Known deferrals from MVP Phase 1, tracked so they are not lost: - [x] Backend: background-refresh OCM on *warm* tiles (return DB rows immediately, refresh out-of-band) to meet the <300ms cold budget — done in #33 - [x] Backend: GIN index (jsonb_path_ops) on `stations.connectors` for indexed connector filtering — done in #33 - [x] Frontend: friendlier empty/loading state with the Unsplash hero when a viewport has no stations — done in #34 - [ ] Deploy: host the backend (self-host Docker per DEPLOYMENT.md) so the Vercel frontend shows live data; set CORS + PUBLIC_API_BASE_URL — see #30 - [ ] Phase 2: one-tap reports (POST /reports gated by get_current_user) + live reliability
Known deferrals from MVP Phase 1, tracked so they are not lost:
stations.connectorsfor indexed connector filtering — done in perf(api): background OCM refresh on warm tiles + connectors GIN index #33