On Thu, Aug 3, 2023 at 9:41 AM Simon McVittie <s...@debian.org> wrote: > https://release.debian.org/britney/pseudo-excuses-experimental.html#glib2.0 > says that if the experimental version landed in unstable now, it would be > unable to migrate.
I triggered retries for those now. I believe those are just flaky results and that the autopkgtests will pass. > Are you aware of anything else blocking those, or should I be opening a > transition-tracker bug? 44 Blocker -------------- We need special handling for mipsel/mips64el for gnome-shell. At least this should be simpler than the gjs/s390x case years ago since at least our team-managed GNOME Shell extensions are arch:all so it reduces the number of architecture-specific removals we need. Soft blocker: My opinion is that we should upload Dash to Dock at the same time. I'll ping Jonathan again to upload the update since he started on it. Otherwise, I am comfortable with kicking the other extensions out of Testing given that we set the hard gnome-shell version dependencies for the 43 cycle so it's relatively clear to users what is happening. 44 Notes ------------ Add gnome-shell-extension-gsconnect to your transition list; the 44 compatible version in Experimental isn't compatible with 43. I think we would be ok with keeping the new xdg-desktop-portal-gnome in Experimental because of how it interacts badly with other installed desktops (still not actually resolved upstream). I do not foresee issues if we push GNOME 45 stuff to Unstable before GNOME Shell. There are a few low level packages that we would need to verify compatibility: gnome-control-center, gnome-settings-daemon, gsettings-desktop-schemas, gtk4, libadwaita. We need to do that anyway, but there's a greater chance of issues skipping a version. So yes, go ahead and open the transition bug. 45 -- GNOME Shell 45 Beta will be released in a few days. Unfortunately, every GNOME Shell extension will be broken by this update and will need to adapt to changes in the new version. The GNOME Shell developers in general request that distros like Debian Unstable provide GNOME Shell Beta to early users so that bugs can be identified in time for the .0 release instead of the bugs getting reported after .0 or .1, which then forces GNOME Shell maintainers to work on fixing major bugs for .1 or .2 and it delays when they can start development for the next GNOME release, which then encourages them to land big changes after Beta. However, I imagine our users will be rather upset if all their extensions no longer work and there aren't even upstream fixes available yet. So I want to be more conservative for GNOME Shell this time and wait until at least RC to release to Unstable. Mutter 45 Beta has one new dependency, libei, so I plan to work with the Debian X team to package that in a few days. We need a new mozjs series. Fortunately gjs seems more stable and I've been assured that we should be able to keep using the old gjs with new GNOME Shell or the other way around right now. Thank you, Jeremy Bícha