# Changelog ## v0.66.0 — 2026-09-23 17:06 - **E2E results in the Gitea job summary (#141).** The run page's "Job summaries" panel showed vitest results (vitest's built-in `github-actions` reporter auto-appends `$GITHUB_STEP_SUMMARY` when `GITHUB_ACTIONS=true`) but nothing about the Playwright suite, so a failed e2e run required digging into raw logs to see which specs broke. The CI E2E step now also runs Playwright's built-in `json` reporter (`--reporter=list,json`, output pinned to git-ignored `test-results/e2e.json` via `PLAYWRIGHT_JSON_OUTPUT_FILE` — no new dependencies, no config/TS changes), and a new non-fatal "E2E job summary" step (`if: always()`) parses that report with an inline node script and appends spec-level passed/failed/skipped counts plus the file + title of every failed spec to `$GITHUB_STEP_SUMMARY` — mirroring exactly how vitest gets there. The append is guarded (`|| true`, missing/malformed report = silent no-op) so it can never fail or mask the e2e step's real exit code; test behavior, timeouts, workers and retries are untouched. ## v0.65.0 — 2026-09-23 12:58 - **Per-airport job pools with real-time refresh.** The market was a global bag of jobs (top-up to `MARKET_TARGET_OPEN_PACKAGES = 10` + coverage spawn at uncovered grounded airports), so the board showed work you could never take and hid nothing. Now open jobs are generated PER AIRPORT where a plane can actually pick them up — the pool is the set of grounded planes' `locationIcao`, flying planes' `destIcao` and queued-leg ICAOs — queued legs count even while their plane is in the shop (the work will be waiting at its next stop); a fleet fully repairing with empty queues has no pool, and orphaned pools simply expire out. Each pool airport holds exactly `POOL_TARGET_PER_AIRPORT = 10` open jobs in a fixed weight composition (`POOL_BANDS`: 2×[500,4000) + 6×[4000,18000) + 1×[18000,48000] + 1×(48000,80000] kg), so the board always mixes jobs you can take now with heavier ones that are what you buy better planes for. `spawnPoolPackages` in `lib/sim/engine.ts` replaces both old spawners and runs on every tick (including frozen ones) like v0.64.0's top-up; seeded and spawned packages come from the same new `generatePoolPackage` helper in `lib/game/seed.ts`, so they are indistinguishable — `buildSeed` now seeds a full pool at the starter's home hub, and `PACKAGE_COUNT_MIN/MAX` + `MARKET_TARGET_OPEN_PACKAGES` are retired. - **Open jobs live on the wall clock.** New `created_real_ms INTEGER NOT NULL DEFAULT 0` column (guarded ALTER backfilled with `Date.now()` for existing rows): an open job expires `JOB_LIFETIME_MS = 1 h` after creation regardless of time scale or fleet state — checked every tick OUTSIDE the advance gate, so a paused world still refreshes its board. Loaded jobs keep their sim-clock deadline logic unchanged (delivery check, expiry floor, abandon penalty all untouched). Taken slots (loaded OR ignored) stamp `slot_free_at_real_ms` and hold their pool slot for a refill cooldown — `REFILL_DELAY_MS = 15 min`, or `SMALL_REFILL_DELAY_MS = 1 min` for the <4 t band — so the board visibly refreshes instead of instantly refilling; expired/delivered/abandoned never block. ponytail: one shared per-airport cooldown (the max stamp) rather than a per-band table, add when per-band refill timing matters. - **Visibility + UI.** `/api/state` and SSE now filter open packages to pool origins (`poolAirports` in `lib/game/state.ts`; all non-open statuses pass through unchanged), so the right pane only ever shows jobs you can take — no disabled Load buttons for foreign-origin work. `PackageState` gains `expiresRealMs` (open → `createdRealMs + JOB_LIFETIME_MS`, else null; OpenAPI updated) and the open-job row renders a wall-clock countdown via the existing pure `formatDeadlineRelative` formatter (no new timer — the ~1 s poll re-renders anyway); loaded jobs keep their sim-deadline display. - **Test hooks.** `/api/test/expire-package` rewritten for the wall clock: `{packageId}` ages an OPEN job's `createdRealMs` past its lifetime (`HOOK_AGE_MS = JOB_LIFETIME_MS + 60_000`) and runs a tick, so it comes back already expired (200; unknown/non-open ids 404); no body creates one already-aged hub job (201). CI env: `docker-compose.ci.yml` sets `FLIGHT_JOB_REFILL_MS=5000` + `FLIGHT_JOB_SMALL_REFILL_MS=2000` under the existing `FLIGHT_TEST_HOOKS=1` gate — the lifetime stays 1 h everywhere on purpose, so `state.spec.ts`'s mid-suite container restart can't race a whole board expiring. - **Tests.** Unit: `lib/sim/engine.test.ts` rewritten around the new functions — `expireLoadedPackages (sim clock)` and `expireOpenPackages (v0.65.0 wall clock)` suites (both with inclusive-boundary edges), the frozen-top-up test now asserts exactly 10 open jobs all at EDDF, and a six-test `spawnPoolPackages` suite pins band composition, determinism under an injected RNG + wall clock, pool membership for flying/queued airports (departure airport excluded while its plane is airborne), the taken-slot cooldown ladder, and the full tickOnce pipeline. `lib/game/seed.test.ts`: buildSeed now asserts the exact hub-pool composition (10 jobs, band counts, weights in-band, `createdRealMs` = seed moment) plus a "board always offers jobs beyond the starter" pin; the replay test mirrors `generatePoolPackage`; five full-object equality tests pin `createdRealMs` to avoid millisecond-boundary flakes. E2E: `e2e/sim.spec.ts`'s expiry tests are wall-clock now (hook ages + ticks, then polls the pool back to target — no scale or clock activation); `e2e/right-pane.spec.ts` replaces the LEMD Load-disabled test with an absence assertion (a non-pool-origin open job never reaches `/api/state`); `e2e/hover.spec.ts`'s antimeridian pair moves NZAA→NFTF to EDDF→NFTF (same ±180 eastward crossing, sampled maxJump 359.98) because the hook must create jobs at a pool airport; `e2e/expiry.spec.ts` pins the new hook contract (aged job expired by response time, unknown id 404, already-expired 404). ## v0.64.0 — 2026-09-23 10:06 - **Market keeps topping up while the fleet is grounded (#135 follow-up).** Root cause of a soft-lock introduced by v0.62.0's frozen clock: in `tickOnce` (`lib/sim/engine.ts`) every package-CREATION path (market top-up + coverage spawn) sat behind the same `advanceMs > 0` gate as expiry/delivery/repairs, while removal paths did not — `ignorePackage` runs on any request regardless of advance. A fully grounded fleet therefore froze the clock AND stopped all spawning, so repeated ignores could drain the market below `MARKET_TARGET_OPEN_PACKAGES = 10` (down to zero) and no code path ever created new jobs again: a player with a grounded fleet and an empty market was stuck forever. Fix: only the two spawn calls moved OUT of the gate — `spawnOpenPackages` + `spawnCoveragePackages` now run on EVERY tick, including frozen ones, passing the current (possibly frozen) `clockMs`; `expirePackages`, `deliverPackages` and `finishRepairs` stay inside it. Both spawns were already idempotent no-ops when satisfied (count-gated at target / set-gated by uncovered grounded airports), so advancing ticks behave exactly as before and the clock stays EXACTLY frozen while grounded — only new packages appear, created at game time. No schema change, no migration, no other engine behavior touched. - **Tests.