Deepen comuline analysis: repo timeline, dormancy evidence, upstream drift

This commit is contained in:
hermes 2026-10-12 12:59:31 +00:00
commit 1032de68be
4 changed files with 20 additions and 6 deletions

View file

@ -23,7 +23,7 @@ updated: 2026-10-12
- **Architecture:** daily cron → Cloudflare Workers (Hono + Bun) → Neon Postgres (Drizzle ORM) + Upstash Redis read cache → public REST API with OpenAPI spec → Vite/React web app (plus native apps) consuming it via a typed `openapi-fetch` client.
- **Coverage (from live API):** 113 stations — 95 daop-1 (Jabodetabek), 17 daop-6 (Yogyakarta area), 1 daop-2 (Purwakarta); 111 typed KRL + 2 LOCAL (Cikampek, Purwakarta). Sample row counts: MRI = 1,045 departures; DP = 730; CLT = 683; JAKK = 367; TNG = 84; BEK = 0.
- **No live tracking:** schedules and departures only — no GPS/position data anywhere (bundle + API verified). "Real-time" in their copy means "current schedule lookup", not live trains.
- **Data frozen:** schedule rows show last updates 2024-11-13 (bulk) + 2025-03-30 (partial passes); nothing newer. Repo activity stops 2025-03-30 (api) / 2025-04-03 (web).
- **Data frozen — an ops stop, not a breakage:** last successful sync ~2025-03-30 (rows show 2024-11-13 bulk + 2025-03-30 passes — both matching the last merge day). No automatic sync exists in the repo at all (the README's "daily cron" lived outside the codebase; no cron triggers/workflows/scheduled handler anywhere). The official KRL side stayed alive and drifted (endpoint renames, token rotation) — adaptation fixes sit **unmerged** (PR #16, Jan 2026). Details: [[comuline-krl-data/analysis|Analysis]].
## Files