4 Commits

Author SHA1 Message Date
julian 54e691a4c9 fix(release): latest.json entry per installer type — .deb booths could never self-update
Release desktop / bundle (push) Successful in 5m26s
tauri-plugin-updater resolves the download target as {os}-{arch}-{installer}
first (linux-x86_64-deb — the bundler stamps the installer type into the
binary, verified with `strings` on a local .deb) and only then bare
linux-x86_64. Our manifest carried only the bare key, pointing at the
AppImage. A .deb install therefore downloaded the AppImage, verified its
signature, then failed install_deb()'s is_deb check with
InvalidUpdaterFormat — after the download, before any relaunch. This, not
version drift or swallowed errors, is why v0.1.0→v0.1.6 never self-updated.

latest.json now carries linux-x86_64-deb, linux-x86_64-rpm (when built) and
linux-x86_64 (AppImage), each with its own .sig. A .deb update ends in a
polkit password prompt (pkexec dpkg -i) — the intended admin gate on a
root-installed package. README + wiki updated; wiki also records the v0.1.6
LIVE field verification.

Claude-Session: https://claude.ai/code/session_01FWncR69HgGPuei1dLrW3cU
2026-09-04 15:28:08 +02:00
julian 52862db8ad chore(resources): bump stage TAG to 8fa66c9
Build & push images / images (push) Successful in 2m48s
Claude-Session: https://claude.ai/code/session_01FWncR69HgGPuei1dLrW3cU
2026-09-04 12:16:16 +02:00
julian 8fa66c9911 fix(desktop): WS ticket auth for the live feed; desktop logs never reached the server
Build & push images / images (push) Successful in 2m51s
Release desktop / bundle (push) Successful in 41m19s
The v0.1.4 Origin fix cleared only the first of two gates in /api/ws's
preHandler. The second, req.jwtVerify(), reads the HttpOnly cookie — which
tauri-plugin-websocket (a bare tungstenite client, no cookie jar) can never
send. Every desktop handshake 401'd and use-live-feed reconnected every 10s
(confirmed in the park-2 server log).

- routes/ws.ts: POST /api/ws/ticket (cookie + CSRF auth) mints a 30s,
  single-use, in-memory ticket; the WS preHandler accepts it via an
  x-ws-ticket header after the Origin check, then the same report:read
  role check. Browser cookie path unchanged; JWT stays out of JS.
- platform-ws.ts: fetch a ticket before connect, send it with the Origin
  header; connect failures now go through logClient (rate-limited).
- logger.ts: flush read the CSRF token from document.cookie, null on
  desktop, so every desktop POST /api/logs 403'd and was dropped silently —
  no desktop client log had ever reached app_logs. Stash moved to a
  dependency-free lib/desktop-csrf.ts shared by api.ts and logger.ts.
- backend-config.ts: ConnectScreen probe uses the unauthenticated /health
  (now also returns app: "parking-system") instead of accepting any 401.
- README: local-AppImage release gate — tauri dev runs at
  http://localhost:5173, not tauri://localhost, so none of these
  origin-dependent bugs reproduce there.
- wiki: new section + log entry; four citation corrections.

Requires the server image with this commit deployed before the new desktop
build connects (the ticket endpoint must exist).

Claude-Session: https://claude.ai/code/session_01FWncR69HgGPuei1dLrW3cU
2026-09-04 12:09:30 +02:00
julian 70e1e9939f chore(resources): bump stage TAG to 5c6a21e
Build & push images / images (push) Successful in 2m50s
Promotes park-buzi + park-2 to the runtime-configurable desktop backend
address (ConnectScreen) and the desktop CSRF fix. build-images.yml
confirmed green for this sha before bumping.
2026-09-04 11:27:17 +02:00
13 changed files with 388 additions and 75 deletions
+51 -17
View File
@@ -129,8 +129,23 @@ jobs:
# The Tauri updater fetches a manifest describing the newest version, its
# notes, and per-target {signature, url}. The URL points at the MIRROR
# repo (mca/public_releases) — that's the unauthenticated endpoint field
# appliances actually reach; see the workflow header for why. Adjust the
# platform keys you actually ship.
# appliances actually reach; see the workflow header for why.
#
# ONE ENTRY PER INSTALLER TYPE — this is what made every in-app update
# v0.1.0→v0.1.6 fail. tauri-plugin-updater looks up
# `{os}-{arch}-{installer}` FIRST (linux-x86_64-deb / -rpm / -appimage,
# from the running app's detected bundle type) and only then the bare
# `linux-x86_64`. The booths run the .deb, and the manifest used to
# carry ONLY `linux-x86_64` → the AppImage. So a .deb install found the
# "update", downloaded the AppImage, verified its signature fine, then
# handed the bytes to install_deb(), which checks they're a .deb
# (infer::archive::is_deb) and bails with InvalidUpdaterFormat — after
# the download, before any relaunch, with the error swallowed client-
# side until v0.1.6. Now each installer gets its own signed asset; the
# bare key stays for an AppImage install. .deb/.rpm updates run
# `pkexec dpkg -i` / `rpm -U`, so the operator sees a polkit password
# prompt — intended: updating a root-installed package IS an admin
# action on this box (see wiki/decisions/desktop-shell-tauri.md).
env:
SERVER_URL: ${{ github.server_url }}
MIRROR_REPO: mca/public_releases
@@ -138,22 +153,41 @@ jobs:
run: |
set -e
VERSION="${TAG#v}"
APPIMAGE=$(cd dist && ls *.AppImage | head -1)
SIG=$(cat "dist/${APPIMAGE}.sig")
ASSET_URL="${SERVER_URL}/${MIRROR_REPO}/releases/download/desktop-latest/${APPIMAGE}"
cat > dist/latest.json <<JSON
{
"version": "${VERSION}",
"notes": "Parking System ${TAG}",
"pub_date": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
"platforms": {
"linux-x86_64": {
"signature": "${SIG}",
"url": "${ASSET_URL}"
}
}
ASSET_BASE="${SERVER_URL}/${MIRROR_REPO}/releases/download/desktop-latest"
cat > /tmp/latest.js <<'JS'
const fs = require("fs");
const [version, tag, base] = process.argv.slice(2);
const files = fs.readdirSync("dist");
const pick = (ext) => files.find((f) => f.endsWith(ext));
const entry = (f) => ({
signature: fs.readFileSync(`dist/${f}.sig`, "utf8").trim(),
url: `${base}/${f}`,
});
const deb = pick(".deb"), rpm = pick(".rpm"), appimage = pick(".AppImage");
if (!deb || !appimage) {
console.error(`missing bundle in dist/: deb=${deb} appimage=${appimage}`);
process.exit(1);
}
JSON
const platforms = {
"linux-x86_64-deb": entry(deb),
...(rpm ? { "linux-x86_64-rpm": entry(rpm) } : {}),
"linux-x86_64": entry(appimage),
};
fs.writeFileSync(
"dist/latest.json",
JSON.stringify(
{
version,
notes: `Parking System ${tag}`,
pub_date: new Date().toISOString().replace(/\.\d+Z$/, "Z"),
platforms,
},
null,
2,
) + "\n",
);
JS
node /tmp/latest.js "${VERSION}" "${TAG}" "${ASSET_BASE}"
echo "latest.json:"; cat dist/latest.json
- name: Create release + upload assets (Gitea API)
+29
View File
@@ -44,6 +44,35 @@ see that workflow's header and `wiki/decisions/desktop-shell-tauri.md`). The upd
signing pubkey live in `tauri.conf.json`; the private signing key is held outside the repo, never
committed.
**The manifest carries one entry per installer type** (`linux-x86_64-deb`, `linux-x86_64-rpm`,
and bare `linux-x86_64` for AppImage). The updater picks the entry matching how the running app
was installed — a `.deb` install will only ever accept a signed `.deb`. Booths run the `.deb`,
so an in-app update ends in a **polkit password prompt** (`pkexec dpkg -i`): that is expected,
and it is the right gate — the package lives in `/usr/bin`, root-owned, and the operator is not
supposed to be able to replace it silently. Cancel the prompt and the app keeps running the old
version; the failure is logged to the server's Logs viewer.
## Release gate — run the REAL bundle locally before tagging
`tauri dev` loads the SPA from `http://localhost:5173`, a plain http origin. The shipped bundle
loads it from `tauri://localhost`, a *secure* custom-scheme origin — and every desktop-only bug
found in the field on 2026-09-03/04 (relative-URL DOMException, mixed content, missing WS
`Origin`, the reqwest-vs-webview cookie split, the WS handshake that can't carry the cookie)
depends on that difference. **Dev mode cannot reproduce any of them**, so "works in `tauri dev`"
carries no information about a release. Before pushing a `vX.Y.Z` tag:
1. `pnpm --filter @parking/server dev` (local backend; `.env` must have `COOKIE_SECURE=0` and
`tauri://localhost` in `WS_ALLOWED_ORIGINS`).
2. `pnpm --filter @parking/desktop bundle` and run the produced AppImage from
`src-tauri/target/release/bundle/appimage/` (WSLg is enough).
3. On the ConnectScreen enter `127.0.0.1:3000`, **Test** must say reachable, then **Save**.
4. Log in. The booth header must show **LIVE** (not "JASHTË LINJË") within a few seconds.
5. Perform one mutation (e.g. change your UI language) — it must succeed (proves CSRF).
6. Open Setup → Logs and confirm a `frontend`-sourced row from this desktop session exists
(proves the desktop log channel; historically it was silently 403'd).
Only then tag. If a release still fails in the field, the gap is in this list — fix the list.
## Not here (deliberately)
Kiosk lockdown (fullscreen/no-decorations) and launching Fastify from the shell are out of scope for
+1 -1
View File
@@ -30,7 +30,7 @@ describe("health + login", () => {
it("GET /health is open", async () => {
const res = await app.inject({ method: "GET", url: "/health" });
expect(res.statusCode).toBe(200);
expect(res.json()).toEqual({ status: "ok" });
expect(res.json()).toEqual({ status: "ok", app: "parking-system" });
});
it("login with bad credentials is rejected", async () => {
+77 -6
View File
@@ -1,7 +1,8 @@
import { randomBytes } from "node:crypto";
import type { FastifyInstance } from "fastify";
import type { Db } from "@parking/db";
import type { LedgerEvent } from "@parking/shared";
import { roleHasPermissions } from "../auth.js";
import { requireAuth, roleHasPermissions } from "../auth.js";
import {
deviceEvents,
type LaneStatusEvent,
@@ -31,11 +32,57 @@ import { getOccupancy } from "../occupancy.js";
// an Origin allowlist: the handshake's Origin must be same-origin (or an explicitly
// allowed booth UI origin). Non-browser clients (no Origin) are rejected too.
// See auth.ts, event-log.ts (emitLedger), capacity-occupancy.md.