** Unit: `lib/sim/engine.test.ts` — "a frozen tick still tops up the market so a grounded fleet never starves (#135 follow-up)": fully grounded fleet (advance=false), open market drained to zero via direct DB writes, one tick → asserts BOTH that `sim.clockMs` is unchanged (freeze preserved) AND that the open count is back at ≥ `MARKET_TARGET_OPEN_PACKAGES` with ≥1 open package originating at the grounded starter's airport (EDDF coverage). E2E: `e2e/sim.spec.ts` — "a grounded fleet never starves: ignoring every open package cannot empty the market (#135 follow-up)": self-heals to a fully grounded fleet, drains by calling `POST /api/packages/[id]/ignore` for every open package in repeated passes (pre-fix this reached zero and stayed there; post-fix each request's tick re-tops-up before its removal, so it can't), then `expect.poll`s that the open count from `/api/state` is above 0 within ~5 s of real time while asserting `simState.clockMs` stays EXACTLY equal to its pre-drain value and no plane took off — pinning both starvation-fixed and clock-still-frozen. ## v0.63.0 — 2026-09-23 08:43 - **Overlay colors retuned for the Fiord default (ISS-118).** The first half of this issue landed in v0.4.0 (`DEFAULT_STYLE_ID = 'fiord'` + top-bar switcher, e2e already pins `toHaveValue('fiord')`); this completes the second half — auditing every color drawn on the map against Fiord's actual palette. Key finding: OpenFreeMap's Fiord is a DARK cool-toned basemap (land #45516E slate, water #38435C navy — verified byte-identical between the self-hosted `public/styles/fiord.json` and `tiles.openfreemap.org/styles/fiord`), not a light one. WCAG contrast audit of all nine overlay colors against both base surfaces: seven already read fine (≥ 3.3:1 on land AND water) and are untouched — amber flying leg #f59e0b, gold dotted queue line #d4a017 (ISS-098 picked it for light+dark), sky-blue grounded plane + hover halo #38bdf8, green flying plane #22c55e, yellow hover ring/trace #facc15, green target ring #4ade80. Two were weak and got brightened one step within the same hue family: airport dot fill rose-600 #e11d48 (1.68:1 on slate land — the weakest element on the map) → rose-500 #f43f5e (2.15:1), and the hovered-route highlight red-500 #ef4444 (2.10:1, below the 3:1 graphical-object bar for an emphasis layer) → red-400 #f87171 (2.86:1 on land / 3.57:1 on water). White strokes and all other paint values unchanged; no new constants — both colors are single-use inline paint values, matching the existing style of this file. - **Tests.** No test changes needed: `components/map.test.ts` pins only #d4a017 (untouched), e2e specs query rendered features by id/layer rather than color, and `lib/mapStyles.test.ts` already pins `DEFAULT_STYLE_ID = 'fiord'`. ## v0.62.0 — 2026-09-23 07:36 - **Fixed 24x world clock (#135).** Finalized the closed issue #117 spec: the world clock runs at the fixed `WORLD_TIME_SCALE = 24` while any plane is flying or in repair, and stays FROZEN when the whole fleet is grounded — replacing v0.56.0's idle slowdown (`IDLE_WORLD_TIME_SCALE = WORLD_TIME_SCALE / 24`, ISS-128). **Production rate is now exactly 24x regardless of legacy stored values:** `tickOnce` in `lib/sim/engine.ts` computes `advanceMs = advance ? min(elapsed, MAX_TICK_ELAPSED_MS) × scale : 0` where `scale = FLIGHT_TEST_HOOKS === '1' ? sim.timeScale : WORLD_TIME_SCALE` — pre-v0.52.0 saves hold `timeScale: 1000` (the old UI ladder), which is why the live instance still ran at ×1000; only deployments with the env var set honor the stored override written by `POST /api/test/set-timescale` (range [1, 100_000], kept for tests). A frozen tick skips expiry/delivery/repair/spawn entirely and only updates `lastRealTickMs`, so resuming activity never fast-forwards the frozen span. The client mirror (`advanceSimClock` in `lib/useSimClock.ts`) holds its anchor value while grounded, so polls re-anchor without any jump. Removed: `IDLE_WORLD_TIME_SCALE` from `lib/worldClock.ts`, and — **separate finding** — `FLIGHT_TEST_HOOKS: "1"` from the production `docker-compose.yml`: it enabled the `/api/test/*` hooks in prod AND (pre-gate) made the live instance honor its stale stored 1000; e2e runs only in CI against `docker-compose.ci.yml`, which sets the var itself. Unchanged on purpose: the ~1 s background tick (started at boot), the test hook endpoint, and the inert legacy columns `paused`/`lastSelectedTimeScale` (nothing reads them; no migration — per #117 saves reset when multiplayer ships). The OpenAPI description of `/api/state` ("while no plane is flying or in repair the clock stays frozen") now matches the code. - **Tests.** Unit: `lib/sim/engine.test.ts` — both rate branches pinned: with `FLIGHT_TEST_HOOKS=1` stubbed (the file's default, like CI) stored overrides are honored; a new production test unsets the var and proves a legacy stored `timeScale: 1000` still advances at exactly ×24. Grounded-fleet tests pin a FROZEN clock (no idle creep, resume-without-catch-up after takeoff, raised timeScale ignored while grounded); describe renamed to "grounded fleet freezes the world clock (#135)" and pins `WORLD_TIME_SCALE = 24`. `lib/useSimClock.test.ts` — the grounded case asserts staying exactly at the anchor value. E2E: `e2e/sim.spec.ts`'s idle test becomes an exact freeze (delta === 0 after ~2.5 s of idleness); `e2e/state.spec.ts`'s restart test tightens from bounded delta to EXACT clock equality across `docker compose restart`; `e2e/right-pane.spec.ts` verifies the frozen clock, then asserts the abandon-penalty preview matches the server exactly (zero drift under a frozen clock — replaces the DRIFT_MS slope tolerance); `e2e/events.spec.ts` relaxes strict clock advance between SSE frames to non-decreasing (a grounded fleet pushes identical-clockMs frames; advance itself is pinned in sim.spec.ts). The activateWorld/deactivateWorld wrappers are unchanged — a repairing plane still means full scale. ## v0.61.0 — 2026-09-23 07:12 - **Special remote airports (ISS-097).** The world dataset was curated to `large_airport` only, so genuinely empty regions — Antarctica, the high Arctic, Alaska, northern Canada and far islands — had no markers at all. New pure function `isRemote(airports, icao)` in `lib/airports.ts` (reusing the existing `haversineKm` from `lib/geo.ts`): an airport is remote when no other member of the given list lies within `REMOTE_RADIUS_KM = 1000`. The curation in `scripts/build-world-airports.mjs` now runs over ALL fixed-wing rows: every large_airport (unchanged clustering) plus every non-large row that is remote against the kept set; same-station twins (two strips <40 km apart at one base) collapse via the same greedy 40 km clustering. `data/airports_world.csv` grows from 1119 → **1483 rows** (+364 remote airports), including Rothera (EGAR), Halley (EGAH) and Showa skiway (AT25) in Antarctica; Grise Fiord (CYGZ), Alert (CYLT), Ny-Ålesund (ENAS — Svalbard's only marker within 1000 km of a large airport) and Wrangel Island (UHMZ) in the Arctic; Bethel (PABR) and other Alaska strips; Canadian-Arctic smalls; and far islands such as Easter (SCIP), Canton (PCIS), Henderson/Pago Pago (PMDY) and Ascension (FHAW). The only change to a pre-existing row is an upstream `home_link` refresh on CYLW. New airports become contract origins/destinations and map markers unchanged — the most isolated airport in the new set (Easter Island) has its nearest neighbour 2599 km away, well under the 9000 km job cap, so every origin stays flyable; `filterWorldAirports` already accepts small/medium/seaplane types, so no runtime change was needed. - **Tests.