robert/ttrpg-initiative-tracker
1
0
Fork 0
Code Issues Pull Requests Releases 4 Wiki Activity
152 Commits 2 Branches 5 Tags
rework-backend
Commit Graph

8 Commits

Author SHA1 Message Date
david raistrick da25f46e3e fix(docker): single container (caddy+node), ESM adapters fix blank page 
Docker: moved all docker files to docker/ tree (was conflated with
upstream Dockerfile at root + server/Dockerfile). Single container now:
caddy (front, serves static + proxies /api /ws) + node backend (internal
:4001). Node never exposed. entrypoint.sh runs both. Compose: one service.

Blank page root cause: storage adapters had inconsistent module systems.
firebase.js = ESM (export). ws.js + memory.js = CJS (module.exports).
CRA prod build = ESM strict -> CJS runtime crash, blank root. Dev mode
lenient, masked bug. First ws prod build (docker) = first exposure.
Never dev/prod split intended; just inconsistency from M2 era.

Fix: all adapters ESM. ws.js lazy-loads 'ws' pkg via dynamic import()
(Node/jest only; browser uses global WebSocket). index.js static
imports. server jest: added babel.config.js (preset-env, node target)
to transform ESM for jest.

Test: src/tests/StorageEsm.test.js — 4 tests grep all adapters for
module.exports / require(). Regression guard catches CJS leak.

Verified: docker page renders (root 4534 chars, UI visible).
server 24 green, shared 90 green, FE ESM 4 green.
 2026-07-01 19:03:59 -04:00
david raistrick c1d982b4a4 fix(BUG-8): ws adapter auto-reconnect after drop 
WS adapter had no reconnect. WS dies (idle/error/close) → wsReady=null,
subscribers dead forever, display frozen until full reload.

Changes (src/storage/ws.js):
- onClose: schedule reconnect via setTimeout(500ms), ensureWs re-arms.
  Guard: disposed flag stops reconnect after dispose.
- onOpen: resubscribe all existing doc/coll subscribers (backend state
  may have changed). Re-fetch current values on RECONNECT only (skip
  first connect — initial REST fetch in subscribe* already did). Added
  everConnected flag to distinguish first vs reconnect.
- reconnectTimer unref'd (Node) to avoid hanging event loop.
- dispose(cb): set disposed, clear timer, close ws, then cb.

Also fixed test teardown leaks:
- server/index.js close(): terminate all wss.clients before wss.close().
  Reconnect test spawned new ws to server; old close hung on live conn.
- both ws test factories: port 0 (OS picks free) instead of module-local
  nextPort counter. Parallel jest workers collided on EADDRINUSE.

Tests: ws-reconnect GREEN (1.7s), ws-contract 23 GREEN. No regression.
server suite 24/24. shared 90/90.
 2026-07-01 18:26:42 -04:00
david raistrick 7467a8d30f feat(M5): docker-compose full stack (caddy + node backend + sqlite) 
docker-compose.yml: two profiles.
  - backend: backend (node+ws+better-sqlite3, /data volume) + frontend
    (Caddy static build, STORAGE=ws, same-origin proxy)
  - firebase: existing Dockerfile + nginx (upstream path, untouched)
  Run: docker compose --profile backend up --build. OrbStack local now,
  remote docker context later.

server/Dockerfile: node:18-alpine, workspaces (shared dep), rebuild
better-sqlite3 for musl, DB at /data/tracker.sqlite.

Dockerfile.ws: CRA build STORAGE=ws → caddy:2-alpine serves /srv.
No backend URL baked (same-origin).

Caddyfile: handle /api/* + handle /ws → backend:4001 (path preserved,
mutually-exclusive handles so try_files SPA fallback never shadows proxy).
handle { static try_files } last. HTTP basic auth block optional.

src/storage/ws.js: same-origin defaults. Empty baseUrl = relative fetch
(Caddy proxy). wsUrl derives from window.location (http→ws/https→wss).
Fallback localhost for bare npm start dev.

.dockerignore: add data/ scratch/ tmp/ (never bake into image). Keep
Caddyfile in context (frontend build COPYs it).

