Bug#1107734: unblock (pre-approval): mutter/48.3.1-2
Control: tags -1 confirmed moreinfo
On 2025-06-13 09:25:46 +0100, Simon McVittie wrote:
> Package: release.debian.org
> Severity: normal
> X-Debbugs-Cc: mutter@packages.debian.org
> Control: affects -1 + src:mutter
> User: release.debian.org@packages.debian.org
> Usertags: unblock
>
> [ Reason ]
> New upstream bug fix release
Please go ahead and remove the moreinfo tag once the package is
available in unstable.
>
> [ Impact ]
> If not accepted we'll be missing several known bug fixes (memory leaks,
> performance fixes, behaviour fixes, etc.) and it will be harder to do
> subsequent stable updates.
>
> I'm also hoping that some of the fixes in the new upstream release will
> address, or at least help, the flakiness of mutter's test suite and
> autopkgtests (#1093969) - but that's difficult to determine because it's
> always been a probabilistic thing, with the failures being most frequent
> on slower architectures that don't run autopkgtests on experimental
> packages.
>
> [ Tests ]
> A functionally equivalent package has been in experimental for a while.
> I've been running that on my GNOME laptop since it was uploaded without
> observing any problems.
>
> There is also autopkgtest coverage, which passed so far.
>
> [ Risks ]
> High-visibility key package, a core part of our default desktop
> environment.
>
> This is from an upstream maintenance branch, and they have a schedule
> for maintenance branch updates (the next one will be in early August).
> In my experience this upstream has generally been good at avoiding
> regressions. All changes are targeted and can be reverted if necessary.
>
> There is one known bug related to the changes, mutter#4033 upstream:
> the proposed version improves tracking of drag-and-drop cursors in some
> situaions (ensuring that the cursor changes between pointer, drag hand,
> or "cannot drop" when appropriate), but this is known not to work for
> older versions of GTK (most commonly from Flatpak or Snap apps that use
> an outdated runtime). A fix has been proposed upstream, but not yet
> reviewed or merged. If I'm reading correctly, 48.1 in trixie already
> has this bug, so it is not a regression in the proposed version.
ACK. This issue can be fied in a later update (if upstream comes up with
a fix).
Cheers
>
> [ Checklist ]
> [x] all changes are documented in the d/changelog
> [x] I reviewed all changes and I approve them
> [x] attach debdiff against the package in testing
> - debdiff is testing vs. experimental, what I'm proposing is
> the same plus a changelog update. I've excluded the changelog
> from the debdiff and pasted a draft changelog below.
>
> --------
>
> mutter (48.3.1-2) UNRELEASED; urgency=medium
>
> * Add a detailed list of changes to the previous changelog entry
>
> -- Simon McVittie <smcv@debian.org> Sun, 08 Jun 2025 17:55:09 +0100
>
> mutter (48.3.1-1) experimental; urgency=medium
>
> * New upstream bugfix release
> - Improve performance of use-cases like fullscreen video playback
> on older GPUs such as the one in the Pinebook Pro, by substituting
> opaque dmabuf formats for their equivalents with alpha, such as
> XRGB for ARGB (mutter#2471 upstream)
> - Allocate the correct height for members of a vertical layout by
> updating size request mode when changing orientation
> (mutter!4339 upstream)
> - Only check for xkbfile when supporting X11
> (no practical effect on Debian, where we do support X11)
> (mutter!4409 upstream)
> - Fix several memory leaks, while ensuring objects are still destroyed
> in the correct order
> (mutter!4432 upstream)
> - Don't log a warning when files are removed from the ICC profile
> directory, fixing one source of flaky automated tests
> (mutter!4431 upstream)
> - Don't take control of logind session when running on Xorg, fixing the
> tty being left in an inconsistent state if mutter is killed
> (mutter!4391 upstream)
> - Fix gdctl crash when communicating with Xorg sessions
> (mutter#4023 upstream)
> - Avoid unexpectedly low framerate when moving/resizing windows after
> session locking or after heavy system load
> (mutter#4029 upstream)
> - Fix drag and drop of tabs into tiled/maximized state
> (mutter!4468 upstream)
> - Fix Xwayland windows becoming unresponsive to events
> (mutter!4475 upstream)
> - Use frame-sync candidate that likely matches scanout candidate, making
> variable refresh rate work better
> (mutter!4412 upstream)
> - Add an early-return from build_and_scan_frame_mask() for zero-sized
> textures, possibly fixing one source of flaky automated tests
> (mutter!4039 upstream)
> - Avoid leaking the session manager file's file descriptor down to
> apps launched by gnome-shell
> (gnome-shell#8345, mutter!4405 upstream)
> - Restore tiled/maximized windows to the correct monitor
> (mutter#3939 upstream)
> - Let the compositor control drag-and-drop cursor feedback for native
> Wayland
> (mutter!4390 upstream)
> - Related issue mutter#4033 upstream, mainly affecting older GTK
> versions like those in outdated Flatpak and Snap runtimes, is not
> fixed here; see mutter!4433 upstream for a possible solution
> - Propagate all touchpad gesture events to apps
> (mutter#3505 upstream)
> - Fix a race condition when registering an implict grab
> (mutter#4055 upstream)
> - Fix Wine games failing to enter fullscreen across multiple monitors
> (mutter!4389 upstream)
> - Update window decorations on override-redirect attribute change
> (mutter!4287 upstream)
> - In the test suite, fall back to a reasonable pipewire socket name on
> non-systemd systems
> (mutter#4044 upstream)
> - Translation updates: be, sk
>
> -- Jeremy Bícha <jbicha@ubuntu.com> Fri, 06 Jun 2025 13:51:45 -0400
--
Sebastian Ramacher
Reply to: