[Antruly]

Summary

Adds the regression test for uvp_run's signal-death branch, which #283 merged
without: a fixture that signals itself, asserting exit_code: null,
exit_signal: "SIGTERM", a process_exit unknown and the "was killed by
SIGTERM before the probe answered" wording. Skipped on Windows with the reason
in the test title — the platform reports a code for a terminated process.

Refs #281

Why this is a separate PR

The owner's review of #283 noted the signal branch was implemented but untested
and suggested exactly this fixture. I wrote it, but the commit (4839475) was
pushed after the owner merged #283 at a5ce0dd (5d0a3da), so it is on the
PR branch and not on main. A merged branch is not amended; this is the
change rebased onto main as its own branch, and the correction is recorded on
#283.

Problem and expected behavior

A child that ends by signal produced exit_code: null with exit_signal set.
Nothing pinned that: the other two fixtures exit with a code, so a regression
that dropped the signal — or reported a null code as a code — would have passed.
Expected: the observed signal and the null code are both asserted, and the
unknown names the signal.

Change and scope

• tests/conformance/fixtures/killsItselfWithSigterm.mjs — the self-signalling
  child.
• tests/conformance/handlers.test.ts — one case for it, skipped on Windows with
  the reason in the title.

No source change: this is coverage for behaviour already on main.

Contract and boundary impact

• Earliest changed layer: none (tests only).
• Tool contract changes: none.
• libuvcpp feature mapping changes: none.
• Generated-code impact: none.
• Evidence, unknown, or error-contract semantics: none.
• Generated metadata, package, or installation impact: none.

Expected effects: none — this PR changes test files only and starts no service
outside the test process.

Evidence and regression coverage

• Tests added: the signal case. Observed on main as merged (5d0a3da) with the
  case added: npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts → 129 passed.
• Mutation: forcing the reason back to resource_limit fails the exited-child
  case; dropping exitInfo() fails 2 cases (expected undefined to be 3,
  expected undefined to be null) — the second is this new one.
• The exact command a reviewer can run:
  npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts
• Remaining proof gaps: the case runs only on non-Windows hosts, for the reason
  the title names.

For evidence-bearing changes:

• Observed, derived, and inferred claims remain distinguishable in the output.
• Coverage that was not examined is reported as unknown, not complete.
• A check that could not run yields undetermined, not a fabricated yes or no.
• Resource kinds and provenance survive the boundary unchanged.

Validation performed

• npm run compile — exit 0.

• npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts — 129 passed.

• npm run typecheck — exit 0; npm run lint — exit 0 (see below); npm run format:check — exit 0.

• Test isolation: both readiness cases now pass binary: svc.path (b3bd315).
  Before, filtering to either case spawned build/svc — the launcher an earlier
  case installed — and failed with an ENOENT unrelated to the case; observed by
  the owner at 4adf419. -t "signal death" and -t "bound-bound" each pass
  alone now, and the full file remains 108 passed.

• The static lane's first run failed on max-lines-per-function in
  tests/conformance/handlers.test.ts: the three readiness cases pushed the
  launch and stop callback to 163 lines. Fixed at 4adf419 by moving them into
  their own describe — no test changed, and the suite stayed at 108 passed.

• A template change was generated, built, and run against a real libuvcpp install.

• Not applicable (no template or libuvcpp-facing change).

Compatibility, safety, and release

• Breaking changes or migration steps: none.
• libuvcpp version floor impact: none.
• New network exposure or listening behaviour: none.
• Package or release metadata impact: none.

Author readiness

Human owner: @Antruly

• I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work.
• I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps.
• I declare this PR ready for initial maintainer review.

Before merge

• This PR addresses a concrete problem or an agreed enhancement.
• Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff.
• The PR has one focused outcome and the title follows type(scope): outcome.
• Related issue is linked, or the reason for not linking one is stated above.
• Tests cover the changed observable behaviour and its meaningful failure paths.
• Owning docs, contracts, and generated metadata are updated where needed.
• User-visible CLI or MCP changes include representative output.
• I checked the final diff for secrets, unrelated cleanup, and unsupported claims.

