Debian Patches

Status for gtk4/4.22.2+ds-1

Patch Description Author Forwarded Bugs Origin Last update
debian/reftest_compare_surfaces-Report-how-much-the-images-diffe.patch reftest_compare_surfaces: Report how much the images differ
In unattended/non-interactive/autobuilder environments where the images
are not trivially accessible, this provides a way to distinguish between
totally different rendering and more subtle issues.		 Simon McVittie <smcv@debian.org> yes 2024-08-31
insttests/Revert-build-Drop-the-install-tests-option.patch Revert "build: Drop the install-tests option"
Debian and Ubuntu uses these installed tests as part of their
autopkgtest system

This reverts commit 3121f88265ac61733e257f5335473d6f139f158c.		 Jeremy Bicha <jeremy.bicha@canonical.com> no 2023-01-12
insttests/Revert-testsuite-Remove-leftover-test.in-files.patch Revert "testsuite: Remove leftover test.in files"
This reverts commit b598f21506c31d3913c9705a00eb1e8270e9a810.		 Jeremy Bícha <jeremy.bicha@canonical.com> no 2024-12-06
workarounds/reftests-Allow-minor-differences-to-be-tolerated.patch reftests: Allow minor differences to be tolerated
Based on an earlier patch by Michael Biebl, as used in Debian's GTK 3
packaging, with additional inspiration from librsvg's reftests.

Each .ui or .node reftest can have an accompanying .keyfile file
like this:

[reftest]
tolerated-diff-level=20
tolerated-diff-pixels=1000

If the image differs, but the number of pixels that differ is no more
than tolerated-diff-pixels and the differences are no more than
tolerated-diff-level, then we treat it as a success with warnings, save
the .diff.png for analysis, and use g_test_incomplete() to record the
test-case as "TODO".		 Simon McVittie <smcv@debian.org> yes 2021-02-13
workarounds/Disable-inscription-markup.ui-reftest.patch Disable inscription-markup.ui reftest
It's rendered 1px to the left of where it should be, which is not a
serious problem for practical use of GTK, but is a test failure.		 Simon McVittie <smcv@debian.org> not-needed upstream 2022-08-12
workarounds/tests-Mark-gltexture-as-expected-to-fail-on-big-endian-ma.patch tests: Mark gltexture as expected to fail on big-endian machines Simon McVittie <smcv@debian.org> not-needed 2023-08-22
workarounds/tests-Allow-longer-for-a-dialog-to-open.patch tests: Allow longer for a dialog to open
This hopefully makes the test pass more reliably, especially on riscv64.		 Simon McVittie <smcv@debian.org> invalid debian upstream 2023-08-22
workarounds/nodeparser-Adjust-test-for-pango-1.52.0.patch nodeparser: Adjust test for pango 1.52.0
Adjust test result for changed output resulting from
https://gitlab.gnome.org/GNOME/pango/-/commit/d489fc8c4		 Jeremy Bícha <jeremy.bicha@canonical.com> yes 2024-03-09
workarounds/testsuite-skip-color-mix.patch testsuite: skip color-mix
https://gitlab.gnome.org/GNOME/gtk/-/issues/6907		 Jeremy Bícha <jeremy.bicha@canonical.com> no 2024-08-06
workarounds/scaling-test-Skip-floating-point-pixel-formats-with-Cairo.patch scaling test: Skip floating-point pixel formats with Cairo renderer
If the four chosen pixel values happen to have alpha values that average
to 16/17, GTK will quantize that into 8bpp integer space as
16/17*255 = 240, but Cairo/Pixman does its quantization differently and
gets 16/17*256 = 241.		 Simon McVittie <smcv@debian.org> yes upstream 2024-09-03
workarounds/nodeparser-Mark-failing-tests-on-s390x.patch nodeparser: Mark failing tests on s390x Alessandro Astone <alessandro.astone@canonical.com> not-needed upstream 2025-08-18
x11-scale/gdk-x11-check-surface-scale-on-input-region-opaque-region.patch gdk/x11: check surface scale on input region, opaque region/rect
Since in x11 backend the input region, opaque region and opaque rect are
related to scale, it is necessary to track whether the scale is
different so these regions can be updated without changing the logical
size.		 Yao Wei <medicalwei@gmail.com> yes upstream 2025-12-19
x11-scale/gdk-x11-update-correct-shadow-size-according-to-window-si.patch gdk/x11: update correct shadow size according to window size change
Because update_shadow_size is not updated since the logical size of
shadow is not changed, the shadow is glitched out.