//
// Desktop shell (Tauri) exception — the WS TICKET. The desktop app's HTTP goes
// through tauri-plugin-http (reqwest, its own cookie jar) and its WebSocket
// through tauri-plugin-websocket (bare tungstenite, NO cookie jar at all), so
// the JWT cookie set at login can never ride on the WS handshake — jwtVerify()
// would 401 every connect (found 2026-09-04: the desktop live feed reconnected
// every 10s forever). The JWT is HttpOnly and must stay out of JS, so instead
// the desktop client POSTs /api/ws/ticket (normal cookie + CSRF auth) to get a
// single-use, 30-second random ticket bound to its user, and presents it in an
// `x-ws-ticket` header on the handshake. A browser page cannot set custom
// headers on a WebSocket, so this path is unreachable from a browser and adds
// no CSWSH surface; the Origin allowlist still applies to both paths.
/** Permission required to watch the live feed (a read-only stream of ledger +
* device status). Any role granted `report:read` may watch. */
const WATCH_PERMISSION = "report:read" as const;
/** Handshake header carrying a desktop WS ticket (see file header). */
const WS_TICKET_HEADER = "x-ws-ticket";
/** A ticket is only good for the connect that immediately follows its issue. */
const WS_TICKET_TTL_MS = 30_000;
interface WsTicket {
sub: string;
roleId: string;
expiresAt: number;
}
/** Outstanding tickets. Tiny (one per desktop connect attempt), in-memory only —
* a server restart invalidates them, which is fine: the client just asks for
* another on its next reconnect. */
const tickets = new Map<string, WsTicket>();
function issueWsTicket(sub: string, roleId: string): string {
const now = Date.now();
for (const [key, t] of tickets) {
if (t.expiresAt <= now) tickets.delete(key);
}
const ticket = randomBytes(32).toString("hex");
tickets.set(ticket, { sub, roleId, expiresAt: now + WS_TICKET_TTL_MS });
return ticket;
}
/** Single-use: the ticket is removed whether or not it turns out to be valid. */
function consumeWsTicket(ticket: string): WsTicket | null {
const t = tickets.get(ticket);
if (!t) return null;
tickets.delete(ticket);
return t.expiresAt > Date.now() ? t : null;
}
/**
* Is the handshake's Origin trusted? Same-origin (Origin host === Host header) is
* always allowed; additional origins can be allowlisted via WS_ALLOWED_ORIGINS
@@ -80,19 +127,43 @@ export async function wsRoutes(
laneStatus: LaneStatus,
lanePresence: LanePresence,
): Promise<void> {
// Desktop-only: mint a WS ticket for the signed-in session (see file header).
// Ordinary cookie + CSRF auth — the desktop client CAN do that over HTTP (via
// tauri-plugin-http), it just can't carry the cookie onto the WebSocket.
app.post("/api/ws/ticket", { preHandler: requireAuth }, async (req) => ({
ticket: issueWsTicket(req.user.sub, req.user.roleId),
expiresInMs: WS_TICKET_TTL_MS,
}));
app.get(
"/api/ws",
{
websocket: true,
// Origin allowlist (anti-CSWSH, replaces CSRF — see file header) THEN JWT +
// role. Reject a cross/absent origin before touching the token, so a hijack
// attempt never reaches an authenticated socket. jwtVerify reads the cookie.
// Origin allowlist (anti-CSWSH, replaces CSRF — see file header) THEN
// session (JWT cookie, or a desktop WS ticket) THEN role. Reject a
// cross/absent origin before touching either credential, so a hijack
// attempt never reaches an authenticated socket.
preHandler: async (req) => {
if (!isAllowedOrigin(req.headers.origin, req.headers.host)) {
throw Object.assign(new Error("forbidden origin"), { statusCode: 403 });
}
await req.jwtVerify();
if (!req.user || !roleHasPermissions(req.user.roleId, [WATCH_PERMISSION])) {
const rawTicket = req.headers[WS_TICKET_HEADER];
const ticket = Array.isArray(rawTicket) ? rawTicket[0] : rawTicket;
let roleId: string;
if (ticket !== undefined) {
const t = consumeWsTicket(ticket);
if (!t) {
throw Object.assign(new Error("invalid or expired ws ticket"), { statusCode: 401 });
}
roleId = t.roleId;
} else {
await req.jwtVerify(); // reads the HttpOnly cookie (browser path)
if (!req.user) {
throw Object.assign(new Error("forbidden"), { statusCode: 403 });
}
roleId = req.user.roleId;
}
if (!roleHasPermissions(roleId, [WATCH_PERMISSION])) {
throw Object.assign(new Error("forbidden"), { statusCode: 403 });
}
},
+4 -1
View File
@@ -106,7 +106,10 @@ export async function buildServer(opts: BuildOptions = {}): Promise<FastifyInsta
cookie: { cookieName: TOKEN_COOKIE, signed: false },
});
app.get("/health", async () => ({ status: "ok" }));
// Unauthenticated liveness probe. `app` lets a client (the desktop ConnectScreen
// test — apps/web/src/lib/backend-config.ts) tell THIS server apart from any
// other service that happens to answer on the address the operator typed.
app.get("/health", async () => ({ status: "ok", app: "parking-system" }));
// Local username/password login → JWT in an HttpOnly cookie + CSRF cookie.
await authRoutes(app, db);
+18 -18
View File
@@ -5,18 +5,14 @@
// back in the X-CSRF-Token header (double-submit). See
// wiki/entities/local-jwt-auth.md.