** Unit +4 in `lib/airports.test.ts`: a synthetic radius boundary (equator pair at ~945 km vs ~1052 km), real coordinates (Frankfurt↔Amsterdam is not remote; McMurdo vs Christchurch is), an unknown ICAO returning false, and the committed-dataset curation invariant — every non-large row has no large airport within 1000 km and at least one non-large row exists. ## v0.59.0 — 2026-09-23 05:48 - **SSE frames no longer carry package history (ISS-133).** Measured on the live instance first: each `/api/events` frame was compact JSON at ~597 KB, pushed once per real second regardless of time scale (the background tick is a fixed 1 s interval; ×1000 just advances the clock further per frame) — and 99.8 % of it was `packages`: 1,521 rows because terminal packages are never deleted from the DB (expired 564, delivered 528, ignored 386, abandoned 31), while only 12 were `open`. No component ever reads a package with a terminal status — the market list renders `open`, and on-board cargo is `loaded`/`expired` (`LeftPane`, `cargoWeightFor`) — so the history was pure re-sent weight. New `sseFrame(snapshot)` in `lib/game/state.ts` projects a snapshot down to the client-rendered statuses (`SSE_PACKAGE_STATUSES = open|loaded|expired`); both SSE paths use it — the broadcast in `lib/sim/engine.ts` and the initial frame in `app/api/events/route.ts`, so a connect never costs one full-history frame. Live payload drops ~597 KB → ~10 KB per frame (≈ 60×), bounded forever by fleet/market size instead of growing with play time. `/api/state` and action responses keep the FULL package set unchanged, so every e2e oracle that asserts `delivered`/`abandoned` transitions via state lookups still works; the polling fallback stays a strict superset of what SSE pushes. Deliberately NOT done: hash-skip broadcasts (the snapshot's `clockMs` advances on every tick even while idle — `advanceMs = min(elapsed, 2000 ms) × scale ≥ 1 s` — so no two frames are ever equal), throttling (already exactly 1 Hz), and a diff/patch engine. - **Tests.** Unit +3 in new `lib/game/state.test.ts`: the projection keeps exactly open/loaded/expired (order preserved, terminal rows dropped), passes account/simState/planes through untouched without mutating its input, and pins the status set itself. E2E: `e2e/events.spec.ts` now records every package status seen in any frame plus the largest frame — asserts no terminal status ever appears and frames stay under 200 KB (a full-history regression at ~597 KB fails loudly). ## v0.58.0 — 2026-09-22 21:27 - **Plane markers no longer teleport across the antimeridian (ISS-124).** The follow-up to ISS-120's known limitation: a plane mid-crossing still used the snapshot's wrapped lat/lon (`positionAt` wraps every longitude into [−180, 180]), so its marker jumped +179.x → −179.x between polls — a visible 360° teleport on the map while its route line stayed continuous. `planesGeoJson` in `lib/mapData.ts` now takes the airport index (`airportsByIcao`, same parameter shape as `routesGeoJson`) and re-expresses each FLYING plane's longitude against its departure airport — the same anchor `greatCircleArc` keeps for its first sample — via a new module-private `unwrapLon(lon, prev)` (the ±360 shift until |Δlon| ≤ 180 that `unwrapLons`'s loop body already used; `unwrapLons` now calls it). The unwrap applies only when BOTH leg endpoints exist in the index, mirroring `routesGeoJson`'s guard exactly: a marker is unwrapped if and only if its line is drawn. Grounded planes, repairing planes and flying planes with a missing endpoint keep their raw wrapped position (unchanged behavior); non-crossing legs are byte-identical (no |Δlon| > 180 → the shift never fires). `components/Map.tsx` passes `airportsIndexRef.current ?? new Map()` at both call sites (`applyDynamic` and the source bootstrap in `ensureLayers`) — planes draw immediately at their wrapped position until the airport index loads, then unwrap. - **Tests.** Unit +6 in `lib/mapData.test.ts` (new "antimeridian marker unwrap" block on the real NZAA/NFTF pair): eastward crossing unwraps a mid-leg −179.5 to +180.5 against the origin's +174.8, westward (reversed pair) unwraps below −180, and — the key invariant — every sample of `greatCircleArc(NZAA→NFTF)` fed back as a WRAPPED snapshot position comes out exactly on that arc's own sample, so marker and line agree frame to frame; grounded planes keep their raw coordinate, a flying plane with a missing endpoint stays wrapped (no line drawn either), and non-crossing legs are untouched. E2E +1 in `e2e/map-interactions.spec.ts`: the starter is relocated to Auckland via a fewest-legs chain over `/api/airports` (each hop ≤ 3400 km — within the ATR72's guaranteed range even in worst-case headwind; the tank is topped for free with the `set-fuel` hook before every dispatch so auto-refuel costs nothing), then a hook-created NZAA → NFTF package is loaded and flown at scale 1000 while the marker's longitude is sampled from the planes source every ~500 ms (real time) until the server reports landing — no two consecutive samples may differ by ≥ 180° (the teleport), some sample must sit on the east world copy (lon > 180, proving the crossing was actually traversed unwrapped rather than skipped), and the sampled span must exceed 4°. The plane chains back home afterwards so the other specs start from the canonical state. - **E2E fix (CI run 176).** The new antimeridian test failed deterministically in CI — not a marker bug, but a query trap: MapLibre's `querySourceFeatures` only returns features in LOADED tiles, and GeoJSON sources load per viewport tile; the default camera sits over Europe while NZAA/NFTF are in the Pacific, so the plane feature never appeared in queries and the "flying" poll timed out (the same pattern passes for EDDF→LEMD because that leg is inside the default view). The test now fits the camera over both endpoints first via `fitToAirports` (copied from `e2e/hover.spec.ts`, including its ISS-120 unwrap so a ±180-straddling pair bounds the Pacific, not the Atlantic) before polling and sampling. Pollution guard: relocation + crossing now run in a `try/finally` whose cleanup ALWAYS runs — it waits for any in-flight leg to land (scale 1000), then chains the plane back to EDDF from wherever it is, so a failed poll can no longer leave the starter FLYING and 409-cascade into the eleven later specs' fly-home helpers (the run-176 cascade). The dispatch POST's status is now checked explicitly with a clear message instead of a bare `expect(...).toBe(200)`. Unit +1 in `lib/mapData.test.ts` pins the exact feature properties the e2e `starterMarker` matches on (`properties.id === 'plane-starter'`, `properties.flying === true`). - **E2E fix (CI run 177).