Before creating this update, the F34 package had a higher NVR than the F35 package, resulting in a downgrade (https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/FIFXMOXAGR5VCJ7AKYMJYTLUL6EK7SSA/). This update corrects the upgrade path.

Thanks for the bugfix!

$ age --version
1.0.0

Typical Google. They did this without a soversion bump.

I noticed. It was careless of them, and easy for you to miss.

I checked cctz, geometry-hpp, python-pypet, seqan3, and snappy. Based on source grepping and fedpkg mockbuild --enablerepo=local), none of them uses the part of the API that was changed. That’s less fallout than I expected!

I have no objection to carrying on with the F35 update at this point. I appreciate your work on handling this the right way.

Please consider limiting this update to Rawhide at this point in the release cycle for F35.

There is a breaking API change, mentioned in the release notes (https://github.com/google/benchmark/releases/tag/v1.6.0), in that State::thread_index and State::threads are now accessor methods instead of data members. This means all packages that use these members will need to patched.

In addition, please consider sending a heads-up to the devel mailing list about the change.

I noticed the problem because grpc failed to rebuild into the side tag for OpenSSL 3.0. Here is what the necessary patch looked like for grpc.

Requires: FEDORA-2021-55adef54f6

Please unpush this update from F33 and F34 per https://docs.fedoraproject.org/en-US/fesco/Updates_Policy/#stable-releases. It is ABI-incompatible and requires many dependent packages to be rebuilt. While many use it only as a BuildRequires and will be mostly unaffected, others will go FTI. See https://bugzilla.redhat.com/show_bug.cgi?id=2001356 for an example of the consequences.

(For Rawhide and potentially F35, this update seems reasonable. It should have been announced to the devel list and to dependent package maintainers a week in advance, however, and rebuilds should have been coordinated in a side tag.)

I left negative karma on the F33-F34 updates, so they will not be autopushed. If I were the maintainer, I would consider leaving the update out of F35 at this point in the release cycle, too. However, it is not too late to do the necessary rebuilds if the update is judged to be worth it.

Please unpush this update from F33 and F34 per https://docs.fedoraproject.org/en-US/fesco/Updates_Policy/#stable-releases. It is ABI-incompatible and requires many dependent packages to be rebuilt. While many use it only as a BuildRequires and will be mostly unaffected, others will go FTI. See https://bugzilla.redhat.com/show_bug.cgi?id=2001356 for an example of the consequences.

(For Rawhide and potentially F35, this update seems reasonable. It should have been announced to the devel list and to dependent package maintainers a week in advance, however, and rebuilds should have been coordinated in a side tag.)

Please unpush this update from F33 and F34 per https://docs.fedoraproject.org/en-US/fesco/Updates_Policy/#stable-releases. It is ABI-incompatible and requires many dependent packages to be rebuilt. While many use it only as a BuildRequires and will be mostly unaffected, others will go FTI. See https://bugzilla.redhat.com/show_bug.cgi?id=2001356 for an example of the consequences.

(For Rawhide and potentially F35, this update seems reasonable. It should have been announced to the devel list and to dependent package maintainers a week in advance, however, and rebuilds should have been coordinated in a side tag.)

This update has been unpushed.

Unfortunately, this update causes an ABI change in MorphIO:

    [C] 'function morphio::Property::Properties morphio::readers::h5::load(const HighFive::Group&)' at morphologyHDF5.cpp:86:1 has some indirect sub-type changes:
      parameter 1 of type 'const HighFive::Group&' has sub-type changes:
        in referenced type 'const HighFive::Group':
          in unqualified underlying type 'class HighFive::Group' at H5Group.hpp:22:1:
            type size changed from 64 to 192 (in bits)
            1 base class insertion:
              class HighFive::PathTraits<HighFive::Group> at H5Path_traits.hpp:16:1

so it’s probably best to leave it out of F34.

Fixed failing mock due to GPG issues after F35 branching. Thanks!

See also FEDORA-2021-2f4c74c1ee for python-fastapi, which depends on this update.

See also FEDORA-2021-4d3de3183f, which is a prerequisite.

This update has been unpushed.

This update has been unpushed.

This update has been unpushed.

This update has been unpushed.

Allows python-fastapi-0.63.0-3 to build for Fedora 34.