//
// Desktop shell exception: tauri-plugin-http's fetch() runs through Rust's
// reqwest, which keeps its OWN cookie jar separate from the webview —
// document.cookie on the tauri://localhost page never sees a cookie set on a
// plugin-routed response (open upstream bug, tauri-apps/tauri#13045). The
// cookie itself IS still sent back to the server by reqwest on later
// requests (only the client-side *read* is broken), so the server also
// Desktop shell exception: document.cookie can't see the CSRF cookie there
// (reqwest's separate jar — see lib/desktop-csrf.ts), so the server also
// echoes the token in the login/me response BODY (sessionView's csrfToken —
// see routes/auth.ts) purely as a second channel for the desktop client to
// learn the value; desktopCsrfToken below stashes it in memory and
// setSessionUser() (called wherever a SessionUser is received) keeps it
// current. The browser path is untouched — it still reads document.cookie.
// see routes/auth.ts) and setSessionUser() (called wherever a SessionUser is
// received) stashes it via setDesktopCsrfToken(). The browser path is
// untouched — it still reads document.cookie.
import { getDesktopCsrfToken, setDesktopCsrfToken } from "./lib/desktop-csrf.js";
import { logFailedRequest } from "./lib/logger.js";
import { apiUrl, platformFetch } from "./lib/origin.js";
import { inTauri } from "./lib/tauri-env.js";
@@ -30,16 +26,11 @@ function readCookie(name: string): string | null {
return m ? decodeURIComponent(m[1]!) : null;
}
/** Desktop-only in-memory CSRF stash — see file header. Never persisted (a
* fresh app launch always logs in again, or bootstraps via /api/auth/me
* which re-supplies it). */
let desktopCsrfToken: string | null = null;
/** Update the desktop CSRF stash. Called wherever a SessionUser is received
* (login, fetchMe). No-op / cheap in the browser (the value just goes
* unused there — reads still come from document.cookie). */
function setSessionUser(user: SessionUser): void {
if (user.csrfToken) desktopCsrfToken = user.csrfToken;
if (user.csrfToken) setDesktopCsrfToken(user.csrfToken);
}
/** fetch wrapper: sends cookies, adds CSRF header on mutations, parses errors. */
@@ -50,7 +41,7 @@ export async function apiFetch<T>(path: string, init: RequestInit = {}): Promise
headers.set("content-type", "application/json");
}
if (method !== "GET" && method !== "HEAD") {
const csrf = inTauri() ? desktopCsrfToken : readCookie(CSRF_COOKIE);
const csrf = inTauri() ? getDesktopCsrfToken() : readCookie(CSRF_COOKIE);
if (csrf) headers.set(CSRF_HEADER, csrf);
}
const res = await platformFetch(apiUrl(path), { ...init, headers, credentials: "include" });
@@ -129,10 +120,19 @@ export async function login(username: string, password: string): Promise<Session
export async function logout(): Promise<{ ok: boolean }> {
const res = await apiFetch<{ ok: boolean }>("/api/auth/logout", { method: "POST" });
desktopCsrfToken = null;
setDesktopCsrfToken(null);
return res;
}
/** Desktop only: mint a single-use, short-lived ticket that authenticates the
* live-feed WebSocket handshake in place of the session cookie — the native WS
* plugin has no cookie jar, so the cookie can never ride along (see
* routes/ws.ts and lib/platform-ws.ts). Normal cookie + CSRF auth on the way in. */
export async function fetchWsTicket(): Promise<string> {
const { ticket } = await apiFetch<{ ticket: string }>("/api/ws/ticket", { method: "POST" });
return ticket;
}
/** Persist the current user's UI language preference (restored on next login). */
export function setLanguagePref(language: Lang): Promise<{ language: Lang }> {
return apiFetch("/api/auth/language", { method: "PUT", body: JSON.stringify({ language }) });
+13 -12
View File
@@ -56,27 +56,28 @@ export interface BackendCheck {
detail?: string;
}
/** Probe a candidate origin by hitting /api/version. That route is behind
* requirePermission("site:read") (session cookie + site:read — see
* apps/server/src/routes/site.ts), so a pre-login probe can never get a 2xx;
* we're not checking "is this reachable and mine to use", only "is something
* that speaks our Fastify auth protocol listening here" — a 401 (missing/bad
* JWT) or 403 (valid session, wrong permission) from THIS specific route is
* as strong a signal of that as a 200 would be, and both are expected outcomes
* pre-login. Uses the same tauri-plugin-http path platformFetch does (raw
* fetch from the webview can't reach an arbitrary LAN host — mixed content,
* see origin.ts). */
/** Probe a candidate origin via GET /health — the server's one unauthenticated
* route (server.ts), which answers `{status:"ok", app:"parking-system"}`. We
* require BOTH a 2xx and that `app` value: the previous probe hit an
* auth-guarded route and accepted 401/403 as "ours", which any password-
* protected service on the LAN would also have passed. Uses the same
* tauri-plugin-http path platformFetch does (raw fetch from the webview can't
* reach an arbitrary LAN host — mixed content, see origin.ts). */
export async function testBackendUrl(url: string): Promise<BackendCheck> {
const origin = url.replace(/\/$/, "");
try {
const { fetch: tauriFetch } = await import("@tauri-apps/plugin-http");
const res = await tauriFetch(`${origin}/api/version`, {
const res = await tauriFetch(`${origin}/health`, {
method: "GET",
signal: AbortSignal.timeout(5000),
});
if (!res.ok && res.status !== 401 && res.status !== 403) {
if (!res.ok) {
return { ok: false, reason: "bad_response", detail: `HTTP ${res.status}` };
}
const body = (await res.json().catch(() => null)) as { app?: unknown } | null;
if (body?.app !== "parking-system") {
return { ok: false, reason: "bad_response", detail: "unexpected /health body" };
}
return { ok: true };
} catch (err) {
return {
+25
View File
@@ -0,0 +1,25 @@
// Desktop-only in-memory CSRF token stash.
