This update has been unpushed.

This update has been unpushed.

I just noticed this update. The polymake build unfortunately conflicts with a rebuild due to a normaliz ABI change in this older build: FEDORA-2026-a169bb426a. Upstream made backwards-incompatible changes to the normaliz ABI without bumping the soname, so a rebuild was necessary. I need a polymake that is built against both perl 5.42.2 and normaliz 3.11.1.

There is a new commit on the Rawhide branch of polymake this morning which should reduce build times somewhat. The Rawhide build is ongoing. Can we do the following? Tag the builds from f44-build-side-133261 into your side tag. Run git merge rawhide on the F44 polymake branch. Do one more build in your side tag. Refresh this update. Cancel FEDORA-2026-a169bb426a. Does that sound okay?

Thanks for the polymake build. It works fine.

Works for building and running python-networkx.

Works for building and running python-networkx.

It needed a couple of tweaks. I have made them and kicked off a build. Thanks for alerting me to the situation.

ocaml-ppx-js-style was deliberately broken and orphaned, as announced on fedora-devel-list. I didn't know of the existence of the lem package. Looks like it is fairly new. I will take care not to break it in the future.

Thank you for the lesson on how to use bodhi, @adamwill, and thank you @yselkowitz for your work.

If you look at the polymake code, you'll see that it does a lot of mucking about with perl internals. In the past, even minor perl updates have caused breakage, hence upstream's policy of requiring the specific major+minor perl version that polymake was built with. Thank you for testing.

Thank you. I reran the tests, but they failed again. As usual, I find the test reports to be absolutely incomprehensible. If somebody who knows how to read this stuff could take a look, I would appreciate some help figuring out what is still broken.

Perhaps we should revert that commit for F41, and then fix whatever needs fixing in Rawhide. That would maintain parity with previous F41 builds.

This and distribution-gpg-keys should have been in the same update to avoid the test failures, but the combination does work.

When you file the bugs, please mention which desktop you use (GNOME, KDE, etc.) and whether you are using Wayland or X11.

The menu issue sounds like https://bugzilla.redhat.com/show_bug.cgi?id=2250564. I don't know how to fix it.

I cannot reproduce either of the two crashes you are talking about. Could you please file bugs for those two with ABRT? The backtraces may help me diagnose the issues.

This update has been unpushed.

python-pdm-backend, python-sphinx-theme-builder, and spglib seem okay without modification. On the other hand, python-llama-cpp-python is broken. I don't understand that, though, since when I do a mock build, python-pyproject-metadata isn't even installed. I'll have to investigate further.

I am going to withdraw this update.

Oh my, it seems that python-pyproject-metadata has grown more consumers than I knew about. Sorry about that. I was under the impression that version 0.8.0 contained only bug fixes. Since it appears to contain backwards incompatible changes, I'm leaning toward withdrawing this update. What do you think?

Yes, the transition from ant to maven has been a little rough. Let's try this again.