** The antimeridian test still failed — again not a marker bug: by dispatch time the test's page had been alive ~3 min through the relocation chain, and under dind load the browser's state feed can go stale (the SSE fallback in `lib/useGameState.ts` only re-arms polling on stream error or a one-shot 5 s silence check at connect), leaving the planes source frozen on an old FLYING frame from a relocation hop — which fooled the "wait for flying" poll and made every sampled longitude a stale pre-crossing value (continuous, all ≤ 180, never > 180). The test now reloads the page after relocation (server state persists; on the fresh page only the crossing leg can ever produce a flying frame), and the sampling loop gains a freshness gate: a sample is recorded only if its marker position is within ~700 km of where the SERVER snapshot says the plane is (~5 s of flight at ×1000 — haversine is ±360°-periodic in Δlon, so wrapped server lon vs unwrapped marker lon compare directly), so a frozen frame can no longer pass continuity vacuously; `starterMarker` now also returns latitude for that check. Pre-existing flake fixed in the same run: `e2e/queue.spec.ts`'s chain-refuel lower bound loosened from `1.4 × required2` to `1.3 × required2` — under dind load the post-departure snapshot can land ~0.1 × required2 into leg 2's burn (observed miss: 0.07 %); a wrong top-up level still lands far below the bound, and the exact level remains pinned by the upper bound plus the rep-discounted cash assertion. - **Root cause of the stale-feed freeze fixed in `lib/useGameState.ts` (CI run 178).** The reload + freshness gate above were insurance; run 178 proved the feed can freeze even on a FRESH page — and the reason is now fixed at the source, not worked around. The SSE silence check was one-shot: `setTimeout(startPolling, 5 s)` armed once at connect, cleared by the FIRST event and never re-armed. After that, only a stream ERROR could restart polling — but a half-open connection that simply stops delivering data (observed under dind load; `/api/state` on separate connections kept answering fine throughout) fires no `onerror`, so state froze until reload. The timer is now re-armed on EVERY message: any 5 s gap in the event flow — whenever it happens, not just before the first event — starts the polling fallback (immediate catch-up fetch + interval), which stops again when fresh events resume. Healthy streams pay one set/clearTimeout per event; stalled ones degrade to the pre-v0.57.0 poll loop instead of freezing. The e2e keeps both insurance layers: the reload still guarantees a clean SSE connect at dispatch, and the freshness gate still fails loudly if any future stall outlasts it (a frozen source can no longer pass continuity vacuously). - **Tests.** Unit +1 in `lib/mapData.test.ts` ("NZAA→NFTF crossing leg — full pipeline on stored dataset coords"): runs the exact production pipeline on the STORED CSV coordinates (`loadAirportsGeoJson` → `positionAt` slerp sampled at 51 steps → `planesGeoJson`) and pins what a live browser source must show — marker lons monotonically increasing from +174.8 to +184.9 (the unwrapped NFTF), every consecutive pair < 180° apart, max > 180 (eastward crossing onto the world copy). This is the ground truth the e2e's `lon > 180` assertion encodes: the dataset crosses EASTWARD (NZAA +174.786 → NFTF −175.149, Δ ≈ +10°), so a correct pipeline MUST produce lons above +180 — and any e2e failure of that assertion is now provably a stale browser source, not the math. ## v0.57.0 — 2026-09-22 19:19 - **Push updated state to clients over SSE (ISS-129).** Clients no longer rely on pure polling for live updates: new `GET /api/events` streams full game-state snapshots as Server-Sent Events — the current snapshot immediately on connect, then one after every sim tick (~1 s), so the UI updates push-driven even while idle. New module `lib/sse.ts` is a tiny in-process fan-out registry (`subscribe(fn)` → unsubscribe, `broadcast(text)` with per-subscriber try/catch) — the three process-wide singletons (this subscriber set, the sql.js DB promise (`getGameDb`), and the tick lock (`withLock`)) live on `globalThis`: Next.js can load a module more than once per process (instrumentation.ts and route handlers get separate module instances), so plain module-level state would give each layer its own registry/DB/lock, and background-tick pushes would never reach route-handler subscribers. The broadcast hook lives in `tickNow()` in `lib/sim/engine.ts`: after each locked tick on the singleton path it serializes via the SAME `getStateSnapshot()` that `GET /api/state` returns (no duplicated serialization) and pushes to all subscribers; failures are logged, never fatal. Action mutations reach other clients within one background tick (~1 s); the acting client still gets its immediate post-action snapshot from the POST response as before. The SSE route (`app/api/events/route.ts`) sends `Content-Type: text/event-stream`, `Cache-Control: no-cache`, `Connection: keep-alive` and `X-Accel-Buffering: no` (prod sits behind Traefik), plus a `: ping` heartbeat comment every 15 s so idle proxies don't drop the connection; snapshot-first-then-subscribe ordering guarantees events are strictly ordered by sim clock, and stream cancel unsubscribes + clears the heartbeat; broadcast pushes are wrapped in `data: \n\n` SSE frames at the route (matching the initial snapshot). Client (`lib/useGameState.ts`): one `EventSource('/api/events')` per page — onmessage parses the full snapshot into the shared state (same shape as today); polling stays as an automatic fallback: if no event arrives within 5 s of connect, or on a stream error, the old poll loop starts (immediate catch-up fetch + interval), and it stops again when fresh events flow. EventSource's native reconnect is relied on — no custom reconnect framework. `/api/state` behavior is unchanged (still ticks-then-snapshots; polling remains fully functional). - **API.** OpenAPI spec gains `GET /api/events` (200 → `text/event-stream`) in `lib/openapi.ts`, so the machine-readable spec at `/api/openapi` stays accurate. - **Tests.** Unit: new `lib/sse.test.ts` — subscribe/broadcast/unsubscribe semantics, a throwing subscriber doesn't break others, idempotent unsubscribe; new `app/api/events/route.test.ts` (throwaway-DB pattern from the state route test) — SSE headers + first chunk parses as a snapshot with `simState.clockMs`. E2E: new `e2e/events.spec.ts` (+1) — node-side fetch of `/api/events` against the app under test asserts content-type `text/event-stream`, the first event parses as a valid snapshot (`simState.clockMs`), and within ~5 s at least one MORE event arrives with an advanced clock (background tick pushes while idle since v0.56.0 — no scale hook); stream closed in `finally`. The exact-endpoint-list assertions in `lib/openapi.test.ts` and `e2e/api.spec.ts` move twelve → thirteen endpoints (`/api/events` added). ## v0.56.0 — 2026-09-22 18:28 - **Idle fleet slows the world to real time instead of freezing it (ISS-128).** A fully grounded fleet no longer freezes the sim clock — `tickOnce` in `lib/sim/engine.ts` now advances by `min(elapsed, MAX_TICK_ELAPSED_MS) × IDLE_WORLD_TIME_SCALE` when nothing is flying or in repair, where new constant `IDLE_WORLD_TIME_SCALE = WORLD_TIME_SCALE / 24` (= exactly 1, i.e. real time) lives next to `WORLD_TIME_SCALE` in `lib/worldClock.ts`. Packages keep spawning/expiring and the market stays fresh while idle; an active fleet keeps its full scale including any FLIGHT_TEST_HOOKS override (grounded + hook scale 1000 → advances at ×1, not ×1000), and `lastRealTickMs` still always tracks real time so resuming activity never fast-forwards the idle duration. The client mirror (`advanceSimClock` in `lib/useSimClock.ts`) extrapolates at the same idle rate instead of freezing, so polls re-anchor without a visible jump. - **Tests.