetic scroll helpers

In order to properly accumulate scroll velocities, we need to keep
the kinetic scroll helpers after we have possibly stopped them
in the process of initiating a further scroll flick.

So, instead of stopping (and destroying) those helpers on scroll-begin,
keep them until the next scroll-end if a scroll was initiated before
kinetic scroll finished. This way we can fetch the last velocity when
calculating the extra kick.

In order to ensure the helpers don't live beyond what it is expected,
we now also remove them after a finished hold event.

Fixes the accumulation of scrolling velocity on consecutive scroll
sequences.

3 years agogtkscrolledwindow: Refactor kinetic scroll animation
Carlos Garnacho [Tue, 15 Mar 2022 20:59:12 +0000 (21:59 +0100)]
gtkscrolledwindow: Refactor kinetic scroll animation

Do not depend on the kinetic scroll helpers existing or not before
exiting the animation, as we may want to keep those a little bit
longer after stopped.

3 years agogtkkineticscroll: Do not reset velocity after stop()
Carlos Garnacho [Tue, 15 Mar 2022 20:52:35 +0000 (21:52 +0100)]
gtkkineticscroll: Do not reset velocity after stop()

We may want to fetch the last velocity obtained, even though we
preemptively called stop() on a kinetic scroll helper. Keep this
velocity so it can be queried later on.

3 years agoMerge branch 'search-delay-versions' into 'main'
Emmanuele Bassi [Sat, 19 Mar 2022 10:14:04 +0000 (10:14 +0000)]
Merge branch 'search-delay-versions' into 'main'

Fix GtkSearchEntry:search-delay docs

See merge request GNOME/gtk!4585

3 years agoAdd version marker to GtkSearchEntry:search-delay property
Sebastian Dröge [Sat, 19 Mar 2022 09:56:05 +0000 (11:56 +0200)]
Add version marker to GtkSearchEntry:search-delay property

3 years agoFix typo in gtk_search_entry_set_search_delay() docs causing it to not show up
Sebastian Dröge [Sat, 19 Mar 2022 09:55:03 +0000 (11:55 +0200)]
Fix typo in gtk_search_entry_set_search_delay() docs causing it to not show up

3 years agoMerge branch 'macos-fix-scroll-unit-build' into 'main'
Matthias Clasen [Sat, 19 Mar 2022 03:57:48 +0000 (03:57 +0000)]
Merge branch 'macos-fix-scroll-unit-build' into 'main'

macos: fix scroll-unit build

See merge request GNOME/gtk!4584

3 years agomacos: fix scroll-unit build
panoplie [Sat, 19 Mar 2022 01:55:02 +0000 (02:55 +0100)]
macos: fix scroll-unit build

3 years agoMerge branch 'scroll-unit' into 'main'
Carlos Garnacho [Sat, 19 Mar 2022 00:47:08 +0000 (00:47 +0000)]
Merge branch 'scroll-unit' into 'main'

GdkScrollEvent: indicate the delta unit with a new GdkScrollUnit enum

See merge request GNOME/gtk!4508

3 years agoMerge branch 'wip/carlosg/rewritten-events-from-other-toplevels' into 'main'
Matthias Clasen [Sat, 19 Mar 2022 00:34:17 +0000 (00:34 +0000)]
Merge branch 'wip/carlosg/rewritten-events-from-other-toplevels' into 'main'

gtk/main: Make coords out of surface when rewriting events for grabs

Closes #4760

See merge request GNOME/gtk!4566

3 years agoMerge branch 'wip/chergert/gsk-gl-texture-library' into 'main'
Matthias Clasen [Fri, 18 Mar 2022 23:46:17 +0000 (23:46 +0000)]
Merge branch 'wip/chergert/gsk-gl-texture-library' into 'main'

gsk/gl: texture libraries, shader creation

See merge request GNOME/gtk!4583

3 years agogtkeventcontrollerscroll: Map surface scroll unit in discrete steps
panoplie [Thu, 24 Feb 2022 22:36:58 +0000 (23:36 +0100)]
gtkeventcontrollerscroll: Map surface scroll unit in discrete steps

3 years agorecorder: Show scroll events deltas unit
panoplie [Tue, 22 Feb 2022 13:12:44 +0000 (14:12 +0100)]
recorder: Show scroll events deltas unit

3 years agogtkrange: Add scroll unit handling
panoplie [Mon, 28 Feb 2022 20:45:06 +0000 (21:45 +0100)]
gtkrange: Add scroll unit handling

3 years agogtkscrolledwindow: Add scroll unit handling
panoplie [Mon, 21 Feb 2022 15:24:00 +0000 (16:24 +0100)]
gtkscrolledwindow: Add scroll unit handling

On the "scroll" signal, the widget uses
gtk_event_controller_scroll_get_unit() to get the
scroll unit.

When the unit is GDK_SCROLL_UNIT_WHEEL, the
behavior is unchanged: the widget scrolls a
certain number of pixels at each wheel detent
click. This number of pixels is determined by the
window dimensions in get_wheel_detent_scroll_step().

When the delta unit is GDK_SCROLL_UNIT_SURFACE, the
widget directly adds the delta to the number of
scrolled pixels no matter the window dimensions.

3 years agogtkeventcontrollerscroll: Add scroll unit getter
panoplie [Mon, 21 Feb 2022 15:23:24 +0000 (16:23 +0100)]
gtkeventcontrollerscroll: Add scroll unit getter

The scroll unit is accessible through
gtk_event_controller_scroll_get_unit() after the
"scroll" signal reception.

3 years agogdk: Add enum to indicate the unit of scroll deltas
panoplie [Mon, 21 Feb 2022 15:22:57 +0000 (16:22 +0100)]
gdk: Add enum to indicate the unit of scroll deltas

Add a new GdkScrollUnit enum that represent the
unit of scroll deltas provided by GdkScrollEvent.
The unit is accessible through
gdk_scroll_event_get_unit().

3 years agogsk/gl: use consistent library naming
Christian Hergert [Fri, 18 Mar 2022 21:55:13 +0000 (14:55 -0700)]
gsk/gl: use consistent library naming

3 years agogsk/gl: make texture libraries more autonomous
Christian Hergert [Fri, 18 Mar 2022 21:51:41 +0000 (14:51 -0700)]
gsk/gl: make texture libraries more autonomous

This moves a lot of the texture atlas control out of the driver and into
the various texture libraries through their base GskGLTextureLibrary class.

Additionally, this gives more control to libraries on allocating which can
be necessary for some tooling such as future Glyphy integration.

As part of this, the 1x1 pixel initialization is moved to the Glyph library
which is the only place where it is actually needed.

The compact vfunc now is responsible for compaction and it allows for us
to iterate the atlas hashtable a single time instead of twice as we were
doing previously.

The init_atlas vfunc is used to do per-library initialization such as
adding a 1x1 pixel in the Glyph cache used for coloring lines.

The allocate vfunc purely allocates but does no upload. This can be useful
for situations where a library wants to reuse the allocator from the
base class but does not want to actually insert a key/value entry. The
glyph library uses this for it's 1x1 pixel.

In the future, we will also likely want to decouple the rectangle packing
implementation from the atlas structure, or at least move it into a union
so that we do not allocate unused memory for alternate allocators.

3 years agogsk/gl: pin atlases to single texture library
Christian Hergert [Fri, 18 Mar 2022 19:48:43 +0000 (12:48 -0700)]
gsk/gl: pin atlases to single texture library

This removes the sharing of atlases across various texture libraries. Doing
so is necessary so that atlases can have different semantics for how they
allocate within the texture as well as potentially allowing for different
formats of texture data.

For example, in the future we might store non-pixel data in the textures
such as Glyphy or even keep glyphs with color content separate from glyphs
which do not and can use alpha channel only.

3 years agogsk/gl: add more control over shader generation
Christian Hergert [Thu, 17 Mar 2022 23:50:51 +0000 (16:50 -0700)]
gsk/gl: add more control over shader generation

This allows the gskglprograms.defs a bit more control over how a shader
will get generated and if it needs to combine sources. Currently, none of
the built-in shaders do that, but upcoming shaders which come from external
libraries will need the ability to inject additional sources in-between
layers.

