id: Don't treat unresolved IDs as error by default

[mhutter]

This brings the behavior of id in line with GNU coreutils, which only treats unresolved IDs as errors when providing any of the -u, -g or -G flags in combination with -n.

Fixes: #12246

[sylvestre]

could you please add a test to make sure we don't regress? thanks

[mhutter]

could you please add a test to make sure we don't regress? thanks

Any idea how I can test the behavior in the context of a non-existing UID/GID? Are there existing tests that test similar scenarios?

edit: I think I have a solution, will update the PR

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/date/date-locale-hour (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/pr/bounded-memory (passes in this run but fails in the 'main' branch)
Note: The gnu test tests/printf/printf-surprise is now being skipped but was previously passing.

[mhutter]

I'll fix this in the follow-up PR to address GID deduplication.

[oech3]

You cannot use unshare on the CI. See #12120 (comment) .

[mhutter]

Then the test should just short-circuit in CI. That way at least we have a way to test the desired behavior locally.

[mhutter]

@sylvestre WDYT?

One of the CI Checks failed for unrelated reasons (timeout)

[oech3]

Would you add

# some test needs unshare
sudo sysctl -w kernel.unprivileged_userns_clone=1 ||:
sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0 ||:

to one of the *.yml to catch regression?

[mhutter]

@oech3 thanks, didn't realize this was an option on GitHub CI.

• I used || true instead of || : to be consistent with existing code; hope that's okay
• I have added the commands to one of the tests that only runs on Ubuntu; in the Windows builds the commands fail even with || true