//
// tauri-plugin-http's fetch() runs through Rust's reqwest, which keeps its OWN
// cookie jar separate from the webview — document.cookie on tauri://localhost
// never sees the parking_csrf cookie the server sets (open upstream bug,
// tauri-apps/tauri#13045). The cookie IS still sent to the server by reqwest;
// only the client-side READ is broken. So the server echoes the same value in
// the login / me response body (sessionView's csrfToken, routes/auth.ts) and
// the desktop client keeps it here, echoing THIS in X-CSRF-Token instead of
// reading document.cookie.
//
// One module, no imports, so BOTH echo sites can share it without a cycle:
// api.ts (sets it, uses it for apiFetch mutations) and logger.ts (uses it for
// the /api/logs flush — which api.ts imports, so it can't import api.ts back).
// Never persisted: a fresh launch re-learns it via login or /api/auth/me.
let token: string | null = null;
export function setDesktopCsrfToken(value: string | null): void {
token = value;
}
export function getDesktopCsrfToken(): string | null {
return token;
}
+7 -1
View File
@@ -14,7 +14,9 @@
// high-signal sources (failed requests, uncaught errors) are always captured.
import { LOG_LEVEL_ORDER, type ClientLogInput, type LogLevel } from "@parking/shared";
import { getDesktopCsrfToken } from "./desktop-csrf.js";
import { apiUrl, platformFetch } from "./origin.js";
import { inTauri } from "./tauri-env.js";
const ENDPOINT = "/api/logs";
const FLUSH_MS = 4000;
@@ -75,7 +77,11 @@ async function flush(): Promise<void> {
flushing = true;
try {
const headers: Record<string, string> = { "content-type": "application/json" };
const csrf = readCookie(CSRF_COOKIE);
// /api/logs is behind requireAuth → assertCsrf on POST. On desktop the
// cookie is unreadable (see desktop-csrf.ts) — without this branch every
// desktop flush 403'd and was dropped here, silently, by design (found
// 2026-09-04: no desktop client log had EVER reached app_logs).
const csrf = inTauri() ? getDesktopCsrfToken() : readCookie(CSRF_COOKIE);
if (csrf) headers[CSRF_HEADER] = csrf;
await platformFetch(apiUrl(ENDPOINT), {
method: "POST",
+34 -8
View File
@@ -17,8 +17,16 @@
// Browser build: plain pass-through to the real WebSocket (this file's
// createPlatformSocket is only called from inside inTauri() callers).
import { fetchWsTicket } from "../api.js";
import { logClient } from "./logger.js";
import { inTauri } from "./tauri-env.js";
/** Rate-limit the "connect failed" log: use-live-feed reconnects every ≤10s
* forever, and each attempt is a fresh adapter, so without this an outage
* would write six near-identical app_logs rows a minute. */
const CONNECT_FAIL_LOG_INTERVAL_MS = 60_000;
let lastConnectFailLogAt = 0;
export interface PlatformSocket {
onopen: (() => void) | null;
onmessage: ((ev: { data: string }) => void) | null;
@@ -64,13 +72,21 @@ class TauriSocketAdapter implements PlatformSocket {
try {
const { default: TauriWebSocket } = await import("@tauri-apps/plugin-websocket");
if (this.#closed) return; // close() called before connect resolved
// Runs on Tauri's native (Rust) side, NOT inside the webview page — there
// is no page context to auto-attach an Origin header the way a real
// browser WebSocket would. The server's anti-CSWSH check (routes/ws.ts)
// rejects any handshake with a missing/mismatched Origin, so it must be
// set explicitly here to match what WS_ALLOWED_ORIGINS expects
// (tauri://localhost — see apps/server/.env.example).
const conn = await TauriWebSocket.connect(url, { headers: { Origin: "tauri://localhost" } });
// The native WS plugin is a bare tungstenite client: no page context AND
// no cookie jar. Two consequences, both handled via explicit headers:
// - Origin: nothing auto-attaches `Origin: tauri://localhost` the way a
// browser WebSocket would, and routes/ws.ts's anti-CSWSH check rejects a
// missing/mismatched Origin — so set it to match WS_ALLOWED_ORIGINS.