** Unit: `lib/sim/engine.test.ts` — the grounded-fleet tests now pin real-time advance clamped to one idle-rate tick (no freeze, no fast-forward), resume-without-catch-up after takeoff, and that a raised timeScale is ignored while grounded; new describe "idle slowdown" pins `IDLE_WORLD_TIME_SCALE = WORLD_TIME_SCALE / 24 = 1`. `lib/useSimClock.test.ts` — the frozen-while-idle test now asserts real-time advance at the idle rate. E2E: `e2e/sim.spec.ts`'s freeze test becomes a bounded real-time delta (advances > 0, ≥ half the wall span clamped to one tick, ≤ wall span + one clamp, clearly below ×24); `e2e/state.spec.ts`'s restart test replaces byte-identical clock equality with derived bounds — never backwards (not re-seeded) and ≤ `MAX_TICK_ELAPSED_MS` of clamped downtime + the live wall span; `e2e/right-pane.spec.ts`'s penalty preview no longer races a frozen clock — it verifies the idle rate, then allows the formula's slope over one UI poll interval (~1 s at ×1) instead of an exact string match. The `activateWorld`/repair wrappers in actions/panes/sim specs are unchanged (a repairing plane still means full scale); only their "freezes" comments were updated. ## v0.55.0 — 2026-09-22 15:38 - **Favicon (ISS-122).** The app shipped with no favicon at all (`app/layout.tsx` set only title/description; no `app/icon.svg`, no `public/favicon.ico`). New hand-written single-file `app/icon.svg` — a bold white top-down airplane silhouette (fuselage + swept wings + tail, one straight-line path) on a pastel blue (#93c5fd) rounded square, 64×64 viewBox. Next App Router's file convention auto-serves it at `/icon.svg` and injects `` into the head; no layout/metadata changes needed. - **Tests.** E2E +1 in `e2e/panes.spec.ts`: page load asserts exactly one `head link[rel~="icon"]` pointing at `/icon.svg`, and a direct `GET /icon.svg` returns 200 with content-type `image/svg+xml`. No unit test — static asset. - **E2E hardening (CI run #167).** Two e2e-only fixes: the favicon assertion now tolerates the content-hash query string Next App Router production builds append to the auto-injected icon `` (`/icon.svg?icon..svg` — matched with `/^\/icon\.svg/`, canonical `GET /icon.svg` check unchanged), and `e2e/hover.spec.ts`'s local `flyHomeViaApi` self-heal is replaced by the established `e2e/panes.spec.ts` pattern (set-health repair-gate reset + re-read state AFTER an in-flight leg lands) — the old version checked `locationIcao` against the stale pre-landing snapshot, so a retry after a mid-flight interruption left plane-starter grounded at the destination and its package load 409'd ("plane not at the origin"). - **E2E hardening (CI run #168).** Retry-cascade fix in `e2e/actions.spec.ts`: its local `flyHomeViaApi` self-heal now also tops plane-starter's tank back to full via the free `POST /api/test/set-fuel` hook after repositioning home. A transient `/api/state` 500 (issue #127) killed attempt 1 mid-flight; the retry's self-heal then flew the starter home with a low tank, so the test's own dispatch auto-refueled AFTER its cash baseline was captured and the exact-cash "no reward for expired cargo" assertion failed by €1190.56 — a full tank makes every subsequent dispatch top-up cost exactly €0. - **E2E hardening (ISS-127).** e2e `getState` hardened against transient `/api/state` 500s — internal retry ×3/500 ms keeps poll callbacks total. ## v0.54.0 — 2026-09-22 15:20 - **Long priority jobs are now always finishable (ISS-119).** `generatePackage` in `lib/game/seed.ts` drew a flat 6–72 h expiry window (priority ×0.5 → 3–36 h) independent of leg distance, so some jobs' deadlines were shorter than the flight itself and could never be completed (legs go up to MAX_JOB_LEG_KM = 9000 km; the fastest catalog plane is ~920 km/h ≈ 10 h for a max leg). The drawn window is now clamped UP after special scaling: deadline ≥ now + `travelDurationMs(legKm, FASTEST_CATALOG_SPEED_KMH)` + `EXPIRY_BUFFER_MS` (flat 1 h), reusing the existing `travelDurationMs` from `lib/sim/movement.ts`. New constant `FASTEST_CATALOG_SPEED_KMH` in `lib/game/catalog.ts` — data-driven max of `PLANE_TYPES` speeds, so a job must be finishable with the best available aircraft (NOT the starter's 510 km/h, which would wreck long-job economics); adding a faster plane tightens the floor automatically. `EXPIRY_BUFFER_MS` lives next to the deadline tunables in seed.ts. The window is a single uniform draw, so clamping up covers both requirements: short legs (drawn × scale already above the floor) are byte-identical to before, and when the raw draw falls below the floor the deadline lands exactly on it — which stays ≤ the scaled ceiling for any pair of airports on Earth (max leg ~20000 km → floor ≈ 22.8 h < the 36 h priority ceiling), so no separate ceiling raise is needed. Forced `deadlineOffsetMs` skips the floor, same as it skips special scaling (the e2e expiry hook keeps working). - **Tests.** Unit +5 in `lib/game/seed.test.ts` (new "expiry floor" block): pins the constants (fastest = max of catalog speeds and > starter speed; buffer exactly 1 h); a seeded sweep over 200 unforced packages on the real dataset asserting deadline ≥ now + travel time at max speed + buffer for every one (specials roll naturally, ~15% priority); the farthest within-cap pair in the real dataset (~9000 km) forced with both `priority` and null across 50 seeds each — invariant holds everywhere AND at least one package per special lands exactly on the floor (deterministic for fixed seeds, proving the clamp engages rather than holding vacuously); short legs (EDDF→EHAM ≈ 370 km → floor ≈ 1.4 h < the priority minimum window) pinned byte-identical to a replay of the pre-v0.54.0 draw order (no economy regression); forced deadline offsets still skip the floor. E2E +1 in `e2e/expiry.spec.ts`: the EXISTING `POST /api/test/create-package` hook forces EDDB→KSMF (~9000 km, at MAX_JOB_LEG_KM) with `special: 'priority'`; 10 packages are created and every one asserted to expire ≥ travel time at max speed + buffer after creation (and ≤ the 36 h priority ceiling), then all ignored for cleanup — no new server hooks. ## v0.53.0 — 2026-09-22 14:47 - **Antimeridian routes no longer chord across the map (ISS-120).** Routes whose great circle crosses ±180° (e.g. Auckland → Fua'amotu, Tonga) rendered as a line spanning the whole map: `greatCircleArc` sampled `positionAt`, which wraps every sample's longitude into [−180, 180], so consecutive samples jumped +179.x → −179.x and MapLibre — which draws LineStrings as-is with no ±180 splitting — connected them across the globe. New module-private `unwrapLons` in `lib/mapData.ts` re-expresses each arc's sample sequence in continuous longitudes (first sample keeps its raw lon; every later one is shifted by ±360 relative to the previous UNWRAPPED value until |Δlon| ≤ 180), applied inside `greatCircleArc` so all four line builders (`routesGeoJson`, `queuedRoutesGeoJson`, `hoverRouteGeoJson`, `hoverTraceGeoJson`) are fixed at once. One line stays one LineString — no splitting; out-of-range values render on MapLibre's adjacent world copies (`renderWorldCopies` is default-on). `queuedRoutesGeoJson` gets a second global pass after concatenation: each leg's arc unwraps relative to its OWN start airport, so at a joint the sequence can jump ±360 (leg 1 ends at dest+360k, leg 2 restarts at the raw dest) and the re-unwrap keeps multi-leg chains continuous. Non-crossing routes are byte-identical (no |Δlon| > 180 anywhere → the pass is a no-op). - **Tests.