This makes sure that the shadow size is recalculated when the scale
is changed.		 "Yao Wei (魏銘廷)" <yao.wei@canonical.com> yes upstream 2025-10-03
x11-scale/gdk-x11-update-cursor-size-on-the-extent-after-scale-chan.patch gdk/x11: update cursor size on the extent after scale change
This will update the cursor size when the scale is changed.
When gnome-shell under Xorg is started in 2x, scaling the UI to
1x the cursor size is correctly scaled to 1x, but it does not
work the other way round.		 "Yao Wei (魏銘廷)" <yao.wei@canonical.com> yes upstream 2025-10-03
x11-touch/xi2-Start-drag-grab-with-pointer-only-event-mask.patch xi2: Start drag grab with pointer-only event mask
For a touch drag we prefer usin the emulated pointer events. Indeed,
gdk_x11_drag_handle_event ignores any native touch events.

However, the initial device grab was created with an event mask that still
allowed touch events, so the server would keep sending native touch events
while the application would stop responding to touch events.

The device grab added in gdk_x11_drag_update_cursor would correctly set an
event mask for pointer-only events, but we stop handling inputs before ever
reaching that point.

We thus change the initial grab event mask to only accept pointer events.		 Alessandro Astone <alessandro.astone@canonical.com> yes 2025-09-25
x11-touch/xi2-Do-not-discard-emulated-pointer-events-during-drag-an.patch xi2: Do not discard emulated pointer events during drag-and-drop
The filter for XIPointerEmulated XI_Motion events, which was meant to
filter-out emulated scroll events, was also inadvertently filtering-out
the emulated drag-and-drop motion updates.		 Alessandro Astone <alessandro.astone@canonical.com> yes 2025-09-19
x11-touch/xi2-Expose-a-logical-touch-device.patch xi2: Expose a logical touch device
Create a new GdkDevice used for reporting all touch events from the XI2
Virtual Core Pointer. It tracks the same XI2 pointer device, but reports
to have source GDK_SOURCE_TOUCHSCREEN.

Add a GdkX11SeatXI2 class extending GdkSeatDefault to manage it.		 Alessandro Astone <alessandro.astone@canonical.com> yes upstream 2025-09-18
printdialog-Keep-GTask-alive-for-portal-repsonse.patch printdialog: Keep GTask alive for portal repsonse
We might return/unref the GTask before the portal response.
The portal response handler uses the GTask data so it must be kept alive.		 Alessandro Astone <alessandro.astone@canonical.com> yes 2026-03-11
testsuite-Don-t-build-waylandsocket-test-if-Wayland-is-di.patch testsuite: Don't build waylandsocket test if Wayland is disabled
Debian builds a copy of GTK for the Debian installer, which is a
constrained X11-only environment (currently using GTK 2 but hopefully
upgrading to GTK 3 or 4 in future).		 Simon McVittie <smcv@debian.org> no 2026-03-15
print-Fix-listing-printers-with-synchronous-backends.patch print: Fix listing printers with synchronous backends
By getting the current list of printers first, and only later calling
request_printer_list(), we are assuming that request_printer_list() itself
will never synchronously add new printers. The cpdb backend is mostly, if not
entirely, synchronous so this assumption breaks.		 Alessandro Astone <alessandro.astone@canonical.com> yes 2026-03-20
accessibility-Fix-regression.patch accessibility: Fix regression
When using GtkApplication with windows attached, the top element
gets a "filler" role instead of a "frame", and there are some
name-calculation issues due to a wrong naming. This results in
accessibility being broken, unable to detect when the focus has
changed from a container to another and, thus, screen readers
won't say the container label/role/description when switching
from a widget in a container to a different widget in another
container.

All this is caused because in commit https://gitlab.gnome.org/GNOME/gtk/-/commit/ef2da952330aa6f1cbc5121caf8ba90cf637c837 ,
the method gtk_at_context_realize() is called inside
gtkapplication-wayland.c before the gtk_widget_realize_at_context()
method in gtkwidget.c is called. This results in an incomplete
widget initialization.

This patch creates a new read-only property to notify when an
ATContext has been realized. With this, we can now ensure that
the code is called only when the ATContext is realized.

This backport to debian includes dropping the unused REALIZED signal,
which was later forwarded to upstream as well: https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/9746		 Sergio Costas Rodriguez <sergio.costas@canonical.com> yes upstream https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/9722 2026-03-24

All known versions for source package 'gtk4'

Links