Three-stage refactor of the game-engine plumbing (game logic untouched):
Stage 1 — lock-free persistence + admin serialisation. Remove the file
lock from repo/fs (the .lock file, the Read/Write-vs-*Safe duality and the
dead ReadSafe polling) and replace the two-step rename with a single atomic
rename so concurrent reads are torn-free without a lock. Serialise the
state-mutating admin writers (init/turn/banish) with one shared router
LimitMiddleware, rewritten to block on the request context instead of a
racy shared 100ms timer.
Stage 2 — remove the obsolete immediate-command path end to end. Players
submit through PUT /api/v1/order; the legacy PUT /api/v1/command path is
deleted across game (route, handler, 24 command factories, Ctrl), backend
(Commands handler/route, engineclient.ExecuteCommands), gateway (dispatch +
executeUserGamesCommand + routing entry), the FlatBuffers/model contract
(UserGamesCommand[Response]) and transcoder, plus every affected
OpenAPI/README/FUNCTIONAL/ARCHITECTURE doc. The integration proxy test is
converted to the order path.
Stage 3 — flatten the REST->engine wrapper. Replace the executor adapter,
the controller package functions and RepoController with one concrete
controller.Service; drop the single-implementation Repo and Storage
interfaces (repo.Repo / fs.FS are now concrete). Handlers depend on a thin
handler.Engine seam and own the domain->REST projection; storage is
resolved once at startup instead of per request.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ARCHITECTURE.md §15 "Verification order" specifies signature verification
(step 4) before payload_hash (step 5), but the authenticated-edge
decorator chain wrapped the payload-hash gate outside the signature gate,
so the hash was checked first. gateway/README.md and gateway/docs/flows.md
had drifted to match the code (hash-first), leaving ARCHITECTURE.md as the
lone source describing the intended order.
Swap the two decorators in server.go so the signature gate runs first, and
align README + flows.md to ARCHITECTURE.md. Signature-first is the
cryptographically sound order: the signature covers the payload_hash field,
so the request is authenticated before any of its content is processed.
Observable side effect: a request carrying a tampered payload_hash whose
signature was computed over the original hash is now rejected at the
signature gate (UNAUTHENTICATED "invalid request signature") instead of the
hash gate (INVALID_ARGUMENT). Security is unchanged — both refusals happen
before the payload is handled. The four payload-hash unit tests re-sign
over the tampered hash so they keep exercising the hash gate; the
cross-service integration test signs over the overridden hash and already
accepts both codes.
Refs #39
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Serve the whole stack behind one host: site at /, game UI at /game/,
gateway REST at /api + /healthz, Connect at /rpc (prefix stripped by the
edge Caddy). The built artifact is domain-agnostic — the UI talks to the
gateway same-origin via relative URLs, so the same bundle runs under any
host with no rebuild and with CORS disabled.
- Rename the Connect proto service galaxy.gateway.v1.EdgeGateway ->
edge.v1.Gateway; regenerate Go + TS; public path /rpc/edge.v1.Gateway.
- Move the game UI under base path /game (env BASE_PATH); make the
manifest, service-worker scope, WASM loader, and all navigation
base-aware via a withBase helper.
- Relative API + /rpc Connect prefix; Vite dev proxy mirrors the strip.
- Rewrite the edge Caddy (dev + prod) for path-based routing; empty CORS
allow-lists (same-origin); single host.
- New VitePress project site (site/): i18n en/ru with switcher, LaTeX
math, minimal monospace theme; built and served at /.
- dev-deploy compose/Makefile + CI (dev-deploy, prod-build, new
site-build) build and seed the site; probes hit /, /game/, /healthz.
- Sync docs (ARCHITECTURE, gateway README/openapi, dev-deploy &
local-dev READMEs, CLAUDE.md, ui/PLAN).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Browser fetch-streaming layers close response bodies they consider
idle after roughly 15-30 s without incoming bytes. Safari is the
most aggressive, but the symptom matters everywhere: a quiet
SubscribeEvents stream (lobby, between turns, mailbox empty) gets
torn down by the browser, the EventStream singleton reconnects with
backoff, and any push event that fires inside the reconnect window
is lost because `push.Hub` queues are not persisted across
subscription closes. The user-visible failure mode is the
intermittent "Fetch API cannot load … due to access control checks"
console error (a misleading WebKit symptom — CORS headers are
actually present) plus missed turn-ready / mail-received toasts.
Server-side fix: a silence-based heartbeat at the
`authenticatedPushStreamService` wrapper layer. After the signed
`gateway.server_time` bootstrap event, gateway wraps the bound
stream with `heartbeatingStream`. Every tail Send (fan-out, future
variants) resets the silence timer; when the timer elapses, a
goroutine emits `gateway.heartbeat` with only `EventType` set —
everything else stays at proto3 defaults, so the wire frame is
~45 bytes amortised. A `sendMu` serialises the heartbeat goroutine
with tail Sends because grpc.ServerStream.Send is not goroutine-safe.
The heartbeat is intentionally UNSIGNED: heartbeats carry no
payload, dispatch to no handler on the client, and an injected
heartbeat trivially causes no user-visible state change. TLS still
protects the wire and real events keep the signed envelope
unchanged. Documented in `docs/ARCHITECTURE.md` § 15 alongside the
per-scale bandwidth projection (100…100 000 clients × 15…60 s).
Config: new `GATEWAY_PUSH_HEARTBEAT_INTERVAL` (default `15s`,
`0s` disables). Telemetry: new
`gateway.push.heartbeats_sent{outcome}` counter so operators can
budget bandwidth and spot a sudden `outcome=error` bump as an
upstream-failing-before-flush signal.
Client (`ui/frontend/src/api/events.svelte.ts`): early `continue`
on `event.eventType === "gateway.heartbeat"` before `verifyEvent`,
`verifyPayloadHash`, or dispatch — empty signature would otherwise
trip SignatureError and reconnect. A leading heartbeat still flips
`connectionStatus` to `connected` and resets backoff, because
receiving one is proof the stream is healthy.
Tests:
- `push_heartbeat_test.go`: unit tests for the wrapper — zero
interval returns nil, heartbeat fires after silence, real Send
resets the timer, Stop / context-cancel halt the goroutine,
Send errors propagate.
- `server_test.go`: integration tests through the full gateway
pipeline — heartbeat fires after the configured silence window,
zero interval keeps the stream silent.
- `config_test.go`: default applied, env-override parsed,
negative value rejected.
- `events.test.ts`: heartbeat skipped before verification + not
dispatched to handlers; leading heartbeat still flips
`connectionStatus` to `connected`.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Backend now owns the turn-cutoff and pause guards the order tab
relies on: the scheduler flips runtime_status between
generation_in_progress and running around every engine tick, a
failed tick auto-pauses the game through OnRuntimeSnapshot, and a
new game.paused notification kind fans out alongside
game.turn.ready. The user-games handlers reject submits with
HTTP 409 turn_already_closed or game_paused depending on the
runtime state.
UI delegates auto-sync to a new OrderQueue: offline detection,
single retry on reconnect, conflict / paused classification.
OrderDraftStore surfaces conflictBanner / pausedBanner runes,
clears them on local mutation or on a game.turn.ready push via
resetForNewTurn. The order tab renders the matching banners and
the new conflict per-row badge; i18n bundles cover en + ru.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Extend pkg/model/lobby and pkg/schema/fbs/lobby.fbs with public-games
list, my-applications/invites lists, game-create, application-submit,
invite-redeem/decline. Mirror the matching transcoder pairs and Go
fixture round-trip tests.
- Wire the seven new lobby message types through
gateway/internal/backendclient/{routes,lobby_commands}.go with
per-command REST helpers, JSON-tolerant decoding of backend wire
shapes, and httptest-based unit coverage for success / 4xx / 5xx /
503 across each command.
- Introduce TS-side FlatBuffers via the `flatbuffers` runtime dep, a
`make fbs-ts` target driving flatc, and the generated bindings under
ui/frontend/src/proto/galaxy/fbs. Phase 7's `user.account.get` decode
now uses these bindings as well, closing the JSON.parse vs
FlatBuffers gap that would have failed against a real local stack.
- Replace the placeholder lobby with five sections (my games, pending
invitations, my applications, public games, create new game) and the
/lobby/create form. Submit-application uses an inline race-name
form on the public-game card; create-game keeps name / description /
turn_schedule / enrollment_ends_at always visible and the rest under
an Advanced toggle with TS-side defaults.
- Update lobby/+page.svelte to throw LobbyError on non-ok result codes;
GalaxyClient.executeCommand now returns { resultCode, payloadBytes }.
- Vitest binding round-trips, lobby.ts wrapper unit tests, lobby-page
+ lobby-create component tests, Playwright lobby-flow.spec covering
create / submit / accept across all four projects. Phase 7 e2e was
migrated to the FlatBuffers fixtures and to click+fill against the
Safari-autofill readonly inputs.
- Mark Phase 8 done in ui/PLAN.md, mirror the wire-format note into
Phase 7, append the new lobby commands to gateway/README.md and
docs/ARCHITECTURE.md, add ui/docs/lobby.md.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replace the native-gRPC server bootstrap with a single
`connectrpc.com/connect` HTTP/h2c listener. Connect-Go natively
serves Connect, gRPC, and gRPC-Web on the same port, so browsers can
now reach the authenticated surface without giving up the gRPC
framing native and desktop clients may use later. The decorator
stack (envelope → session → payload-hash → signature →
freshness/replay → rate-limit → routing/push) is reused unchanged
behind a small Connect → gRPC adapter and a `grpc.ServerStream`
shim around `*connect.ServerStream`.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ilia Denisov