// - Session: the HttpOnly JWT cookie lives in tauri-plugin-http's reqwest
// jar and can't ride on this handshake, so jwtVerify() would 401 every
// connect (the 2026-09-04 "reconnects every 10s forever" bug). Instead,
// mint a single-use ticket over normal HTTP auth and present it in the
// x-ws-ticket header (see routes/ws.ts).
const ticket = await fetchWsTicket();
if (this.#closed) return;
const conn = await TauriWebSocket.connect(url, {
headers: { Origin: "tauri://localhost", "x-ws-ticket": ticket },
});
if (this.#closed) {
void conn.disconnect();
return;
@@ -87,7 +103,17 @@ class TauriSocketAdapter implements PlatformSocket {
});
this.onopen?.();
} catch (err) {
console.error("Tauri WebSocket connect failed:", url, err);
// logClient, not console.error: console output only reaches app_logs at
// debug/trace level, which is how the ticket-less 401 stayed invisible
// for a full day. A closed-before-connect race isn't a failure.
if (!this.#closed && Date.now() - lastConnectFailLogAt > CONNECT_FAIL_LOG_INTERVAL_MS) {
lastConnectFailLogAt = Date.now();
logClient({
level: "error",
message: `desktop live-feed connect failed: ${err instanceof Error ? err.message : String(err)}`,
context: { kind: "desktop_ws_connect_failed", url },
});
}
this.onerror?.();
this.onclose?.();
}
+2 -2
View File
@@ -49,7 +49,7 @@ REGISTRY=git.infra.msai.al/mca/parking_solution
# Staging booth: pinned immutable stage-<sha>. After each promotion (merge dev → stage, CI builds
# :stage-<sha>), bump this to the new sha and re-sync/deploy from Core. The moving `:stage` tag
# exists as the pointer; we deploy the sha, not the mover.
TAG=stage-7d67934
TAG=stage-8fa66c9
COOKIE_SECURE=0
VISION_ENABLED=1
# Desktop app WS handshake: Origin is tauri://localhost (set explicitly by
@@ -82,7 +82,7 @@ REGISTRY=git.infra.msai.al/mca/parking_solution
# Staging booth: pinned immutable stage-<sha>. After each promotion (merge dev → stage, CI builds
# :stage-<sha>), bump this to the new sha and re-sync/deploy from Core. The moving `:stage` tag
# exists as the pointer; we deploy the sha, not the mover.
TAG=stage-7d67934
TAG=stage-8fa66c9
COOKIE_SECURE=0
VISION_ENABLED=1
# Desktop app WS handshake: Origin is tauri://localhost (set explicitly by
+95 -9
View File
@@ -144,8 +144,11 @@ Per the user's choices — the operator **keeps OS access** (no fullscreen lockd
`VITE_API_BASE` correctly set (below), login still failed with WebKit's generic `"Load failed"`.
Root cause is a separate, deeper issue: WebKitGTK treats `tauri://localhost` as a **secure
origin**, so a plain `http://127.0.0.1:3000` `fetch()` — or a `ws://127.0.0.1:3000` WebSocket —
from inside it is blocked as **mixed content**, a long-standing WebKit limitation
([bugs.webkit.org #171934](https://bugs.webkit.org/show_bug.cgi?id=171934)). `connect-src` in the
from inside it is blocked as **mixed content**. (Nearest upstream ticket:
[bugs.webkit.org #171934](https://bugs.webkit.org/show_bug.cgi?id=171934) — note that one is
specifically about *loopback* addresses from https pages; a LAN IP such as `192.168.1.50:3000`
would stay mixed content even if it were fixed, so the plugin route below is the right
architecture for a remote booth regardless, not a stopgap.) `connect-src` in the
CSP does **not** override this — it's a different browser security layer entirely, so the request
never even reaches the network layer to be diagnosable via server logs. **Fix:** two Tauri plugins
route the SPA's traffic through Tauri's native (Rust) side instead of the webview's own
@@ -167,7 +170,9 @@ Per the user's choices — the operator **keeps OS access** (no fullscreen lockd
- **Gotcha (found immediately after shipping the above): the native WS plugin sends no `Origin`
header.** `tauri-plugin-websocket`'s `connect()` runs on Tauri's Rust side, not inside the
webview page — there's no page context to auto-attach `Origin: tauri://localhost` the way a real
browser `WebSocket` would. The server's anti-CSWSH check (`routes/ws.ts`, `isAllowedOrigin`)
browser `WebSocket` would. (The **HTTP** plugin, by contrast, *does* attach that Origin itself —
`tauri-plugin-http/src/commands.rs`, "ensure we have an Origin header set" — so only the WS
path needs the explicit header.) The server's anti-CSWSH check (`routes/ws.ts`, `isAllowedOrigin`)
treats a missing Origin as untrusted and 403s the handshake before touching auth — the live feed
showed **"JASHTË LINJË"** (offline) in the desktop app while the browser showed **"LIVE"**, same
server, same moment. **Fix (two parts, both needed):** `platform-ws.ts`'s `connect()` call now
@@ -332,8 +337,9 @@ runtime-persisted** value.
anyway (the WebKit mixed-content fix above), and those plugins run on the Rust side, **outside**
`connect-src`'s jurisdiction entirely. The real access boundary moved to
`capabilities/default.json`'s `http:default` scope, which is now wildcarded
(`http://*`, `https://*`, `http://*:*`, `https://*:*` — all four forms needed, a known Tauri
scope-matching quirk drops bare `http://*` matches for a `host:port` URL otherwise). `websocket:
(`http://*`, `https://*`, `http://*:*`, `https://*:*` — all four forms needed: the scope is a
URLPattern, and a pattern with no port matches only the scheme's *default* port, so `http://*`
covers `:80` (Caddy) while `http://*:*` is what covers `:3000`). `websocket:
default` already had no scope restriction. Net effect: **the app can now reach any host the
operator types in, and nothing else** — same shape of guarantee as before, just operator-directed
instead of build-directed.
