I dropped the gnome-contacts update from here so we can get the rest pushed to stable.

Thanks!

@remyabel Is this a regression compared to the previous F37 build? Can you file a ticket for this at https://gitlab.gnome.org/GNOME/gnome-contacts please?

@markec Sorry, I have no idea, you'll have to ask stransky.

mcatanzaro and mcrha are aware of this and are looking into it (looks like a webkitgtk regression).

(The last one passes my own smoke testing at least.)

Next try; mutter-43~beta-4.fc37 has https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2594 backported in addition to the earlier fixes.

I backported https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2588 in addition to the first patch. Let's see if it improves things.

Works fine in my testing

@adamwill Can you help here, please? Looks like update.install_default_update_live test needs updated needles.

See the various emails I've sent in the morning and discussion on the devel list. (Short version: They are all FTBFS.)

pango 1.50.1 is in updates testing, https://bodhi.fedoraproject.org/updates/FEDORA-2021-baf1c3a19c if you guys want to test the fix and leave karma there. Filing negative karma here doesn't affect anything as the update is already in stable.

Weird -- thanks for testing in any case!

Aha, found it: https://gitlab.gnome.org/GNOME/connections/-/issues/90

Is it a regression in the new version?

Do you have a link to the ticket you filed?

I added back the blocker bug association as I think we should pull this in as the blocker fix as well, in addition to the glib2 update.

This update has been unpushed.

Weird. The sqlite3_value_free issue sounds like maybe BricsCAD is bundling sqlite (tracker links against sqlite as well) and then the dynamic linker ends up resolving tracker's sqlite symbols to the wrong sqlite library and that causes things to blow up.

Sounds like it's safest to unpush this update for now to avoid weird issues with 3rd party programs.

@ozeszty Oh, that's a good point (sorry, I missed your comment earlier). I am not entirely sure what would be the best solution here though. To avoid the duplicate indexers on KDE, I guess a solution would be if libtracker-sparql package stopped recommending tracker package -- but at the same time, doing so would regress the search in gtk file chooser dialog on KDE.

Thanks for fixing this, @bruno! Let's get this stable to fix the broken deps in the repos.