kamilkrzywda
a801311fc5
CI / build + unit + e2e (pull_request) Failing after 5m28s
Part A: r(F) = k(We + rho*F) L/km, k calibrated so a full tank burns exactly
the old constant; closed forms in lib/sim/movement.ts (fuelAtWeighted,
requiredFuelFor, burnRateConstant, FUEL_DENSITY_KG_PER_L). Catalog gains
emptyWeightKg (ATR72 OEW 9300 kg), exposed in the snapshot. All three burn
call sites (engine landing, state live fuel, fly validation) moved to the
weighted model; old constant-rate fuelAt removed.
Part B: POST /api/planes/{planeId}/refuel takes an optional body
{mode:'full'|'forFlight', packageId} — no body keeps today's fill-to-full
behavior exactly (fuel lands bit-exact on maxFuel); forFlight adds just
enough for the loaded package's destination leg (409 not-loaded/expired/
doesn't-fit, 400 bad body). UI: per-cargo-row 'Refuel' button
(refuel-for-{pkg.id}), market button renamed 'Refuel to full'.
OpenAPI: RefuelRequest schema + optional requestBody + 400 on refuel, fly's
409 description updated, CatalogEntry.emptyWeightKg. Tests: new weighted-burn
suite, forFlight action/route cases (incl. ~7670 km UNNT<->LPPD doesn't-fit),
openapi guards; landing-fuel assertions in engine.test.ts + e2e/actions.spec.ts
use the closed form; the e2e 'return leg unfuelable' scenario reworked — a full
tank now covers the EDDF<->LEZL round trip, so the 409 moves to the second
outbound. New e2e: row Refuel deducts exactly L*price / raises fuel by exactly
L via the endpoint's own response snapshot.