🤖 Antruly-dev-1 · AI Engineer 1 · agent:dev-ai-1

[Antruly]

🤖 Antruly-dev-1 · AI Engineer 1 · agent:dev-ai-1

The first CI run found one real problem, now fixed at 4adf419: the Static checks lane failed on max-lines-per-function in tests/conformance/handlers.test.ts — the three readiness cases grew the launch and stop callback to 163 lines against a 150 limit. Moved them into their own describe("a process that ends before readiness"); no test changed, npm run lint exits 0 locally, and the suite is 108 passed. CI re-ran on the new head.

[Antruly]

Review at 4adf419

The signal branch is now covered, and the case passes.

• npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts → 129 passed; npm run typecheck exit 0; npm run lint exit 0 (warnings only, pre-existing).
• The fixture ends by SIGTERM, and the assertions pin exit_code: null (not 0), exit_signal: "SIGTERM", the process_exit reason, and the "was killed by SIGTERM before the probe answered" wording.
• 4adf419 is a pure test reorganization: it moves the launch/stop cases into their own describe so the suite stays inside the lint budget. No behaviour change; the fixture and assertions are unchanged.
• The Windows skip carries its reason in the test title, so a skipped run and a passing run do not look alike.

Non-blocking (my judgment): the case is not runnable in isolation. writeFakeBinary(workspace, "sigterm", …) installs build/sigterm, but uvp_run is called without binary, so it spawns the manifest's build/svc — the launcher installed by the earlier "refuses to bind" case. npx vitest run … -t 'signal death' therefore fails with spawn …/build/svc ENOENT (exit 1). It is fine in the full-file run, but a developer iterating on this one case gets an error unrelated to it. Passing binary: svc.path (or writing the launcher under svc) would make it self-contained; the silent-child case from #283 shares the pattern.

No blocking findings.

[Antruly]

🤖 Antruly-dev-1 · AI Engineer 1 · agent:dev-ai-1

You were right about the isolation, and it was not only the signal case: the silent-child case had the same dependency on build/svc. Both now pass binary: svc.path (b3bd315), so each names the fixture it just wrote.

Evidence at b3bd315: npx vitest run tests/conformance/handlers.test.ts -t "signal death" → 1 passed | 107 skipped; -t "bound-bound" → 1 passed | 107 skipped; full file → 108 passed; npm run lint and npm run typecheck exit 0. The PR body records the finding and the fix. CI is re-running on this head.

[Antruly]

Review at b3bd315

The signal branch is covered, and the case is now self-contained.

• npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts → 129 passed; npx vitest run tests/conformance/handlers.test.ts -t 'signal death' → 1 passed on its own. The earlier head failed that filter with spawn …/build/svc ENOENT; b3bd315 passes binary: svc.path, so the case names the fixture it wrote (the same fix is applied to the silent-running case).
• The fixture ends by SIGTERM, and the assertions pin exit_code: null (not 0), exit_signal: "SIGTERM", the process_exit reason, and the "was killed by SIGTERM before the probe answered" wording.
• 4adf419 is a pure test reorganization to stay inside the lint budget; b3bd315 adds the binary path. npm run typecheck and npm run lint exit 0 (warnings only, pre-existing).
• The Windows skip carries its reason in the test title, so a skipped run and a passing run do not look alike.

No blocking findings.

[Antruly]

🤖 Antruly-dev-1 · AI Engineer 1 · agent:dev-ai-1

Nothing has changed at b3bd315 since your review — the tree is the one you tested, CI is green (15 pass, the two classifier lanes skipped), and the PR body matches. The approval is the only gate left.

[Antruly]

Approved.

Reviewed head b3bd315f11c7be38a567841408154d34bc6fb739.

• Fixes the coverage gap from feat(run): report the exit status when a service dies before readiness #283: the SIGTERM fixture asserts exit_code: null, exit_signal: "SIGTERM", the process_exit reason, and the wording. Windows skip reason is in the title.
• npx vitest run tests/conformance/handlers.test.ts src/domain/unknown.test.ts → 129 passed; the signal case also passes when the runner is filtered to it (-t 'signal death'), because b3bd315 passes binary: svc.path.
• npm run typecheck, npm run lint exit 0 (warnings only, pre-existing); all selected lanes green at this head; readiness declaration present.
• git merge-tree --write-tree origin/main pr284 exit 0 against 5d0a3da.

Merging with --match-head-commit.