ui/_docs/03_implementation/reviews/batch_06_review.md

Code Review Report

Batch: 6 — AZ-463, AZ-469, AZ-476, AZ-477 Date: 2026-05-11 Verdict: PASS Mode: Full (per-batch invocation by /implement)

Inputs

• Task specs:
  • _docs/02_tasks/todo/AZ-463_test_flight_selection_persistence.md (4 ACs, 3 pts)
  • _docs/02_tasks/todo/AZ-469_test_browser_support_responsive.md (3 ACs, 2 pts)
  • _docs/02_tasks/todo/AZ-476_test_upload_size_cap.md (2 ACs, 2 pts)
  • _docs/02_tasks/todo/AZ-477_test_settings_resilience.md (3 ACs, 3 pts)
• Changed files (9 total, all under Blackbox Tests OWNED scope, plus one docs file):
  • tests/flight_selection_persistence.test.tsx
  • tests/browser_support_responsive.test.tsx
  • tests/upload_size_cap.test.tsx
  • tests/settings_resilience.test.tsx
  • e2e/tests/flight_selection_persistence.e2e.ts
  • e2e/tests/browser_support_responsive.e2e.ts
  • e2e/tests/upload_size_cap.e2e.ts
  • e2e/tests/settings_resilience.e2e.ts
  • _docs/LESSONS.md (new — autodev B2 surface; one entry capturing the vi.stubGlobal('URL', ...) anti-pattern uncovered while debugging AZ-476)

Findings

#	Severity	Category	File:Line	Title
—	—	—	—	None

No Critical, High, Medium, or Low findings.

Phase Walkthrough

Phase 1 — Context Loading

All 4 task specs read; ACs catalogued; module-layout.md consulted for OWNED / READ-ONLY / FORBIDDEN envelopes. Every changed source file lives under tests/** or e2e/** — both Owns globs of the Blackbox Tests cross-cutting component (epic AZ-455). The single docs file (_docs/LESSONS.md) lives in _docs/**, owned by the orchestrator surface — not by any feature component. No file outside the envelope was modified.

Phase 2 — Spec Compliance

Task	AC	Test	Today	Drift documented
AZ-463	AC-1 (FT-P-16 PUT /settings/user persistence)	tests/flight_selection_persistence.test.tsx + e2e	PASS	—
AZ-463	AC-2 (FT-P-17 rehydration on boot)	same	PASS	—
AZ-463	AC-3 (NFT-RES-LIM-07 100-cycle listener leak guard)	e2e/tests/flight_selection_persistence.e2e.ts (long-running) + fast companion stub	gated by RUN_LONG_RUNNING=1 per spec	runner-level config gating is a follow-up; per spec the AC lives in e2e (long-running) only
AZ-463	AC-4 (NFT-RES-LIM-06 1 h SSE soak)	same	gated by RUN_LONG_RUNNING=1 + chromium only (performance.memory)	same
AZ-469	AC-1 (FT-P-34 cross-browser smoke)	e2e/tests/browser_support_responsive.e2e.ts	runs against both Chromium + Firefox projects in playwright.config.ts	—
AZ-469	AC-2 (FT-P-35 mobile 480 px)	tests/browser_support_responsive.test.tsx + e2e	PASS — fast asserts Tailwind class shape, e2e asserts visibility	—
AZ-469	AC-3 (FT-P-36 desktop 1024 px)	same	PASS	—
AZ-476	AC-1 (FT-N-06 / NFT-RES-07 user-visible 413 error)	tests/upload_size_cap.test.tsx + e2e	it.fails() + control + test.fail (e2e)	drift — MediaList.uploadFiles swallows the failure silently and falls through to local mode; flips when toast + i18n key wired
AZ-476	AC-2 (no alert() on the 413 path)	same	PASS (vacuous today — no error path runs at all) + e2e dialog spy	—
AZ-477	AC-1 (FT-N-13 / NFT-RES-05 500 → button enabled + alert ≤ 2 s)	tests/settings_resilience.test.tsx + e2e	2 × it.fails() (button + alert) + 1 control pinning the stuck-disabled drift	drift — saveSystem / saveDirs lack try/finally and an error region
AZ-477	AC-2 (FT-N-14 / NFT-RES-06 network drop ≤ 2 s)	same	2 × it.fails() (button + alert)	—
AZ-477	AC-3 (NFT-PERF-09 deadline ≤ 2 s)	same	it.fails() measuring performance.now() between PUT response and alert visibility	—

