root/fusion
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

FN-6740: document quarantine ratchet follow-ups

Browse Source 
Document the same-day flaky-test quarantine audit and ownership plan.\n\n- Record the 2026-06-19 quarantine batch expiration date and ledger/config lockstep status.\n- Split follow-up ownership across CLI, core, and dashboard quarantine rescue/delete work.\n- Add an FNXC comment explaining why the coordinated audit should stay scoped by subsystem.\n\nFiles changed:\n docs/testing.md | 4 ++++\n 1 file changed, 4 insertions(+)

Fusion-Task-Id: FN-6740

Fusion-Task-Lineage: eba08900-18d6-4768-ac8a-c673f89537ca
This commit is contained in:
gsxdsm
2026-06-19 14:07:21 -07:00
parent 259cdfb531
commit 2da55a8994
1 changed files with 4 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
docs/testing.md
View File

@@ -199,6 +199,10 @@ Legitimate legacy exceptions must be recorded in `scripts/lib/test-timeout-appea
**2026-06-19 CLI affected-lane rescue (FN-6734):** a broad `@runfusion/fusion` lane reported default 5s test-body timeouts and `fusion-test-workers-*`/fixture `ENOTEMPTY` cleanup noise while isolated files exposed closeable real-store handles and a runtime-dist mock that was sensitive to package-lane module graph ordering. The rescue closed each real `TaskStore`/`AgentStore` before removing its temp fixture, kept task-list truncation coverage under the default timeout by reducing filler size rather than assertions, and preloaded the built `@fusion/core` barrel with `vi.importActual` before `vi.doMock` so complete dist artifacts exercise the CLI surface while partial stale dist skips cleanly. Prove this class with targeted file runs, `pnpm --filter @runfusion/fusion test`, the timeout-appeasement guard, bounded temp-prefix cleanup output, and the normal workspace gate/build; leave the CLI quarantine array empty when no file is actually quarantined.
<!-- FNXC:TestQuarantine 2026-06-19-14:15: FN-6740 audited the same-day quarantine ledger as a coordinated deletion-ratchet batch. The ledger had 14 entries (3 dashboard, 6 core, 5 CLI) and every entry was mirrored in its package Vitest exclude; keep follow-up rescue/delete work scoped by subsystem so ledger/config edits remain lockstep and do not collide. -->
**2026-06-19 quarantine audit (FN-6740):** the 2026-06-19 ledger batch expires on **2026-07-03**. FN-6740 found no ledger/config half-state and chose no inline rescue/delete. The five CLI files (`extension-goal-tools`, `extension-mission-goal-tools`, `extension-task-tools`, `extension`, `research-extension-tools`) are explicitly deferred to FN-6734's outcome and must not get a duplicate rescue task. The six core files (`activity-analytics`, `db`, `store-concurrent-writes`, `store-create-summarize-deferred-hook`, `vitest-teardown-worker-root-cleanup`, `settings-export`) are tracked by FN-6741 as a shared worker-root/temp-redirect cleanup-race follow-up, with the repeat after FN-6600/FN-6486 treated as a product-race signal. The dashboard files are split by likely root cause: FN-6742 owns `session-cross-tab` cleanup `ENOTEMPTY`, FN-6743 owns the third-repeat QuickEntryBox focus-restoration race after FN-6514/FN-6642, and FN-6744 owns WorkflowNodeEditor duplicate-merge-seam concurrency. Until those follow-ups rescue with root-cause evidence or delete under the ratchet, leave all corresponding ledger entries and package excludes in lockstep.
**2026-06-16 rescue (FN-6514):** `packages/dashboard/app/components/__tests__/QuickEntryBox.test.tsx` was rescued before its 2026-06-30 deletion deadline. The file still caught real quick-entry behavior regressions, but it leaked jsdom descriptors for `window.innerWidth`, `window.matchMedia`, `document.visibilityState`, `URL.createObjectURL`, and `URL.revokeObjectURL`; a mobile viewport helper could leave later tests in the same dashboard backfill shard observing `innerWidth=375` and mismatched responsive assertions. The rescue removed the ledger/config quarantine entries in lockstep, captured each original `PropertyDescriptor` at module load, restored those descriptors (or deleted own properties that were originally absent) in `afterEach`, and added a guard test that mutates all rescued globals before asserting they return to their original descriptors. Reusable pattern: any test file that changes jsdom globals with `Object.defineProperty` or spies on replaceable globals must snapshot the original descriptor at the top of the file, restore it in every `afterEach`, and prove the invariant with a guard test; do not use timeout bumps, retries, worker changes, or blanket `vi.restoreAllMocks()` when module mocks depend on stable implementations.
**Gate eviction:** a flake inside the merge gate cannot block all merges while red — it is evicted by removing its line from the `engine-core` allow-list (no quarantine entry needed unless it should also leave the non-blocking tier).