Comments

26 Comments

I think the test is failing, because there is a mismatch between gcc and annobin. For some reason, the test is installing gcc-11.2.1-1.fc35 even though the latest build in updates is gcc-11.3.1-2.fc35.

I ran the test case manually in a container with gcc-11.3.1-2.fc35 and it passed for me.

I tested the redhat-rpm-config package and it fixes the problem with clang + openmp + rpath as expected.

Hi all, thanks for testing. A new qt-creator build has been added to the update can you re-test and see if the issue has been fixed?

There is also this annobin test missing: baseos-qe.koji-build.scratch-build.validation

I'm planning to rebuild the other packages once the freeze is over.

This update has been unpushed.

I re-enabled the buildroot overrides, so that no more packages will be built with the old LLVM. I thin k that should prevent problems like what happened with qt6-qttools.

Why would it be easier to create a new update instead of updating this one?

Let's wait on the pocl rebuild until the maintainer responds to my question on FEDORA-2021-3925fa6655.

We rebuilt a pocl pre-release snapshot against llvm 12.0.0 before this update. Should we wait until this update goes stable to rebuild 1.7.0 against llvm 12.0.0? Or should we do it now and obsolete this update?

We're going to need to rebuild pocl, because a newer version was added to this update: FEDORA-2021-3925fa6655

Can you share the dnf output from when you installed clang?

Thanks for testing.

Actually we should just re-test once https://bodhi.fedoraproject.org/updates/FEDORA-2018-689de58112 goes stable if that is the problem. I don't want to modify that update and reset the karma.

We might want to just add this into the llvm update for f28. f28 currently has llvm-6.0.0-rc2, but I think this build of Mesa was built while I had the llvm-6.0.0 override enabled.

This is the same issue as: https://bugzilla.redhat.com/show_bug.cgi?id=1514446 I'm still not sure exactly why this is happening.