Miro, why did you cancel the rebuild for Python 3.15 after 30 minutes? Was there something wrong with it?

I cancelled it after 30 minutes because the previous attempts hung for 17+ hours on the same log line (just after PASS: test-error.sh). The other attempts were restarted semi-automatically, and I canceled them, knowing they wouldn't finish.

See all attempts at https://src.fedoraproject.org/rpms/gnulib/c/bc44a76a602e108c769ae7d1a497b078f4dae565?branch=rawhide

Also, started another rebuild now. Happy to leave it building for a couple of days, but normally this package builds for  ~1 hour, so I don't expect it to finish.

6+ hours later, the build is still stuck just after PASS: test-error.sh.

Hello,

I tested the gnulib build using the Koji side tag [1], and it also hangs. If we ignore this unit test, the build still hangs in various other places.

The first failure occurs in test-execute.sh. The output is shown below [2].

Updating to the latest git snapshot did not help. I will try to contact upstream, as suggested by the test output [3].
In any case, I do not think this problem is related to Python. It may be better to disable the tests for now. My current understanding is that the issue is related to the GNU toolchain (GCC, binutils, and glibc), specifically in tests that rely on process forking. The failures appear to be intermittent, which may indicate a race condition or another issue in the test code itself. 

As you may have noticed, I can now write more fluently :) because I am using ChatGPT to help correct my English.

Best regards,

[1]
fedpkg mock-config --target=f45-python > sidetag.cfg
fedpkg mockbuild --root=sidetag.cfg

[2]
cat test-execute.sh.log

test-execute-main: test-execute-child subprocess failed: No such file or directory
test-execute-main.c:424: assertion 'fd_out >= 0 && fd_out < 15' failed
./test-execute.sh: line 4: 215724 Aborted (core dumped) ${CHECKER} ./test-execute-main${EXEEXT} ./test-execute-child${EXEEXT} $i
test-execute.sh: test case 19 failed
FAIL test-execute.sh (exit status: 1)

[3]
Please report test failures in this directory to bug-gnulib

I canceled my build again.

---

Sergio, the f45-python side tag has been merged; no need to use it anymore.

I propose this PR: https://src.fedoraproject.org/rpms/gnulib/pull-request/4

I'm no longer a Fedora packager, so you need to upload the sources before starting the build.

Hello,

Please note that this comment was generated automatically by https://forge.fedoraproject.org/releng/tooling/src/branch/main/release-process/bug-filing/follow-policy/follow_policy.py
If you feel that this output has mistakes, please open an issue at https://forge.fedoraproject.org/releng/tickets/issues

This package fails to install and maintainers are advised to take one of the following actions:

 - Fix this bug and close this bugzilla once the update makes it to the repository.
   (The same script that posted this comment will eventually close this bugzilla
   when the fixed package reaches the repository, so you don't have to worry about it.)

or

 - Move this bug to ASSIGNED if you plan on fixing this, but simply haven't done so yet.

or

 - Orphan the package if you no longer plan to maintain it.


If you do not take one of these actions, the process at https://docs.fedoraproject.org/en-US/fesco/Fails_to_build_from_source_Fails_to_install/#_package_removal_for_long_standing_ftbfs_and_fti_bugs will continue.
This package may be orphaned in 7+ weeks.
This is the first reminder (step 3) from the policy.

Don't hesitate to ask for help on https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/ if you are unsure how to fix this bug.

Hello,

Please note that this comment was generated automatically by https://forge.fedoraproject.org/releng/tooling/src/branch/main/release-process/bug-filing/follow-policy/follow_policy.py
If you feel that this output has mistakes, please open an issue at https://forge.fedoraproject.org/releng/tickets/issues

This package fails to install and maintainers are advised to take one of the following actions:

 - Fix this bug and close this bugzilla once the update makes it to the repository.
   (The same script that posted this comment will eventually close this bugzilla
   when the fixed package reaches the repository, so you don't have to worry about it.)

or

 - Move this bug to ASSIGNED if you plan on fixing this, but simply haven't done so yet.

or

 - Orphan the package if you no longer plan to maintain it.


If you do not take one of these actions, the process at https://docs.fedoraproject.org/en-US/fesco/Fails_to_build_from_source_Fails_to_install/#_package_removal_for_long_standing_ftbfs_and_fti_bugs will continue.
This package may be orphaned in 4+ weeks.
This is the second reminder (step 4) from the policy.

Don't hesitate to ask for help on https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/ if you are unsure how to fix this bug.

It is caused by glibc commit 7594b4e ("posix: Add POSIX aliases to some spawn functions", Lucas Chollet, committed to master 2026-04-07 during the 2.44 development cycle).

gettext, m4, and gnulib at the least are also getting stuck during tests.

To reproduce you can do as little as:
```
cat << EOF > test.c
#include <spawn.h>
 
int
posix_spawn_file_actions_addchdir (posix_spawn_file_actions_t *fa,
                                   const char *path)
{
  return posix_spawn_file_actions_addchdir_np (fa, path);
}
 
int
main (void)
{
  posix_spawn_file_actions_t fa;
  posix_spawn_file_actions_init (&fa);
  return posix_spawn_file_actions_addchdir (&fa, "/");
}
EOF
 
gcc -D_GNU_SOURCE -O2 -o test test.c && ./test
```

_USE_XOPEN2K24XSI is only defined when _XOPEN_SOURCE >= 800, and _GNU_SOURCE implies _XOPEN_SOURCE 800. With -O2 it hangs, otherwise it crashes.

The posix_spawn_file_actions_addchdir replacement becomes an infinite loop with the glibc 2.44 prerelease header-only POSIX alias

I updated Gnulib to the stable-202607 branch and the latest git snapshot, but we are still unable to complete the test suite. 

The Fedora Koji build consistently hangs after PASS: test-error.sh:
https://koji.fedoraproject.org/koji/taskinfo?taskID=148168073

This appears to be the same infinite loop issue described in this bug

Sorry, I meant bug #2502638

OK, https://src.fedoraproject.org/rpms/gnulib/pull-request/5 has one fix. After uploading the source, the build will be successful ( https://koji.fedoraproject.org/koji/taskinfo?taskID=148625601 )

I would appreciate help from an proven packager to merge the fix, upload the source, and build it in Rawhide.

Fixed in gnulib-0-57.20260302git4a3650d.fc45

Sorry I mean gnulib-0-60.20260801git9748dc4.fc45

(In reply to Sergio Basto from comment #12)
> OK, https://src.fedoraproject.org/rpms/gnulib/pull-request/5 has one fix.
> After uploading the source, the build will be successful (
> https://koji.fedoraproject.org/koji/taskinfo?taskID=148625601 )
> 
> I would appreciate help from an proven packager to merge the fix, upload the
> source, and build it in Rawhide.

thanks Sergio and Peter, this package was orphaned in rawhide, I took the ownership and added peter as an admin.

sergio, your user is not part of packager groups, so I was unable to add you

(In reply to Filipe Rosset from comment #15)

I made several improvements to `gnulib.spec`, and it's a bit disappointing that they didn't make it in. I split the changes into seven separate commits, each of which can be reviewed and merged independently.

Personally, I would have preferred to use a stable branch instead of the main branch, and I was waiting for upstream to merge the build fix into the stable branch.

The commits are available here:
https://src.fedoraproject.org/rpms/gnulib/pull-request/6