[lpyu001]

Preserve the exit status returned by showfixtures() and show_fixtures_per_test() instead of unconditionally returning 0 from pytest_cmdline_main().

Collection errors require additional handling because Session.perform_collect() records them in session.testsfailed without raising. Extracted the existing collection-error check from pytest_runtestloop() into _raise_on_collection_errors() and reused it for the fixture display commands. With --continue-on-collection-errors, the commands now return ExitCode.TESTS_FAILED when collection errors occurred.

Added regression tests for both --fixtures and --fixtures-per-test. Pre-fix:

$ python -m pytest --fixtures test_bad.py
...
ERROR test_bad.py
=============================== 1 error in 0.09s ===============================
exit=0

After the fix, collection errors return ExitCode.INTERRUPTED (2), or ExitCode.TESTS_FAILED (1) with --continue-on-collection-errors. Exit statuses produced by internal errors and pytest.exit() are also preserved instead of being replaced with 0.

Closes #15047.

• Include documentation when adding new features.
• Include new tests or update existing tests when applicable.
• Allow maintainers to push and squash when merging my commits. Please uncheck this if you prefer to squash the commits yourself.
• Add text like closes #XYZW to the PR description and/or commits (where XYZW is the issue number). See the [github docs](https://github.com/proxy/help.github.com/en/github/managing-work-on-github/managing-your-work-with-issues/linking-a-pull-request-to-an-issue) for more information.
• If AI agents were used, they are credited in Co-authored-by commit trailers.
• Create a new changelog file in the changelog directory, with a name like <ISSUE NUMBER>.<TYPE>.rst. See [changelog/README.rst](https://github.com/pytest-dev/pytest/blob/main/changelog/README.rst) for details.
• Add yourself to AUTHORS in alphabetical order.

[bluetech]

Thanks, I made some tweaks but it should be good for pytest 10.0.

[hafizahmedzia]

Additional checks of @lpyu001's fix compared the status received by pytest_sessionfinish with the final subprocess exit, using both --fixtures and --fixtures-per-test.

On Linux/CPython 3.12.14, comparing 56a92c19d4a71152c594726efc9deb5f0e126e99 with 185e3a7f299308e454afca4c5befb307bc3f139a:

• Syntax-error collection: hook/process status is 0/0 on the parent and 2/2 on the head.
• Successful collection followed by a sessionfinish hook calling pytest.exit(returncode=9): the hook receives 0 on both; the parent exits 0 and the head exits 9.
• Empty-collection controls stay at hook/process 0/0.

All 12 subprocess cases behaved this way. Plugin autoload was disabled, each process recorded the source checkout imported, and test bodies were not executed. This specifically checks preservation of the sessionfinish override. The full pytest suite and CI were not run. This does not change the maintainer's pytest 10.0 timing.