Repository navigation
fix(ci): avoid SIGPIPE in native test listing - #132
Conversation
Test-First: scripts/test-native-tests.sh::native-tests handles long listings without SIGPIPE
Regression-For: 7d4d699
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0df81c39b0
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if ! PATH="$tmp/bin:$PATH" OS=Windows_NT CARGO_TARGET_DIR="$tmp/target" \ | ||
| bash "$ROOT/scripts/ci/native-tests.sh" > "$tmp/out" 2>&1; then |
There was a problem hiding this comment.
Run the new regression test from the gate list
This regression test has no caller outside its own file: at the reviewed tree, scripts/ci/gates.sh does not invoke it directly or through any self-test group. Consequently, the normal Linux and release lanes can remain green if the SIGPIPE-prone pipeline is later restored, so the test only protects the fix when someone remembers to run it manually; register it in the appropriate self-test gate.
AGENTS.md reference: AGENTS.md:L131-L146
Useful? React with 👍 / 👎.
|
Queued — the merge queue status continues in this comment ↓. |
The SIGPIPE regression test was wired into no gate lane and never ran in CI. No-Test: gate wiring only; the wired test already exists and is proven red on the merge base Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
@Mergifyio queue |
Merge Queue Status
This pull request spent 1 hour 8 minutes 35 seconds in the queue, including 22 minutes 17 seconds running CI. Required conditions to merge
|
Der native Windows-CI-Testlauf meldet gültige Tests nicht mehr fälschlich als fehlend, wenn die Testliste groß ist.
Die Regression ist mit einer langen simulierten Testliste reproduziert und der kleinste Fix ist umgesetzt.
Die vollständige Linux-Prepush-Bahn ist grün; der Windows-Lauf bleibt für die Merge-Queue offen.
Report
Was geändert wurde
scripts/ci/native-tests.sh: Die Testnamensprüfung liest die gespeicherte Testliste per Here-String statt überprintf | grep -q; dadurch kannpipefailkein SIGPIPE vonprintfmehr als fehlenden Test fehlinterpretieren.scripts/test-native-tests.sh: Regressionstest mit gültigen nativen Tests und einer langen Ausgabe.Belege
bash scripts/test-native-tests.sh— Exit 0 nach dem Fix; auf Merge-Baseorigin/main— Exit 1 mit derselben langen gültigen Testliste.bash scripts/ci/gates.sh lane prepush— Exit 0 (1.656 Rust-Tests bestanden, 7 übersprungen).git ls-remote origin codex/inv-05-ci-scripts— bestätigt0df81c3.NICHT ABGEDECKT
#[cfg(windows)]-Tests und Windows-Arme von Clippy wurden auf diesem Linux-Host nicht ausgeführt; der Windows-Lauf erfolgt in der Merge-Queue.Review: erledigt (Stufe B, siehe Disposition).
Bestandsaufnahme
scripts/ci/native-tests.sh:84:printf ... | grep -qkonnte unterpipefailbei langen gültigen Testlisten SIGPIPE als Testfehler melden; behoben in diesem Paket.STAND.md,docs/PLAN.mdundKNOWN_ISSUES.md.Noch offen
Review disposition
scripts/ci/gates.sh:71(vorher):scripts/test-native-tests.shwar in keiner Bahn eingehängt, der Regressionstest lief in CI nie. Angenommen, behoben in 5c85b75 (anselftest-gatesangehängt;test-gates.sh,test-ci-shape.shund prepush-Hook Exit 0).native-tests.shselbst geprüft: Here-String ist korrekt; Test rot auf origin/main-Skript (Exit 1), grün mit Fix (Exit 0). Kein weiterer Befund.