Smoke verified via OrbStack:
  - GET / → 200 (static SPA)
  - PUT/GET /api/doc roundtrip → JSON persists
  - WS /ws subscribe + change push → both work through proxy

Firebase profile: pre-existing Dockerfile requires .env.local (hardcoded
COPY on main, not changed here). User must create file. Not a regression.
 2026-07-01 14:39:47 -04:00
david raistrick e514a48d6e tests: BUG-8 ws reconnect RED, BUG-7 reorder no-undo doc, ws _test accessor 
server/tests/ws-reconnect.test.js: subscribe, write (fires), force-drop WS,
write again (must still fire). RED on current. wsReady=null after drop,
no reconnect, subscribers dead forever. Display frozen.

src/storage/ws.js: added _test accessor (getWs, forceDrop, getReady,
docSubs, collSubs) for reconnect test. Test-only, no behavior change.

TODO: BUG-7 (reorder no undo), BUG-8 (ws reconnect) added.
 2026-06-30 13:59:58 -04:00
david raistrick f81308a0df tests: consolidate into tests/ dirs, fix import paths 
Move all test files out of source dirs into per-workspace tests/:
- shared/tests/   (3 unit test files)
- server/tests/   (1 integration test)
- src/tests/      (8 characterization + scenario tests + testHelpers)

Fix all relative import paths (App, storage, __mocks__, testHelpers).
Fix jest.config testMatch globs in shared/ and server/ (rootDir +
<rootDir>/tests pattern).

Delete scripts/repro-pause-bug.js (debug scratch, superseded by
turn.pause-add.test.js).

Keep scripts/replay-combat.js + scripts/audit-rotation.js as manual
demo/exploratory tools (NOT unit tests, not deterministic).

No logic changes. All green: shared 49 + 1 validated RED, server 23,
FE 62. Scenario test unchanged (240s timeout, pre-existing slow).
 2026-06-29 16:02:22 -04:00
david raistrick 52866784b2 M2: replace shape-specific backend with generic KV doc store (firebase mirror) 
Backend was endpoint-mapper (POST /api/campaigns ignores path id, etc). Adapter
translation layer brittle, untested, lost doc identity. Generic contract (Layer 2)
test caught 15 bugs immediately.

Rewrite to firebase-mirror KV model:
- server/db.js: docs table (path PK, parent, data JSON). getDoc/setDoc/updateDoc/
  deleteDoc/getCollection/batchWrite. parent = path prefix for collection queries.
- server/index.js: generic REST (GET/PUT/PATCH/DELETE /api/doc, GET /api/collection,
  POST /api/collection addDoc, POST /api/batch). WS subscribe by path (doc|collection),
  broadcast to doc subs at changed path + collection subs at parent path.
- src/storage/ws.js: thin passthrough adapter. norm() strips firebase prefix.
  initial value via REST (independent of WS connect), subsequent changes via WS.
- shared/turn.js kept (M4 use). server/handlers.js + server.test.js removed (logic
  now in App, backend is dumb KV).
- src/storage/contract.js: flush() bumped to 50ms (WS roundtrip > setTimeout(0)).

Layer 2 test (server/ws-contract.test.js): spins fresh backend per test, runs same
storage contract spec against createWsStorage. Catches adapter translation bugs
that firebase-mock Layer 1 tests cannot.

nanoid v5 ESM breaks jest CJS -> replaced with crypto.randomUUID (Node builtin).

Tests: 114 green (39 shared + 19 ws-contract + 56 frontend).
 2026-06-29 13:00:24 -04:00
david raistrick 0e76fb2fc7 M1: backend (Express+ws+better-sqlite3) + integration tests 
- server/db.js: SQLite schema mirroring Firestore doc tree
- server/handlers.js: action -> shared turn fn -> tx persist -> broadcast
- server/index.js: REST endpoints + WebSocket real-time push
- server/server.test.js: 7 integration tests (REST CRUD + combat flow)
- --forceExit for jest (open WS handles)

Backend boots, serves state, persists to SQLite.
 2026-06-28 17:01:53 -04:00
david raistrick e06adaa081 M1: shared turn logic + characterization tests (39 green) 
- npm workspaces: shared/, server/
- shared/turn.js: port turn logic verbatim from App.js (bugs preserved)
- 39 characterization tests lock current behavior
- gitignore: sqlite data, logs
 2026-06-28 16:57:43 -04:00