Bug#1132301: transition: gnome-shell 50

[Simon McVittie]

Package: release.debian.org
Severity: normal
Control: tags -1 + moreinfo
X-Debbugs-Cc: gnome-shell at packages.debian.org
Control: affects -1 + src:gnome-shell
User: release.debian.org at packages.debian.org
Usertags: transition

We now have GNOME Shell 50 in experimental (I'm typing this email into 
it) and I've just done the usual MBF asking extensions authors to 
update. This will also be the GNOME release used in Ubuntu 26.04 LTS.

Jeremy, Marco: are we ready to ask for a transition slot? (Tagged 
moreinfo for now.)

This time, the group of entangled packages is:

* gdm3
* gnome-kiosk
* gnome-remote-desktop
* gnome-session
* gnome-shell
* mutter

all of which are GNOME-team-maintained.

The headline change is that GNOME 50 doesn't support being run as an X11 
session using Xorg, at all: it's Wayland or nothing. X11 apps can still 
run inside the session via Xwayland as usual.

libgdm1 has an ABI break with no SONAME bump (sigh) as a result of its 
X11 functionality being removed, but that library is not widely used and 
it seems that the symbols that were removed were only used within gdm3 
or by gnome-kiosk, so we can probably get away with using a versioned 
Breaks on gnome-kiosk (already present) rather than bumping the package 
name to libgdm1a.

https://release.debian.org/transitions/html/auto-mutter.html tracks the 
mutter SONAME bump. When we upload gnome-shell to unstable, it will 
presumably also generate an auto-upperlimit- tracker.

Usertagged bugs (all Shell extensions right now, but we can use the same 
usertag for any other blockers):
https://udd.debian.org/cgi-bin/bts-usertags.cgi?user=pkg-gnome-maintainers%40lists.alioth.debian.org&tag=gnome-shell-50

As usual, the automated tests for gnome-shell and mutter are not as 
reliable as we would like, so some retries might be necessary.  I'm 
reluctant to disable them completely because they're our only 
opportunity to find out whether things are completely broken on non-x86, 
but we can if we have to.

Thanks,
    smcv