3 years agogsk/gl: rename glyphs to glyphs_library
Christian Hergert [Thu, 17 Mar 2022 23:48:54 +0000 (16:48 -0700)]
gsk/gl: rename glyphs to glyphs_library

This naming style is less likely to collide with shader naming and makes
it clear where it is consumed what it is.

3 years agogsk/gl: allow configuring atlas size
Christian Hergert [Thu, 17 Mar 2022 23:43:58 +0000 (16:43 -0700)]
gsk/gl: allow configuring atlas size

3 years agogsk/gl: check for format as well
Christian Hergert [Mon, 14 Mar 2022 23:22:52 +0000 (16:22 -0700)]
gsk/gl: check for format as well

This could potentially happen if a uniform had never been set.

3 years agogsk/gl: only clear glyph cache durign reclaimation
Christian Hergert [Tue, 15 Mar 2022 22:55:30 +0000 (15:55 -0700)]
gsk/gl: only clear glyph cache durign reclaimation

We don't need to clear the front cache on every frame as we can clear it
specifically when we do reclaimation to avoid unnecessary memset() calls.

3 years agogsk/gl: ignore max_entry_size when zero
Christian Hergert [Tue, 15 Mar 2022 22:53:47 +0000 (15:53 -0700)]
gsk/gl: ignore max_entry_size when zero

If the max_entry_size is zero, then assume we can add anything to the
atlas. This allows for situations where we might be uploading an arc list
to the atlas instead of pixel data for GPU font rendering.

3 years agogsk/gl: make max-frame-age configurable
Christian Hergert [Tue, 15 Mar 2022 22:52:44 +0000 (15:52 -0700)]
gsk/gl: make max-frame-age configurable

This is nice for some texture libraries that we might want to keep around
for longer than say 60 frames such as a glyph cache.

3 years agogtk-demo: Don't hardcode a title font
Matthias Clasen [Fri, 18 Mar 2022 04:59:57 +0000 (00:59 -0400)]
gtk-demo: Don't hardcode a title font

We want a large font size, but we don't have to
hardcode Sans.

3 years agoMerge branch 'filefilter-suffixes' into 'main'
Matthias Clasen [Thu, 17 Mar 2022 16:57:09 +0000 (16:57 +0000)]
Merge branch 'filefilter-suffixes' into 'main'

filefilter: Fix <suffixes> in buildable

See merge request GNOME/gtk!4581

3 years agofilefilter: Fix <suffixes> in buildable
James Westman [Thu, 17 Mar 2022 15:49:39 +0000 (10:49 -0500)]
filefilter: Fix <suffixes> in buildable

A bug in GtkFileFilter's GtkBuildable implementation caused the
<suffixes> tag not to be recognized.

3 years agoUpdate Italian translation
Milo Casagrande [Thu, 17 Mar 2022 08:35:42 +0000 (08:35 +0000)]
Update Italian translation

(cherry picked from commit 3e4bfa2bae9c6e90005b57d9f9fe47d2301f0e0e)

3 years agoUpdate Italian translation
Milo Casagrande [Thu, 17 Mar 2022 08:34:04 +0000 (08:34 +0000)]
Update Italian translation

(cherry picked from commit 9bbf09fb0a9dbd8ae2c9e64a842138f182681d14)

3 years agoUpdate Serbian translation
Мирослав Николић [Thu, 17 Mar 2022 06:04:15 +0000 (06:04 +0000)]
Update Serbian translation

(cherry picked from commit c874f65d95beef98c99297f05bbed64e7e932f02)

3 years agoMerge branch 'wip/carlosg/im-wayland-serials' into 'main'
Carlos Garnacho [Wed, 16 Mar 2022 18:52:24 +0000 (18:52 +0000)]
Merge branch 'wip/carlosg/im-wayland-serials' into 'main'

gtk/imwayland: Use serial to control outbound messages

Closes #3641

See merge request GNOME/gtk!4398

3 years agogtk/imwayland: Use serial to control outbound messages
Carlos Garnacho [Tue, 18 Jan 2022 22:31:09 +0000 (23:31 +0100)]
gtk/imwayland: Use serial to control outbound messages

Following the text-input protocol changes at
https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/115,
use the serial number to have the client push changes to the
zwp_text_input_v3 object only after compositor/client states match.