** Unit +4 in `lib/mapData.test.ts` (new "antimeridian crossing" block, real dataset pairs NZAA/NFTF/NZCH): eastward and westward crossings stay continuous (every consecutive |Δlon| < 180), endpoints exact mod 360 with the first sample still the raw origin, sampled chord length ≈ haversine within 1%, a non-crossing route's lons stay in [−180, 180], and a grounded plane's two-leg queue chain crossing ±180 stays continuous across the joint (the regression `queuedRoutesGeoJson`'s global pass guards). E2E +1 in `e2e/hover.spec.ts`: hovering an antimeridian-crossing package (hook-created NZAA → NFTF) draws one continuous trace — no rendered segment may contain a ±180 longitude jump; `expectHoverRoute`'s endpoint check now compares longitudes on the circle (mod 360, identical for any pair within ±180 so existing routes are unaffected) and `fitToAirports` unwraps crossing pairs so their bounds follow the short arc across the date line instead of spanning the wrong side of the globe. - **Known limitation (follow-up).** Plane MARKERS still use the snapshot's wrapped lat/lon (`lib/game/state.ts` → `positionAt`), so a plane mid-crossing teleports 360° on its marker at the instant it passes ±180 while its route line stays continuous. Fixing needs airport coordinates inside `planesGeoJson` (signature change + `Map.tsx` call site) — deliberately out of scope for this fix. ## v0.52.0 — 2026-09-22 10:45 - **Fixed world clock, no player time controls (ISS-117).** The pause button and the 1×–1000× scale slider are gone from the left pane (`components/LeftPane.tsx`) — the world now always runs at a fixed `WORLD_TIME_SCALE = 24` (new client-safe `lib/worldClock.ts`, re-exported by the engine). The server advances time in a background tick: new `instrumentation.ts` starts `startBackgroundTick()` (1 s interval, singleton-guarded) under `NEXT_RUNTIME === 'nodejs'`, so the sim moves even with no open browser. **Idle-pause:** while no plane is flying or in repair the clock is frozen exactly — pure `shouldAdvanceTime(planes)` in `lib/sim/engine.ts` gates every tick (no catch-up jump on wake; `lastRealTickMs` still updates). Fresh worlds start at year 2000 (`STARTING_CLOCK_MS = Date.UTC(2000, 0, 1)` in `lib/game/seed.ts`; existing saves untouched — no migration). The client clock (`lib/useSimClock.ts`) extrapolates between polls at the fixed rate and freezes while idle, mirroring the server. - **API.** `POST /api/sim` deleted (route + OpenAPI entry; spec now lists twelve endpoints) and its controls removed from the engine (`applyTimeControls`, `clampTimeScale`, MIN/MAX_TIME_SCALE). The `/api/state` snapshot's `simState` is now `{ clockMs }` only — no `timeScale`/`paused`. New test-only hook `POST /api/test/set-timescale` behind the existing `FLIGHT_TEST_HOOKS=1` gate (sibling of `/api/test/set-health`, same 404-when-disabled convention): body `{ timeScale: finite number in [1, 100_000] }` else 400; ticks first under the sim lock and returns `{ clockMs, timeScale }`. The e2e suite's per-spec `setScale` helpers now call it (and verify from the response); every spec restores the world rate in `afterEach`. - **Tests.** Unit: engine tests re-anchored to `STARTING_CLOCK_MS` with a `makeWorldActive` helper (fresh DBs seed a grounded starter, so advancing-world tests put one in flight first) plus new idle-pause coverage (`shouldAdvanceTime` synthetic arrays; grounded fleet freezes the clock exactly; repairing keeps it advancing); actions/seed/db/openapi tests re-anchored to the year-2000 origin and fixed world rate; `lib/timeScale.ts`, its test, `components/left-pane.test.ts` and the old `/api/sim` route + test deleted. E2E: all specs rewritten off `POST /api/sim` — pause tests replaced by idle-pause assertions (e.g. `right-pane.spec.ts` verifies a double-read 1.5 s apart that a grounded fleet freezes the clock exactly, which also makes its penalty-preview math deterministic); `sim.spec.ts` activates the world via set-health + `/repair` for its advance/expiry tests and pins the new hook's validation; `panes.spec.ts` drops the slider-commit test (controls gone) and samples the smooth client clock at the fixed 24× rate, asserting the pause/slider testids are absent; `state.spec.ts` restart test self-heals to a fully grounded fleet first and now asserts the post-restart clock is byte-identical (frozen across the restart); `two-planes.spec.ts` loses its ISS-057 slow-on-landing assertions — the auto-chain observation leg now flies at world rate so polls can see it; `api.spec.ts` endpoint list drops `/api/sim` (thirteen → twelve). ## v0.51.0 — 2026-09-22 08:21 - **Max range display is now guaranteed flyable in all weather (ISS-099).** The spec card's "Max range" (`components/RightPane.tsx`) showed `maxRangeKm` — the weighted-model closed form with NO wind — while dispatch's fuel guard (`flyPlane`, and the engine's auto-chain) uses wind-adjusted burn (up to +20% headwind, v0.36.0), so a leg inside the displayed range could 409 "insufficient fuel" when the departure airport had a headwind. New `MAX_WIND_BURN_FACTOR` (= 1.2 — pinned in tests to be exactly `windBurnFactor`'s maximum) and `guaranteedRangeKm(burnPerKm, emptyWeightKg, maxFuel)` = `maxRangeKm` under worst-case headwind burn (`lib/sim/movement.ts`). The UI now shows the guaranteed range: any leg within it is flyable in all weather. ATR 72-600: 4436 → 3696 km; Boeing 777: 13045 → 10871 km — matching the `maxRangeKm ÷ 1.2` convention seed.ts already uses for its job-generation cap - **Tests.** Unit +15: new `guaranteedRangeKm` block in `lib/sim/movement.test.ts` (3 tests — MAX_WIND_BURN_FACTOR pinned to `windBurnFactor(30, 90, 90)`, ATR ≈ 3697 km ±5%, zero-fuel/invalid-weight guards) and a per-type consistency loop in `lib/game/catalog.test.ts` (12 tests — for ALL six catalog types: `requiredFuelFor(guaranteed range)` ≤ maxFuel under worst-case headwind, tightness at 1.05× > maxFuel, and the no-wind closed form still an exact inverse). E2E: `e2e/right-pane.spec.ts` market-tab test now imports and asserts `guaranteedRangeKm` directly (local mirror deleted) ## v0.50.0 — 2026-09-22 08:00 - **Queued/planned route line is now amber/gold instead of gray (ISS-098).** The dotted queue line in `components/Map.tsx` (a flying plane's not-yet-started legs, ISS-043) was drawn in muted slate `#64748b`. It is now an amber/gold `#d4a017`, readable on both light and dark tiles. The color was extracted from the inline literal to a new exported module-scope constant `QUEUE_ROUTE_LINE_COLOR` so it is testable; the dashed/dotted style (`line-dasharray [0.1, 1.5]`) and width (3) are unchanged - **Tests.** Unit +1: new `components/map.test.ts` pins `QUEUE_ROUTE_LINE_COLOR === '#d4a017'`. No new e2e spec — the existing queued-route rendering paths (`e2e/queue.spec.ts`, `e2e/relocation.spec.ts`) already exercise this layer, and a pure color swap needs no behavioral coverage ## v0.49.0 — 2026-09-22 07:25 - **Airport popup Relocate now works for flying/repairing planes (ISS-096).