Skip to content

test: skip the non-UTF-8 pool name test on macOS, like its siblings - #12914

Merged
bolichen97 merged 1 commit into
mainfrom
test/skip-non-utf8-name-on-macos
Sep 22, 2026
Merged

bolichen97 merged 1 commit into
mainfrom
test/skip-non-utf8-name-on-macos

Conversation

@bolichen97

Copy link
Copy Markdown
Contributor

Problem / Motivation

macOS Tests / Backend Tests (macOS) (3) fails on every PR that runs the on-demand macOS lane since #12709 landed:

test/test_subprocess_pool.py::TestResolutionParity::test_a_name_that_is_not_utf8_round_trips
OSError: [Errno 92] Illegal byte sequence: b'.../test_a_name_that_is_not_utf8_r0/not\xffutf8'

The test creates a directory whose name is not valid UTF-8 to compare the pool child's realpath_spellings against the thread's. APFS/HFS+ do not accept such a name unit, so os.mkdir raises EILSEQ in the fixture, before the resolver is reached. main did not catch it because the macOS lane is on-demand.

Why it matters

Any PR whose author opts into the macOS lane sees a red check unrelated to their change (seen on #12782), and the lane cannot tell a real regression in the pool's resolution parity from this fixture failure.

What changed (motivation → approach → change)

Every other non-UTF-8 path test in the suite already skips on darwin as well as Windows with the same reason string (#11566: test_git_worktree_scope.py, test_governance_updates.py, test_md_notebook.py). This PR gives the pool test the same marker.

posix_names_only cannot simply be widened: it also guards test_a_name_containing_a_newline_resolves, and a newline IS a legal APFS name unit, so that case keeps running on macOS. A second marker, utf8_relaxed_names_only (os.name == "nt" or sys.platform == "darwin"), is applied to the non-UTF-8 case alone.

Area Change
test/test_subprocess_pool.py new utf8_relaxed_names_only marker; applied to test_a_name_that_is_not_utf8_round_trips

Backwards compatibility

Test-only. No code under test changes; Linux runs both cases exactly as before.

Tests

  • Linux: test/test_subprocess_pool.py — 57 passed (both odd-name cases run).
  • Under sys.platform == "darwin" the module's marks were checked directly: the non-UTF-8 case carries skipif(True), the newline case skipif(False).

Manual verification

Not needed beyond the lane itself; the next macOS run on a PR carrying this change is the check.

Related Issues

Follow-up to #12709 (introduced the test) and #11566 (the macOS skip its siblings carry). Seen on #12782.

Checklist

  • Tests added/updated for the change
  • Spec/docs updated where behavior changed (test-only; none)
  • Commit message follows <type>: <summary>

test_a_name_that_is_not_utf8_round_trips (#12709) creates a directory
whose name is not valid UTF-8 to compare the child's realpath against
the thread's. APFS/HFS+ refuse such a name unit: os.mkdir raises
EILSEQ before the resolver is reached, so the macOS lane fails on the
fixture, not on the code under test. Every other non-UTF-8 path test
in the suite already skips on darwin as well as Windows (#11566); give
this one the same marker and keep the newline-name case, which is
legal on APFS, running there.
@bolichen97
bolichen97 requested review from a team and chenmingwei23 September 22, 2026 21:06
@bolichen97
bolichen97 merged commit b35a866 into main Sep 22, 2026
84 of 85 checks passed
@bolichen97
bolichen97 deleted the test/skip-non-utf8-name-on-macos branch September 22, 2026 21:19
@github-actions

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — ✅ PASS

Design-level review of fa8cf69d159ae6cd8120811de5b24a045df3ac72 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Confirm the next on-demand macOS lane run goes green — that run is the only real proof the remaining newline case passes on APFS.

[DESIGN-REVIEWED] fa8cf69

@github-actions

Copy link
Copy Markdown
Contributor

Opus 4.8 Review — ✅ no blocking findings

Reviewed fa8cf69d159ae6cd8120811de5b24a045df3ac72 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] fa8cf69

Verdict parsed from the review's SHA-scoped output markers for commit fa8cf69d159ae6cd8120811de5b24a045df3ac72.

False positive or not applicable? A repository writer can comment:
/ai-review override fable fa8cf69d159ae6cd8120811de5b24a045df3ac72: <one-sentence reason>

@github-actions

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — 🔴 changes requested (blocking)

GPT 5.6 found at least one blocking issue that must be resolved before merging fa8cf69d159ae6cd8120811de5b24a045df3ac72. Adjudication upheld 1 of 1 blocking finding(s); those must be resolved before merging.

This comment is updated in place on each push.

BLOCKING -- test/test_subprocess_pool.py:46 -- Broadened skip disables the macOS parity ratchet
os.name == "nt" or sys.platform == "darwin",
macOS collection -> skipif(True) -> non-UTF-8 resolver parity is never exercised.
Anchor: a-ratchet-may-only-tighten
Fix: Keep the Windows-only marker and avoid creating the invalid directory on Darwin while testing lexical resolution parity.
[BLOCK-MERGE] fa8cf69
[GPT-REVIEWED] fa8cf69

Adjudication (Opus 4.8) — is blocking on each finding proportionate?

The finding is anchored to AUTOSDE rule a-ratchet-may-only-tighten, which carries blocking: true. The changed line (test/test_subprocess_pool.py:46) broadens a skipif from Windows-only to also skip macOS — squarely within the rule's flag category ("a newly added skip / xfail … in a file the diff otherwise only extends"), and file-pattern test/**/*.py matches.

Per the adjudication contract, a finding anchored to an AUTOSDE rule with blocking: true is authoritative and outranks my weighing; I hold no authority to downgrade it.

Harm rung: N/A — governed by an AUTOSDE blocking rule, not my harm/remedy weighing.
Conditions confirmed: broadened skip at test/test_subprocess_pool.py:46; anchor rule a-ratchet-may-only-tighten is blocking: true at AUTOSDE.yaml:717.
Real fix cost: irrelevant — the blocking flag is dispositive.

[ADJUDICATION] fa8cf69d159ae6cd8120811de5b24a045df3ac72 total=1 uphold=1 downgrade=0
UPHOLD F1 test/test_subprocess_pool.py:46 reason=autosde-blocking-rule
[GPT-ADJUDICATED] fa8cf69d159ae6cd8120811de5b24a045df3ac72

False positive or not applicable? A repository writer can comment:
/ai-review override gpt fa8cf69d159ae6cd8120811de5b24a045df3ac72: <one-sentence reason>

wknight1 pushed a commit to wknight1/KiroCrew that referenced this pull request Oct 5, 2026
…irodotdev#12914)

test_a_name_that_is_not_utf8_round_trips (kirodotdev#12709) creates a directory
whose name is not valid UTF-8 to compare the child's realpath against
the thread's. APFS/HFS+ refuse such a name unit: os.mkdir raises
EILSEQ before the resolver is reached, so the macOS lane fails on the
fixture, not on the code under test. Every other non-UTF-8 path test
in the suite already skips on darwin as well as Windows (kirodotdev#11566); give
this one the same marker and keep the newline-name case, which is
legal on APFS, running there.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant