v0.69.0: remove job ignore feature

2026-09-23 22:01:39 +00:00
parent 7d117f6809
commit d2defe4cd3
20 changed files with 48 additions and 171 deletions
@@ -1,5 +1,10 @@
 # Changelog

+## v0.69.0 — 2026-09-23 21:54
+- **Job ignore removed end-to-end.** The "Ignore" button on open jobs (v0.29.0, ISS-047 part B) is gone with everything behind it: the `POST /api/packages/{id}/ignore` route, the `ignorePackage` action, and `'ignored'` from the package status enum (schema, snapshot type, OpenAPI — now twelve endpoints). Packages can only be loaded, delivered, expired, or abandoned.
+- **Legacy 'ignored' rows self-heal.** Production databases already contain rows with `status='ignored'`; nothing ever cleaned them up (`expireOpenPackages` only selected `'open'`). The sweep now selects `IN ('open','ignored')` and flips any row past `createdRealMs + JOB_LIFETIME_MS` to `'expired'`, so every legacy row heals on the first tick after its lifetime lapses (in practice: the first tick after deploy) and no new ones can ever be created. Reads before that first tick are safe: Drizzle does no runtime validation, nothing renders a non-open/loaded/expired package, and `GET /api/state` ticks first anyway.
+- **Tests.** Unit: the `ignorePackage` suite is deleted; terminal-status fixtures drop their redundant `'ignored'` rows (delivered/abandoned already cover "terminal never flips"); the frozen-top-up drain now flips to `'expired'` mirroring `expireOpenPackages`; the taken-slot cooldown ladder keeps two stamped slots, both `'loaded'`. E2E: the Ignore UI test is deleted; `sim.spec.ts`'s soft-lock regression drains via the existing `/api/test/expire-package` hook (loading can't drain a pool — the starter's payload fits only the lightest band) and `expiry.spec.ts` cleans up its created packages through the same hook. OpenAPI endpoint count thirteen → twelve everywhere it is asserted.
+
 ## v0.68.0 — 2026-09-23 20:32
 - **Special remote airports re-curated per #144.** The old remote pass kept EVERY non-large fixed-wing row with no large airport within 1000 km, so regions without big cities (central Australia, Canada edges) got filled with dozens of tiny strips. `scripts/build-world-airports.mjs` now greedily thins the same candidate pool to one special per ~1000 km region: candidates are sorted by priority (scheduled service first, then type rank medium < small < seaplane, then numeric id) and a candidate is accepted only if it is at least `REMOTE_RADIUS_KM` from EVERY already-accepted special — no special airport within 1000 km of another. This subsumes the old 40 km remote clustering (deleted). Regenerated `data/airports_world.csv`: 1483 → 1161 rows (1119 large unchanged, specials 364 → 42; min pairwise special distance 1001.6 km).
 - **Tests.** New committed-dataset invariant in `lib/airports.test.ts` (same style as the existing remote-rule test): every pair of non-large rows is at least `REMOTE_RADIUS_KM` apart (pairwise haversine, O(n²) — n≈42). The existing "no large airport within 1000 km" invariant stays and still holds. No runtime code changes in `lib/airports.ts`.