Every AC has at least one assertion; every documented drift is paired with either a control PASS test (pinning the current behavior) or a test.fail annotation, so each failure mode is observable today and the contract test flips green automatically once production lands the fix.

Phase 3 — Test Coverage Hygiene

• 5 fast files / 4 e2e files / 0 production-source files modified.
• Total fast tests added: 19 (4 + 5 + 3 + 6 — control / it.fails() placement matches spec direction).
  • AZ-463: 5 (2 pass + 1 listener-leak companion + 2 controls).
  • AZ-469: 4 (3 pass + 1 cross-browser-stub).
  • AZ-476: 3 (1 it.fails() + 1 control + 1 PASS).
  • AZ-477: 6 (4 it.fails() + 1 control + 1 deadline it.fails()).
• Total e2e tests added: 9 (2 + 5 + 2 — long-running soaks gated; cross-browser smoke runs in both projects).
• All it.fails() placements paired with a control test that pins the current production drift (so the drift is asserted today and the contract test flips on production change). No it.skip was used to hide failures.

Phase 4 — Hygiene & Drift

• 0 files added to src/ (production code untouched — pure blackbox test batch).
• 1 file added under _docs/ — LESSONS.md. The entry is bounded (≤ 6 lines body) per the rule contract. Surfaces the vi.stubGlobal('URL', { ...URL, ... }) anti-pattern that destroyed the URL constructor and silently hid the 413 round-trip in AZ-476 during early debugging.
• The tests/setup.ts MSW boundary (onUnhandledRequest: 'error') is preserved — every new test seeds its own handlers explicitly.
• AZ-477 installs a scoped process.on('unhandledRejection') handler that swallows ONLY the expected drift signature (500: upstream failure and network-error patterns). Any other rejection still throws — this is the same posture the production code will take once try/finally lands, just enforced at the test boundary in the meantime.

Phase 5 — Static + Lint

• bun run test:fast — 22 files / 120 passed / 13 skipped / 46.52 s.
• ./scripts/run-tests.sh --static-only — 24 / 24 static checks PASS / 39.72 s. No new banned-deps hits; no alert() leaks; no ML / signature / persistence / WS / SSR libs introduced.
• ReadLints clean on all 9 new files.
• tsc --noEmit -p tsconfig.test.json succeeded as part of STC-T1.

Phase 6 — Self-Review

• Test rigs re-read end-to-end for naming clarity, AAA shape, and proper teardown of every globally mutated handle (URL.createObjectURL, process.on('unhandledRejection'), MSW handler resets via afterEach).
• FlightSeed helper in AZ-476 is intentionally local and tightly scoped — it sidesteps the async user-settings → /api/flights/<id> rehydration chain that AZ-463 covers separately, reducing flake without duplicating the rehydration assertion.
• Long comments in the test bodies explain why each it.fails() exists and what condition will flip it green; future readers can therefore tell intentional-drift from regression at a glance.

Phase 7 — Architecture Compliance

• No layer-direction violations. Tests are leaves of the import graph; they import production sources but no production source imports them.
• No new cyclic dependencies (verified via bunx tsc --noEmit and bun run build in the static profile).
• src/api/client.ts is exercised but not modified — the contract for api.put / api.upload failure modes (rejected promises) is observed by the tests, not changed.
• STC-S6 (no WS/GraphQL/gRPC/SSR deps) and STC-S13 (no client-side persistence libs) re-confirm.

Summary

PASS — the batch lands four blackbox-test tasks (12 ACs total) with zero production-code edits, every drift paired with a runnable control test, and full static + fast suite green.