Hi all, Last week I uploaded glibc to jammy-proposed. This gummed up the autopkgtest queues a good deal as usual but once all is done there are two regressions (well three currently but I think one will go away on retry): libassuan/2.5.5-1 and libgpg-error/1.43-3. These fail in the same way: they have a test that tries to install wine32.
wine32 depends on libc6:i386. The pinning set up by autopkgtest means that the native libc6 from proposed is installable (is already installed in fact) but libc6:i386 from proposed is not. This sounds like a bug in autopkgtest but anyway, it causes autopkgtest to flip into "try all packages from proposed" mode, so wine32 is installed from proposed whereupon the tests fail because of https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1006827 (fallout from fixing https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1006539). There are a few ways this could potentially be fixed: fix autopkgtest to put foreign versions of packages in the /etc/apt/preferences.d/autopkgtest-RELEASE-POCKET file, fixing wine somehow (maybe wine32 should depend on wine now? maybe l_inlt.nls needs to move back to libwine), change libassuan and libgpg-errors tests to depend on wine as well, removing the wine package from proposed on the assumption that a fix will come from a sourceful change to wine. What I can't see how to do is to fix this via a hint: we could badtest libassuan/2.5.5-1 and libgpg-error/1.43-3 but then wine would migrate despite the fact that it really does seem to have a regression. I think the fixes that make most sense to me are either forcing glibc to migrate anyway or removing wine from proposed. I'll have a poke at fixing autopkgtest but that might take a little while to get anywhere. Cheers, mwh
-- Ubuntu-release mailing list Ubuntu-release@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-release