** The Relocate row in the map's airport popup (`components/Map.tsx`) previously listed only grounded planes and showed "No grounded planes to relocate." when none was available. It now lists ALL planes from the latest snapshot, and the click handler picks the endpoint by the SELECTED plane's current status: `available` → `POST /api/planes/{id}/fly` (immediate relocation, unchanged), `flying`/`repairing` → `POST /api/planes/{id}/queue` (the destination is appended to the plane's flight queue and chains on landing). Success text is "Relocation started." for fly and "Added to queue." for queue; the no-planes fallback guard is reworded to "No planes to relocate." No API changes — both endpoints existed since v0.25.0/v0.46.0 - **Tests.** E2E +1 (`e2e/relocation.spec.ts`): dispatches the starter plane on a leg via the API so it is flying, slows the sim to 1x so the leg cannot land mid-test, opens the airport popup for a different airport, selects the flying plane from the (now full) list and clicks "Relocate here", then asserts via state that the plane's `queue` contains the clicked airport's ICAO while it is still flying its original destination ## v0.48.0 — 2026-09-22 06:55 - **Leg distance shown on contracts (ISS-095).** `PackageState` gains `distanceKm` (`lib/game/state.ts`) — the great-circle origin→dest distance, computed server-side in `getStateSnapshot()` from the airport GeoJSON already loaded for live-plane math (same `haversineKm` as flight movement; 0 if a coord is missing). Shown in two places via new `formatKm()` (`lib/format.ts`, integer km, no space before "km"): the on-board contract line (`EGCC → GMME 554km €23,826.00 …`) and the job-list row (`€37,408.00 215km Perishables 0.4t …`). OpenAPI spec updated in lockstep; nothing stored - **Tests.** Unit: `app/api/state/route.test.ts` asserts every package's `distanceKm` equals the haversine of its airports' coordinates (computed with the same lib/geo helper); `lib/openapi.test.ts` pinned key list gains `distanceKm`; new `formatKm` test in `lib/format.test.ts`. E2E: `right-pane.spec.ts` +2 assertions (job-row distance span matches /^\d+km$/, generic and hook-created package), `actions.spec.ts` +1 (on-board line) ## v0.47.0 — 2026-09-21 23:01 - **Job payoff retuned so weight dominates (ISS-090).** `REWARD_PER_KG` in `lib/game/seed.ts` rises from €0.5/kg to €2.0/kg — the linear form is unchanged, only the constant moves. Before, the weight term was small vs base+distance, so heavier jobs barely paid more (a 1000 km leg: €3,250 @500 kg → €5,000 @4,000 kg — per-kg rate fell from €4.59 to €1.01) and there was no sense in taking heavy jobs; after the retune the same leg pays €4,000 / €11,000, so weight clearly drives payoff. Safe for viability: fuel burn (`lib/sim/movement.ts`) is weight-independent, so this only raises income - **Tests.** `seed.test.ts`: stale test name "… 0.5 * weightKg …" made generic; new literal pins in the `rewardFor` formula test — `rewardFor(1000, 4000) === 11_000` and `rewardFor(1000, 500) === 4_000` — so future retunes are deliberate. All other reward assertions recompute through the exported constants and pass unchanged ## v0.46.0 — 2026-09-21 22:32 - **Queue consumption on dispatch (ISS-088).** `flyPlane` (`lib/game/actions.ts`) now consumes `queue[0]` when the requested destination IS the first queued leg, so a landing never chains a zero-distance self-hop (the "phantom hop" — a→a completing instantly before reaching the real next leg). Flying to a non-first queued leg stays an explicit override and leaves the queue untouched; the persisted queue is written back in the same update - **Engine auto-chain hardening.** `deliverPackages` (`lib/sim/engine.ts`) gains two guards: (1) defensive skip — if `queue[0]` equals the arrival airport (only reachable from legacy/corrupt rows now that dispatch consumes), it drops the leg without flying; (2) full-tank backstop — if the next queued leg does not fit even a full tank, the plane grounds as `'available'` with the queue kept intact instead of chaining an unflyable leg (mirrors `flyPlane`'s 409 without throwing mid-tick) - **Selling a queued plane is rejected.** `sellPlane` (`lib/game/actions.ts`) returns 409 `the plane has queued legs — cancel queued legs first` for a grounded plane with a non-empty queue, instead of silently discarding the queued legs - **Test hook: `/api/test/set-cash`.** New route under `app/api/test/` (sibling of `set-fuel`/`add-cash`) setting account cash to an ABSOLUTE value `{ amount }` — same `FLIGHT_TEST_HOOKS=1` gate, strict JSON validation, post-tick state snapshot. Deliberately NOT in the OpenAPI spec: test hooks are out-of-spec by convention and `lib/openapi.test.ts` pins exactly 13 public paths - **Tests.** Unit +5 (373 total): `flyPlane` consumes queue[0] on dispatch / skip-ahead keeps the queue intact; `sellPlane` 409s a grounded plane with queued legs leaving the row untouched; engine full-tank backstop grounds with the queue intact (FGSL ~4267 km from LEMD — beyond ATR range); engine defensive self-leg drop. E2E: `e2e/queue.spec.ts` +3 — three queued legs fly in FIFO order across landings (dispatch response snapshot proves queue[0] consumed), a mid-flight drained tank is topped up before the chained leg at exactly the rep-discounted cost (expectations computed with the same lib helpers as the engine, via `/api/test/set-fuel` drain + new `set-cash` hook), cancelling the only queued leg mid-flight leaves a plain landing with no auto-chain; `e2e/state.spec.ts` +1 — a staged two-leg queue survives `docker compose restart flight` byte-identical; `e2e/two-planes.spec.ts` reworked — the 240 s poll that absorbed the phantom hop is replaced by explicit assertions: the fly response snapshot shows only `[b]` queued, and every state observed while waiting for the chain must never show location a with dest a ## v0.45.0 — 2026-09-21 21:12 - **Broken planes cannot fly (ISS-089).** A plane at `health ≤ 0` is now grounded until repaired instead of flying on fumes: `flyPlane` (`lib/game/actions.ts`) rejects it with 409 `the plane is broken (health 0) — repair it first` BEFORE any auto-refuel math, so fuel and cash stay untouched; the engine's auto-chain (`deliverPackages`, `lib/sim/engine.ts`) skips chaining when a landing's wear breaks the plane — it stays grounded at the arrival airport with its queue kept intact for a later dispatch (same shape as an unaffordable top-up) - **Repair is gradual.** `repairPlane` no longer jumps health to 100 — it grounds the plane in `'repairing'` with the gate at `clock + REPAIR_DURATION_MS` and leaves health where it was. The engine's `finishRepairs` pass (`lib/sim/engine.ts`) now heals along a linear curve each tick: `health = min(100, max(startHealth, round(f × 100)))` with `f` the fraction of the repair duration elapsed — monotonic (never drops below the start value), and the flip to `'available'` at exactly 100 happens when the gate passes. The left-pane health bar doubles as the repair progress indicator; no UI change - **REMOVED the broken-plane ×1.5 burn mechanic.