@@ -351,10 +357,12 @@ runtime-persisted** value.
a `tauri-plugin-http` response is stored in that reqwest jar and IS correctly re-sent by
reqwest on later requests (so plain session auth — GETs — silently worked) — but it is **never**
synced into the webview's own cookie store, so `document.cookie` on the `tauri://localhost` page
can never see it. This is an open, unresolved upstream Tauri bug
([tauri-apps/tauri#13045](https://github.com/tauri-apps/tauri/issues/13045),
[#11518](https://github.com/tauri-apps/tauri/issues/11518)) — not something fixable on our side by
changing how/when we read the cookie. Since `api.ts`'s `apiFetch` reads the readable `parking_csrf`
can never see it. Upstream: [tauri-apps/tauri#13045](https://github.com/tauri-apps/tauri/issues/13045)
(open — asks for exactly this jar→webview sync) and
[#11518](https://github.com/tauri-apps/tauri/issues/11518) (closed, without adding a sync) — not
something fixable on our side by changing how/when we read the cookie. The reqwest jar itself
IS persisted (`.cookies` in the app cache dir), so a desktop session survives an app restart
just like the browser's 30-day cookie does. Since `api.ts`'s `apiFetch` reads the readable `parking_csrf`
cookie via `document.cookie` to echo it in `X-CSRF-Token` (double-submit — see
[[local-jwt-auth]]), this meant **every mutating request from the desktop app was silently sending
no CSRF header at all**, pre-dating this runtime-URL change (it was equally true against the old
@@ -369,3 +377,81 @@ runtime-persisted** value.
`assertCsrf()` checks server-side (and reqwest still sends it correctly, per above) — this only
fixes how the desktop *client* learns what value to put in the header, so browser behavior and
server verification are both completely unchanged.
### Live feed needs a WS *ticket*, not the cookie — and desktop logs never reached the server (2026-09-04, v0.1.6)
A retrospective of the 2026-09-03/04 run found that v0.1.4's Origin fix cleared only the **first**
of two gates in `routes/ws.ts`'s preHandler, and that the diagnostic channel everyone was staring
at was itself broken on desktop. Booth evidence: `docker logs park-2-server-1 | grep /api/ws` showed
a fresh handshake every 10 s (use-live-feed's capped backoff), i.e. every connect rejected.
- **Gate two: `req.jwtVerify()` reads the HttpOnly `parking_token` cookie — which the WebSocket
plugin cannot send.** `tauri-plugin-websocket` is a bare tokio-tungstenite client with **no
cookie jar at all** (its source has no cookie handling); the cookie lives in
`tauri-plugin-http`'s reqwest jar and is HttpOnly besides, so JS can't copy it across either.
Origin OK + no cookie → 401 → reconnect forever. **Fix: a single-use WS ticket.** The desktop
client `POST`s `/api/ws/ticket` over normal HTTP auth (cookie + CSRF, which it CAN do) and gets
a 32-byte random ticket bound to its user, valid 30 s, single-use, in-memory only; it presents
it in an `x-ws-ticket` header on the handshake (`platform-ws.ts`), and the preHandler accepts
ticket-or-cookie *after* the Origin check, then does the same `report:read` role check for both.
A browser page can't set custom WebSocket headers, so the ticket path is unreachable from a
browser and adds no CSWSH surface. **Rejected:** echoing the JWT in the login body and sending
it as `Authorization: Bearer` (fastify-jwt would accept it) — that puts the session token in JS,
which HttpOnly exists to prevent; the ticket keeps it out. Verified locally with an 11-case
handshake script: ticket/no-cookie → 101 + hello; reused/bogus/absent → 401; ticket + bad Origin
→ 403; cookie path unchanged. **Field-verified 2026-09-04:** v0.1.6 on the park-2 booth against
image `stage-8fa66c9` shows **LIVE** — the first desktop build to do so.
- **Desktop client logs had never reached `app_logs`.** `logger.ts`'s flush read the CSRF token
from `document.cookie` (null on desktop — the same jar split as above), so every
`POST /api/logs` from the desktop 403'd under `requireAuth`→`assertCsrf`, and the flush drops
failures by design (loop safety). Consequences: the 2026-09-03 "route update-failure logging
through logClient" fix wrote to a dead channel, and the v0.1.5 CSRF fix patched `api.ts` but not
`logger.ts`. **Fix:** the stash moved to a dependency-free `lib/desktop-csrf.ts` (so `logger.ts`
can read it without importing `api.ts`, which imports `logger.ts`), and the flush uses it when
`inTauri()`. `platform-ws.ts`'s connect failure now goes through `logClient` too (rate-limited
to one row/min — reconnects are every ≤10 s), instead of `console.error`, which only forwards at
debug/trace.
- **ConnectScreen probe now hits `/health`.** The v0.1.5 probe hit an auth-guarded route and
treated 401/403 as "ours" — any password-protected service on the LAN would have passed it, and
the comment claiming no unauthenticated route existed was wrong (`/health` has been there all
along). `/health` now also returns `app: "parking-system"`, and the probe requires both a 2xx
and that value.
- **Why every one of these was found in the field:** `tauri dev` loads `http://localhost:5173`,
not `tauri://localhost`, so the relative-URL error, mixed content, the missing Origin, and the
cookie-jar split *cannot* reproduce in dev mode. The pre-tag gate is now: build the bundle
locally, run the AppImage against a local server, log in, confirm **LIVE**, do one mutation,
and confirm a desktop-sourced row appears in the Logs viewer (`apps/desktop/README.md`).