This specifically is more lenient to compositors pushing multiple
.done events ahead of the client replying to them.

Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/3641
3 years agoMerge branch 'wip/carlosg/im-text-location' into 'main'
Matthias Clasen [Wed, 16 Mar 2022 11:54:26 +0000 (11:54 +0000)]
Merge branch 'wip/carlosg/im-text-location' into 'main'

gtkimcontextwayland: Add native surface offset to input coordinates

Closes #4668

See merge request GNOME/gtk!4573

3 years agogtkimcontextwayland: Add native surface offset to input coordinates
Carlos Garnacho [Wed, 16 Mar 2022 08:55:20 +0000 (09:55 +0100)]
gtkimcontextwayland: Add native surface offset to input coordinates

We were missing the surface offset (e.g. shadows) at the time of expressing
the text caret location in surface coordinates. Add this offset so the
coordinates are as expected by the compositor.

Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4668
3 years agoMerge branch 'gtk/gss-search-delay' into 'main'
Matthias Clasen [Tue, 15 Mar 2022 22:40:54 +0000 (22:40 +0000)]
Merge branch 'gtk/gss-search-delay' into 'main'

searchentry: Make search delay editable

Closes #4133

See merge request GNOME/gtk!4563

3 years agosearchentry: Make search delay editable
Ondřej Míchal [Fri, 11 Mar 2022 00:15:49 +0000 (02:15 +0200)]
searchentry: Make search delay editable

The default search delay of 150ms can be too low at times[0], leading app
developers to add additional delay while handling the search-changed
signal[1].

Based on past work from hugsie[2].

Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/4133

[0] https://github.com/getting-things-gnome/gtg/issues/281
[1] https://github.com/getting-things-gnome/gtg/pull/587
[2] https://gitlab.gnome.org/GNOME/gtk/-/issues/678

3 years agoUpdate Norwegian Bokmål translation
Kjartan Maraas [Tue, 15 Mar 2022 15:26:33 +0000 (15:26 +0000)]
Update Norwegian Bokmål translation

(cherry picked from commit 72a557087cd824f3e90c9b22ba05a0bef833a6a6)

3 years agoMerge branch 'main' into 'main'
Matthias Clasen [Tue, 15 Mar 2022 14:37:14 +0000 (14:37 +0000)]
Merge branch 'main' into 'main'

gtkplacessidebar: Prevent calling g_object_unref on null

See merge request GNOME/gtk!4569

3 years agoApply 1 suggestion(s) to 1 file(s)
Matthias Clasen [Tue, 15 Mar 2022 14:35:09 +0000 (14:35 +0000)]
Apply 1 suggestion(s) to 1 file(s)

3 years agoMerge branch 'wip/jimmac/symbolics-gnome-42-sync' into 'main'
Matthias Clasen [Tue, 15 Mar 2022 14:29:27 +0000 (14:29 +0000)]
Merge branch 'wip/jimmac/symbolics-gnome-42-sync' into 'main'

icons: update symbolics

See merge request GNOME/gtk!4571

3 years agogtk/main: Fix handling of !owner_events grabs
Carlos Garnacho [Tue, 15 Mar 2022 13:32:10 +0000 (14:32 +0100)]
gtk/main: Fix handling of !owner_events grabs

These are meant to always redirect events to the grabbing surface,
even for other surfaces of the same client. We weren't doing that
(instead letting the event go through unmodified), fix this handling
so GTK sees the events consistenty.

3 years agogtk/main: Make coords out of surface when rewriting events for grabs
Carlos Garnacho [Mon, 14 Mar 2022 12:29:11 +0000 (13:29 +0100)]
gtk/main: Make coords out of surface when rewriting events for grabs

If a grab is held on a toplevel surface tree, and events happen on a
different surface tree from another toplevel/window group, we rewrite
these events so they look like generated on the window group that
holds the grab, but it missed that coordinates would fail to be
translated, so these would stay unchanged and "pointing" to random
parts of the toplevel that is holding the grab and handling the events.

Since off-surface coordinates are not specially meaningful, and in
fact impossible to obtain in some backends, just fake the coordinates
making it sure that all rewritten events point outside the surface.

The grabbing window will still handle the events, but the coordinates
in these will be harmlessly moot.

Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4760