TASK: Fix C10S FTBFS

[yselkowitz]

What does the ELN SIG need to do?

• The following packages are building in rawhide and ELN -- and therefore packages tracked in #176 are excluded -- but are failing in CentOS Stream 10. They need to be fixed for a successful conclusion to the bootstrap.

• [x] alsa-sof-firmware: ALSA dependencies were built out of order, need rebuilding

• [x] clang: i686 OOM, https://src.fedoraproject.org/rpms/clang/pull-request/227

• [x] dotnet8.0: runtime_id does not match fedora/eln, may need to be bootstrapped

• [x] gmp - https://issues.redhat.com/browse/RHEL-24891

• [x] grub2: error: cannot open %{sb_cer}

• [x] iproute: rt_dsfield.deprecated needs to be imported post-branching

• [x] ksh

• [x] libecpg - https://issues.redhat.com/browse/RHEL-24889

• [x] libkcapi

• [x] liblouis

• [x] nettle

• [x] perl-DBD-MySQL: waiting for a real mysql package (being renamed from community-mysql)

• [ ] pkcs11-provider

• [x] pygobject3

• [x] python-psutil - https://issues.redhat.com/browse/RHEL-24890

• [x] python-pygit2:

• [x] python3.12: needs bootstrapping without ELN builds in inheritance

• [x] ruby - tests trying to access outside network - https://issues.redhat.com/browse/RHEL-25008

• [x] rubygem-eventmachine

• [x] scap-security-guide: no upstream EL10 content; should be treated as a placeholder

• [x] scipy

• [x] snowball

• [x] tftp

• [x] tracker-miners: meson.build:208:4: ERROR: Problem encountered: Landlock was enabled in build options, but is disabled in the kernel

• [x] varnish: test flakes, retry

[hroncok]

Feel free to ping me for python3.12 bootstrap. From the top of my head, it should be possible to build it like this:

1. build python-pip, python-setuptools, python-wheel in c10s using python3.12 from eln
2. build python3.12 in c10s using results from 1.

[yselkowitz]

@tdawson once alsa-utils-1.2.11-1.eln135 has synced over to the c10s buildroot, please retry alsa-sof-firmware

[yselkowitz]

rubygem-eventmachine: DROP via https://src.fedoraproject.org/rpms/rubygem-mysql2/pull-request/17

[yselkowitz]

libecpg also appears to be a race condition in the Makefiles.

[tdawson]

ruby is trying to access the outside network during it's tests - https://issues.redhat.com/browse/RHEL-25008

[yselkowitz]

ruby is trying to access the outside network during it's tests - https://issues.redhat.com/browse/RHEL-25008

This sounds like a koji/mock configuration difference between Fedora/ELN and CS/RHEL, because the same tests don't fail in the former?

[tdawson]

ruby is trying to access the outside network during it's tests - https://issues.redhat.com/browse/RHEL-25008

This sounds like a koji/mock configuration difference between Fedora/ELN and CS/RHEL, because the same tests don't fail in the former?

I believe you are correct. This was happening with some other package as well. I'm not remembering which. I personally do not like tests that try to access the outside network unless they can be turned off in a real way. The ones that look and see if your network has such and such a configuration are simply too fragile.

[yselkowitz]

tracker-miners: https://src.fedoraproject.org/rpms/tracker-miners/pull-request/8

[yselkowitz]

liblouis: https://src.fedoraproject.org/rpms/liblouis/c/bac0acc0b1517ca1ed08281fee7c3542d482def4?branch=rawhide (mistakenly did fedpkg push instead of fedpkg fork, although given the time constraints I probably would have had to merge the PR myself anyway...)

Update: the build problem is fixed, but now stream koji complains that the noarch -tables subpackage is not identical between buildArch tasks.

[yselkowitz]

tftp: https://src.fedoraproject.org/rpms/tftp/pull-request/4

[yselkowitz]

snowball: https://src.fedoraproject.org/rpms/snowball/pull-request/2

[yselkowitz]

python-pygit2: https://src.fedoraproject.org/rpms/python-pygit2/pull-request/3

[yselkowitz]

pygobject3: https://src.fedoraproject.org/rpms/pygobject3/pull-request/6

[yselkowitz]

liblouis part 2: https://src.fedoraproject.org/rpms/liblouis/pull-request/7

[yselkowitz]

pkcs11-provider test failure is related to softhsm. There have been a number of issues with tests using softhsm due to internal changes in the crypto stack. Disabling the BuildRequires: softhsm is a workaround, at least for the purposes of bootstrapping (there has yet to be a build in c10s).

[tdawson]

ruby: https://src.fedoraproject.org/rpms/ruby/pull-request/170

[yselkowitz]

Bootstrap is complete; remaining issues are now up to the maintainers to solve.