This spawns a mutter WM for the xvfb instance and wraps it in its own
dbus session.

Also use a larger screen size (1080p) - mostly so we can check the
testuite runs with differently sized screens, but alspo because that's
today's default screen size.

This exposes the issue from #4136 in the testsuite.

Duplicate the X11 testsuite into x11full and x11bare.
Run both of them in the Fedora CI runner.

For now they do the same thing (followup commits will change that), but
the goal is to have 2 testruns:

x11full should get as close to simulating a modern full-featured setup
with a compositing WM and all extensions.

x11bare should have the minimum required setup that te testsuite needs,
but nothing more. No compositor, a WM wit the minimal set of features,
etc.

mutter will be needed to get a WM for the x11 testsuite.

lcms2 is needed for upcoming HDR work, so having it available in
varioius branches is neat.

GdkWin32: Plug memory leak

See merge request GNOME/gtk!4747

nodeparser: Fix parsing of color-matrix node

See merge request GNOME/gtk!4745

Negative offsets in the color matrix were inverted because it used the
rect parser.

CI: Include reftest nodes in artifacts

See merge request GNOME/gtk!4743

Allow selectable labels contents to be selected by mnemonic

See merge request !4741

expecially by the mnemonic of another label that targets it.

GNOME/gtk#4927

see also https://bugs.documentfoundation.org/show_bug.cgi?id=137748

checkbutton: fix critical when setting use-underline

Closes #4934

See merge request GNOME/gtk!4740

This needs to make sure that we've created the label before applying the
setting on a child widget.

Fixes #4934

testsuite: Add tests for label overdraw

See merge request GNOME/gtk!4737

This brings back a subset of what quit-mnemonic.ui tested for, but
trying a lot harder to trigger the label overdrawing its allocation,
which will cause the text to be cut off when clipping is happening.

It should not be an issue at all with GTK4, but keeping that test around
is a good idea.

gsk/gl: use critical instead of assert in dispose

Closes #4625

See merge request GNOME/gtk!4736

Instead of asserting only in debug builds (which are generally not
shipped in distributions) we should deliver a critical log-level message
so that these can be found sooner when not developing with jhbuild,
Flatpak, etc.

Also assert that we've setup the state correctly when realizing the
GskGLRenderer object.

Fixes #4625

l10n: Remove po/pot for dropped nicks/blurbs

See merge request GNOME/gtk!4718

Add an explicit harfbuzz dependency

See merge request GNOME/gtk!4732

We are using Harfbuzz api in the file chooser,
so lets make this official.

GdkWin32: Add support for DirectManipulation

See merge request GNOME/gtk!4698

Adds support for Precision TouchPads (PTPs) gestures
and scroll events.

Danial Behzadi authored May 15, 2022 and Administrator committed May 15, 2022

(cherry picked from commit 2b1c09dd)

ci: Update the Fedora image to Fedora 36

See merge request GNOME/gtk!4730

I can't quite figure out what this test was meant
to test, and how to make it do so in a way that
does not fall afoul of rendering issues in the GL
renderer and rounding differences in pango.

Can't win with reftests.

Also remove systemtap from the image, since
we don't need it.

The v36 image is still used in gtk-4-6.

listitemmanager: Clarify warning

See merge request !4731

Lars doesn't have an opinion on this problem anymore.

And when people see this warning, they should know what to do.

Cosmetic fixes to objcopy hack

See merge request !4729

Preserve the elf section name that we would get
without the objcopy hack. Minimizing the differences

Anders Jonsson authored May 14, 2022 and Administrator committed May 14, 2022

(cherry picked from commit 2d8a8e35)

gsk: Serialize textures as tiff

See merge request !4725