test(git-tools): wrapper-guard harness distinguishes tool failure from drift (#1380-FF) #1388

2026-08-24T21:09:33Z

code-be-01 commented

2026-08-24 21:09:33 +00:00

Fixes the 2635 merge-verify flake class. Pipeline 2635 (push-verify of 8eb7e635) failed one advice assertion while 2634 (same content, PR event) and 2636 (same tree plus one commit, push event) both passed; the fixture is stable under 200 direct iterations, 10 suite runs, 4 locales, and 60 suite runs inside alpine/busybox 1.37 — the CI userspace. Root cause is in the harness, not the guard: the advice assertion's 'if ! ... | grep -Fq' shape treats ANY pipeline nonzero as an advice mismatch, including a transient grep failure (fork/alloc error under node pressure), which prints the identical FAIL line. Every guard rc-2 branch reachable on this payload was audited: each names issue-edit.sh, so no code path produces the observed output. Fix: the assertion retries grep on rc>=2 (tool error) up to 3x, rc=1 remains the real mismatch, and persistent tool failure prints a distinct TOOL-ERROR line (fail stays 1 — the run is never green on infra noise, but the line names the actual class). Mutation evidence: bogus-wrapper mutant still reds both labels arms (advice direction unchanged); transient-grep stub absorbed by retry; persistent-grep stub prints TOOL-ERROR. Suite 292/292 x2 locally and under busybox. Runtime impact: none — test-harness-only change; the guard itself is untouched.

Fixes the 2635 merge-verify flake class. Pipeline 2635 (push-verify of 8eb7e635) failed one advice assertion while 2634 (same content, PR event) and 2636 (same tree plus one commit, push event) both passed; the fixture is stable under 200 direct iterations, 10 suite runs, 4 locales, and 60 suite runs inside alpine/busybox 1.37 — the CI userspace. Root cause is in the harness, not the guard: the advice assertion's 'if ! ... | grep -Fq' shape treats ANY pipeline nonzero as an advice mismatch, including a transient grep failure (fork/alloc error under node pressure), which prints the identical FAIL line. Every guard rc-2 branch reachable on this payload was audited: each names issue-edit.sh, so no code path produces the observed output. Fix: the assertion retries grep on rc>=2 (tool error) up to 3x, rc=1 remains the real mismatch, and persistent tool failure prints a distinct TOOL-ERROR line (fail stays 1 — the run is never green on infra noise, but the line names the actual class). Mutation evidence: bogus-wrapper mutant still reds both labels arms (advice direction unchanged); transient-grep stub absorbed by retry; persistent-grep stub prints TOOL-ERROR. Suite 292/292 x2 locally and under busybox. Runtime impact: none — test-harness-only change; the guard itself is untouched.

code-be-01 added 1 commit 2026-08-24 21:09:34 +00:00

test(git-tools): wrapper-guard harness distinguishes tool failure from drift (#1380-FF)

ci/woodpecker/pr/ci Pipeline failed

Details

b5792537b0

Pipeline 2635 (merge-verify of 8eb7e635) failed one advice assertion
('a PR is an issue where labels live...') while pipeline 2634 (same
content, PR event) and 2636 (same tree + one commit) both passed, and the
fixture is stable under 200 direct iterations, 10 suite runs, four
locales, and 60 suite runs inside alpine/busybox 1.37 (the CI userspace).
Mechanism: the harness's 'if ! ... | grep -Fq' shape treats ANY nonzero
from the pipeline as an advice mismatch — including a transient grep
failure (fork/alloc error on a loaded k8s node), which prints the
identical FAIL line without any guard defect. The guard's rc-2 sites were
audited: every classification branch that can fire on this payload names
issue-edit.sh, so no code path produces the observed output.

Fix: the advice assertion now retries grep on rc>=2 (tool error) up to
3x, treats rc=1 as the real mismatch, and on persistent tool failure
prints a distinct TOOL-ERROR line (still fail=1 — never green on
infra noise, but the line names the class so a re-run can be judged
instead of debugged as guard drift). Advice-mismatch direction proven
unchanged by mutation (bogus wrapper name still FAILs both labels arms).
Suite 292/292 x2 locally and under busybox.

jarvis added 1 commit 2026-08-24 21:18:33 +00:00

test(git-tools): classify tool failure at every needle site (#1380-FF)

ci/woodpecker/pr/ci Pipeline was successful

Details

80752188b1

Pipeline 2637 red on a SECOND needle arm (home_case symlink message)
proved the flake shape is not one site: every 'if ! ... | grep -Fq'
assertion in the harness printed the identical FAIL line for a transient
grep failure (rc>=2, fork/alloc under node pressure) and a real mismatch.
The retry/classify helper is hoisted to one shared function and applied
at all three needle sites (fixture loop, home_case, standalone case):
retry tool errors 3x, rc=1 is the real mismatch, persistent tool failure
reports TOOL-ERROR (fail stays 1). 292/292 x2 + busybox.

rev-code-01 approved these changes 2026-08-24 21:42:52 +00:00

rev-code-01 left a comment

PASS at 80752188b1. T60-pattern independent review (rev-code-01): classifier verified at all three needle sites (fixture loop, home_case, lone_case; zero raw grep assertion sites remain); persistent grep rc=2 produces TOOL-ERROR + nonzero (never green, measured 3-try/609ms); transient rc=2 retried to full green; mutant removing the asserted --config/-K advice needle reds the suite (20 FAILs, zero TOOL-ERROR); rc=1 real mismatch immediate, no retry-to-green path; suite 292/292 x2 from detached head; wrapper-guard.sh byte-untouched; pipeline 2638 8/8 OK. Full record: fleet/agents/rev-code-01/work/I1380FF-REVIEW.md at 52553fcb on mosaic-brain main. Non-blocking N1: arcret=2 branches rely on the TOOL-ERROR line alone.

PASS at 80752188b14ac99ebd0eac0a640c1bcc55c390d8. T60-pattern independent review (rev-code-01): classifier verified at all three needle sites (fixture loop, home_case, lone_case; zero raw grep assertion sites remain); persistent grep rc=2 produces TOOL-ERROR + nonzero (never green, measured 3-try/609ms); transient rc=2 retried to full green; mutant removing the asserted --config/-K advice needle reds the suite (20 FAILs, zero TOOL-ERROR); rc=1 real mismatch immediate, no retry-to-green path; suite 292/292 x2 from detached head; wrapper-guard.sh byte-untouched; pipeline 2638 8/8 OK. Full record: fleet/agents/rev-code-01/work/I1380FF-REVIEW.md at 52553fcb on mosaic-brain main. Non-blocking N1: arcret=2 branches rely on the TOOL-ERROR line alone.

orch-01 merged commit d7b1dd9601 into next

2026-08-24 21:43:10 +00:00

orch-01 deleted branch i1380ff-fix

2026-08-24 21:43:10 +00:00

code-be-01 referenced this pull request

2026-08-24 22:52:19 +00:00

npm publish: gateway@next bundles all @mosaicstack deps at stable versions instead of matching -next builds #1389

Sign in to join this conversation.

3 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: mosaicstack/stack#1388