** `BROKEN_BURN_MULTIPLIER` + `conditionBurnMultiplier()` are deleted from `lib/game/catalog.ts` and every burn site reverts to plain wind-adjusted consumption: `flyPlane` dispatch requirement, engine landed fuel + auto-chain next-leg check (`lib/sim/engine.ts`), the live in-flight gauge (`livePlane`, `lib/game/state.ts`). The mechanic was unreachable anyway — broken planes can no longer depart - **Docs.** OpenAPI reworded to match: fly's 409 lists the broken case, queue/repair descriptions + the `PlaneState.health` field note the gradual heal and the cannot-fly rule; route comment in `app/api/planes/[planeId]/repair/route.ts` updated. No endpoints added or removed — count pins unchanged - **Tests.** Unit: new `flyPlane` broken-guard test (409 + fuel/cash untouched) and a worn-but-not-broken control; the ×1.5 broken-burn tests deleted (`lib/game/catalog.test.ts`, `lib/game/actions.test.ts`, `lib/sim/engine.test.ts`) in favor of an engine chain-stop test (a plane that lands broken stays grounded with its queue intact despite enough fuel) and a `finishRepairs` curve test (mid-repair climb to `round(f × 100)`, floor at the start value, exact 100 + `'available'` at the gate); `repairPlane` unit pins updated — health is NOT restored by the action itself. E2E: `e2e/maintenance.spec.ts` reworked per the issue — force broken → pill shows AND dispatch 409s with the tank untouched (the ×1.5 burn flight leg is gone) → click Repair → "in repair" pill with health still ~0 → shorten the gate via the hook → back to service at exactly 100; `e2e/right-pane.spec.ts` drops the dead condition factor from its auto-refuel math ## v0.44.0 — 2026-09-21 17:10 - **World airport dataset (ISS-079).** `data/airports_eu.csv` (242 EU rows) is replaced by `data/airports_world.csv` — the full OurAirports world dump curated to one marker per metro area: keep `large_airport` rows only, then greedy 40 km haversine clustering with one airport kept per cluster ("biggest" proxy: scheduled service → IATA code → Wikipedia page → file order; no pax data exists in the CSV). Result: **1174 large airports worldwide → 1119 kept, 55 dropped into 52 multi-airport clusters** (e.g. Newark ← JFK + LaGuardia, Dallas ← DFW, Warsaw Chopin merged with Modlin). New committed generator `scripts/build-world-airports.mjs` re-derives the CSV from the source mirror (aborts on download failure; asserts EDDF/UNNT/LPPD presence, unique ICACOs, sane row count) - **Loader.** `lib/airports.ts` reads the world file and drops the continent check (type filter kept as a guard); new load-time guard throws on duplicate ICAO — three consumers key features by ICAO with silent-overwrite semantics, so duplicates would corrupt state silently - **Generated jobs are distance-capped.** New `MAX_JOB_LEG_KM = 9000` in `lib/game/seed.ts`: every spawned leg is flyable by at least one catalog plane even in worst-case headwind (b777 full-tank range ~13045 km nominal, ~10871 km with the +20% wind factor — 9000 leaves margin). `generatePackage` rejection-samples unforced origin/dest pairs until they fit (bounded at 25 attempts, then a uniform within-cap fallback); forced endpoints consume no extra draws and stay uncapped (the UNNT→LPPD 409 backstop relies on that), so under-cap RNG streams are byte-identical to before. The engine's coverage spawn caps its random dest through the shared `pickDestWithinCap` helper — one draw in the same stream position as before - **UI/spec/docs.** Map header label → "N world airports" (e2e regex updated), OpenAPI `/api/airports` summary/description reworded to the curated world dataset, PLAN.md dataset mentions updated; initial camera stays EU-tuned (out of scope) - **E2E marker isolation re-verified.** LFRS (Nantes) keeps ≥39 px clearance from its nearest markers (LFRB Brest / LFBD Bordeaux) at the initial zoom [10,52] z4 with the world dataset — no test changes needed beyond the label regex ## v0.43.0 — 2026-09-21 15:57 - **Wear rate reduced 10× per ISS-085** — `WEAR_PER_KM` in `lib/game/catalog.ts` drops from 0.005 to 0.0005, so plane maintenance status degrades 1/10th as fast (~13 mid-range hops → ~130 hops to break a full-health plane) ## v0.42.0 — 2026-09-21 15:02 - **Auto-refuel on dispatch (ISS-082).** Manual refueling is gone — planes top up automatically at dispatch, so a grounded plane can always start any leg that fits a full tank when cash allows. `flyPlane` (`lib/game/actions.ts`) tops the tank to `min(maxFuel, max(fuel, 1.5 × required))` before departure — the same target/price terms as the removed forFlight mode: liters × FUEL_PRICE_PER_L minus the rep-scaled discount (v0.38.0), deducted from cash; unaffordable → 409 `insufficient funds` with nothing mutated, and a leg that doesn't fit even a full tank still 409s `insufficient fuel` (the backstop). The engine's auto-chain (`deliverPackages`, `lib/sim/engine.ts`) does the same at landing: when the next queued leg exceeds the landed fuel it tops up to the identical target when cash allows and chains anyway — only an unaffordable top-up grounds the plane with the queue kept intact - **REMOVED** `POST /api/planes/:planeId/refuel` (route + route tests), the `refuelPlane()` action + `RefuelOptions` (`lib/game/actions.ts`), the left-pane "Refuel to full" section and per-leg Refuel button with their dead state/derived values (`components/LeftPane.tsx`), and the OpenAPI `/api/planes/{planeId}/refuel` path + `RefuelRequest` schema — endpoint count 14 → 13, stale refuel mentions reworded in info/state/queue/repair descriptions, and fly's description + 409 now document the auto-refuel - **Test hook.** New POST `/api/test/set-fuel` (`FLIGHT_TEST_HOOKS=1` only): `{ planeId, fuel }` forces a plane's fuel — finite number in [0, maxFuel] (400 outside / unknown field), 404 unknown plane; sibling of set-health - **Tests.** Unit: `refuelPlane` suite deleted; new `flyPlane` auto-refuel tests (exact top-up target + rep-discounted cost at rep 30, free no-op on a full tank, cash 0 → 409 insufficient funds with fuel unchanged, UNNT→LPPD backstop still 409s); the broken-burn test reworked to pin that ×1.5 burn raises the top-up target; `engine.test.ts`'s chain test flipped — a short tank now chains via an affordable top-up (post-top-up fuel + exact cost from the real helpers) plus a new cash-0 variant grounding with the queue intact. E2E: every `/refuel` setup/retry site (~25 across 9 specs) replaced by the set-fuel hook, heal helpers drop their refuel-and-retry guard (dispatch auto-refuels), right-pane's per-leg Refuel test and two-planes' "refuel section lists both planes" deleted in favor of one new end-to-end dispatch auto-top-up test (set-fuel drops fuel to 500 L → fly → exact cash delta from snapshot + real helpers, landed fuel = weighted burn from the topped start value), panes.spec.ts now asserts Refuel-UI absence in both panes, api.spec.ts pins thirteen endpoints ## v0.41.0 — 2026-09-21 13:26 - **Queue UI fixes (ISS-077).** Three queue-related map/UI fixes, no server logic changed: - **Would-409 controls are disabled.** The on-board "Fly to …" button (`components/LeftPane.tsx`) is additionally disabled while the plane is flying and its destination is already active or queued (title "Already in flight queue"); the Relocate section disables its GO button for the same reason AND renders the selected plane's active/queued airports as native disabled `