### In-app update never worked: the manifest only described the AppImage, the booths run the .deb (2026-09-04)
Every self-update attempt from v0.1.0 through v0.1.6 ended the same way — prompt, download
traffic, then nothing, and the same prompt again next launch. The version-sync (v0.1.2) and
error-logging fixes were real but not the cause. **Root cause:** `tauri-plugin-updater` resolves
the download target as `{os}-{arch}-{installer}` **first** (`linux-x86_64-deb` here — the
bundler stamps `__TAURI_BUNDLE_TYPE_VAR_DEB` into the `.deb`'s binary, verified with `strings`
on a local build), then falls back to bare `{os}-{arch}`. `release.yml`'s `latest.json` carried
**only** `linux-x86_64`, pointing at the **AppImage**. So a `.deb` install found the update,
downloaded the AppImage, verified its signature (which was correct — for the AppImage), then
handed the bytes to `install_deb()`, whose first line checks `infer::archive::is_deb(bytes)` and
returns `InvalidUpdaterFormat`. Before v0.1.6 that error never reached the server (the desktop
log channel was itself broken — see the previous section), so it looked like a silent no-op.
Sources: `tauri-plugin-updater-2.10.1/src/updater.rs` (`get_urls`, `install_inner`,
`install_deb`), `tauri-utils/src/platform.rs` (`bundle_type`).
- **Fix:** `latest.json` now carries one signed entry per installer — `linux-x86_64-deb`,
`linux-x86_64-rpm` (when built), and bare `linux-x86_64` for the AppImage — assembled by a
small Node script in the workflow (the `.sig` files for `.deb`/`.rpm` were already being
produced and uploaded, just never referenced).
- **What a booth update now looks like:** prompt → download → **polkit password dialog**
(`pkexec dpkg -i`) → relaunch into the new version. The prompt is deliberate, not a wart: the
package is root-owned in `/usr/bin`, and under the [[threat-model]] the operator must not be
able to replace the app silently; whoever brings the box online for an update is the admin.
Cancelling the dialog leaves the old version running and logs
`desktop_update_install_failed` to `app_logs`.
- **Rejected:** switching booths to the AppImage so updates need no privilege. It would work
(the updater rewrites the AppImage in place), but the binary would then be operator-writable,
it needs FUSE on the appliance image, and launcher/autostart integration becomes manual —
three regressions to avoid one password prompt.
- **Judgment note for the retrospective:** three fixes were shipped against this symptom
without reading the updater's install path once. The whole chain is ~60 lines of vendored
Rust in `~/.cargo/registry`; it names the exact failure (`InvalidUpdaterFormat`).
+32
View File
@@ -2841,3 +2841,35 @@ value in the login/me JSON body; the desktop client stashes it in memory and ech
reading document.cookie. assertCsrf() itself is untouched — the cookie is still what's verified,
and reqwest was already sending it correctly; this only fixes how the desktop client *learns* the
value. Full detail (including the exact CSP/capability tradeoffs) on [[desktop-shell-tauri]].
## [2026-09-04] fix | Desktop live feed: WS handshake can't carry the cookie → single-use ticket; desktop logs never reached app_logs
Retrospective of the 2026-09-03/04 desktop run (six releases in 26 h) found the v0.1.4 Origin fix
cleared only gate one of two in routes/ws.ts: gate two is req.jwtVerify() reading the HttpOnly
cookie, and tauri-plugin-websocket has no cookie jar at all — so every desktop handshake 401'd and
use-live-feed reconnected every 10 s (confirmed in the park-2 server log). Fixed with a 30-second,
single-use, in-memory WS ticket minted by POST /api/ws/ticket over normal cookie+CSRF auth and
presented in an x-ws-ticket header; Origin check still runs first, browser path unchanged, JWT
stays out of JS. Second finding: logger.ts read the CSRF cookie via document.cookie, null on
desktop, so every desktop POST /api/logs 403'd and was dropped silently — no desktop client log had
EVER reached app_logs, which is why "no logs whatsoever" kept happening and why yesterday's
logClient fix couldn't help. Stash moved to lib/desktop-csrf.ts, shared by api.ts and logger.ts;
WS connect failures now go through logClient (rate-limited). Third: the ConnectScreen probe now
uses the unauthenticated /health (extended with app: "parking-system") instead of accepting any
401. Also corrected four wiki citations (WebKit 171934 scope, tauri#11518 is closed, the HTTP
plugin does set Origin itself, the http-scope "quirk" is URLPattern default-port semantics) and
added a local-AppImage pre-tag gate to the desktop README, since tauri dev cannot reproduce any
of these origin-dependent bugs. Full detail on [[desktop-shell-tauri]].
## [2026-09-04] fix | Desktop in-app update never worked: latest.json described only the AppImage, booths run the .deb
tauri-plugin-updater looks up `{os}-{arch}-{installer}` first (linux-x86_64-deb — the bundler
stamps the installer type into the binary; verified with strings on a local .deb) and only then
bare linux-x86_64. release.yml's latest.json carried only the bare key → the AppImage, so every
.deb install downloaded the AppImage, passed signature verification, then failed install_deb()'s
is_deb check with InvalidUpdaterFormat — invisible until v0.1.6 fixed the desktop log channel.
This, not version drift or swallowed errors, is why v0.1.0→…→v0.1.6 never self-updated.
latest.json now has one signed entry per installer (deb, rpm, AppImage); a .deb update ends in a
polkit password prompt (pkexec dpkg -i), which is the intended admin gate on a root-installed
package. README + [[desktop-shell-tauri]] updated. First real test: tag v0.1.7 and accept the
prompt on the v0.1.6 booth.