Actually fix WNS push, and stop the notification toggle hanging forever (#56)

The pywebpush version bump alone didn't fix WNS: even the latest
release (2.4.0) has no WNS-specific header handling in its own
source, confirmed by inspecting the installed package directly.
Adds the required X-WNS-Cache-Policy header ourselves via
webpush()'s own headers= param, gated to *.notify.windows.com
endpoints.

Also: subscribeToPush()'s permission request and service-worker-ready
wait had no timeout, so a browser that never settles either (seen
live on a fresh Windows/Edge install -- greyed out, no prompt, no
error) left the toggle stuck forever with no feedback. Both now time
out after 20s with an actionable message instead.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

backend/app/services/push_service.py

@@ -2,6 +2,7 @@ import asyncio
 import json
 import logging
 import uuid
+from urllib.parse import urlparse
 
 from pywebpush import WebPushException, webpush
 from sqlalchemy import delete, select
@@ -53,7 +54,23 @@ async def unsubscribe(db: AsyncSession, user_id: uuid.UUID, endpoint: str) -> No
     await db.commit()
 
 
+# #56 correction: bumping to pywebpush's latest release (2.4.0) turned out
+# not to actually fix WNS -- checked the installed package's own source
+# directly and it has no WNS-specific code anywhere; the upstream
+# discussion (web-push-libs/pywebpush#162) apparently never shipped.
+# Worked around here instead, using the `headers` param webpush() already
+# exposes for exactly this: WNS (Windows/Edge push,
+# *.notify.windows.com) has required this header since April 2024, or it
+# 400s with no useful body -- "cache" for a non-zero TTL, "no-cache" for
+# zero (this app never sets a TTL, so always the latter).
+def _is_wns_endpoint(endpoint: str) -> bool:
+    return urlparse(endpoint).hostname is not None and urlparse(endpoint).hostname.endswith(
+        "notify.windows.com"
+    )
+
+
 def _send_one(subscription: PushSubscription, payload: dict) -> None:
+    extra_headers = {"X-WNS-Cache-Policy": "no-cache"} if _is_wns_endpoint(subscription.endpoint) else None
     webpush(
         subscription_info={
             "endpoint": subscription.endpoint,
@@ -62,6 +79,7 @@ def _send_one(subscription: PushSubscription, payload: dict) -> None:
         data=json.dumps(payload),
         vapid_private_key=settings.vapid_private_key,
         vapid_claims={"sub": settings.vapid_subject},
+        headers=extra_headers,
     )
 
 

backend/pyproject.toml

@@ -15,13 +15,13 @@ dependencies = [
     "pydantic-settings>=2.6",
     "argon2-cffi>=23.1",
     "itsdangerous>=2.2",
-    # #56: >=2.0 let the production venv sit on an old 2.0.x that predates
-    # the fix for web-push-libs/pywebpush#162 -- WNS (Windows/Edge push)
-    # has required an X-WNS-Cache-Policy header since April 2024, and an
-    # old pywebpush that never sends it gets a bodyless 400 on every send.
-    # Floored at the actual latest release (merged/shipped 2026-01 through
-    # 2026-08) rather than pinning to whichever exact point release first
-    # included the fix, since that wasn't independently confirmed.
+    # #56: >=2.0 let the production venv sit on an old 2.0.x with no real
+    # downside to bumping the floor -- worth keeping current regardless.
+    # Doesn't by itself fix WNS (Windows/Edge push): despite
+    # web-push-libs/pywebpush#162's discussion, even the latest release
+    # (2.4.0) has no WNS-specific header handling in its own source. The
+    # actual fix is app/services/push_service.py adding the required
+    # X-WNS-Cache-Policy header itself via webpush()'s `headers` param.
     "pywebpush>=2.4.0",
     "redis>=5.0",
     "httpx>=0.27",

backend/tests/test_push.py

@@ -272,3 +272,61 @@ def test_non_gone_push_failure_logs_response_detail_and_keeps_subscription(ws_cl
     assert len(calls) == 1
     assert "Bad Request" in calls[0]
     assert "Ttl value conflicts with X-WNS-Cache-Policy" in calls[0]
+
+
+def test_wns_endpoint_gets_cache_policy_header(ws_client, monkeypatch):
+    calls = []
+    monkeypatch.setattr("app.services.push_service.webpush", lambda **kw: calls.append(kw))
+
+    alice = _register_ws(ws_client, _unique("alice"))
+    room = ws_client.post("/api/rooms", json={"name": _unique("general")}).json()
+
+    bob = _register_ws(ws_client, _unique("bob"))
+    ws_client.post(f"/api/rooms/{room['id']}/join")
+    ws_client.post(
+        "/api/push/subscribe",
+        json={
+            "endpoint": f"https://wns2-by3p.notify.windows.com/w/{_unique('bob')}",
+            "keys": {"p256dh": "p256dh-bob", "auth": "auth-bob"},
+        },
+    )
+
+    ws_client.post(
+        "/api/auth/login", json={"username_or_email": alice["username"], "password": "password123"}
+    )
+    with ws_client.websocket_connect("/ws/chat") as ws:
+        ws.send_json({"type": "join", "room_id": room["id"]})
+        assert ws.receive_json()["type"] == "joined"
+        ws.send_json({"type": "message", "room_id": room["id"], "content": "hello"})
+        assert ws.receive_json()["type"] == "message"
+        ws.send_json({"type": "join", "room_id": room["id"]})
+        assert ws.receive_json()["type"] == "joined"
+
+    assert len(calls) == 1
+    assert calls[0]["headers"] == {"X-WNS-Cache-Policy": "no-cache"}
+
+
+def test_non_wns_endpoint_gets_no_extra_headers(ws_client, monkeypatch):
+    calls = []
+    monkeypatch.setattr("app.services.push_service.webpush", lambda **kw: calls.append(kw))
+
+    alice = _register_ws(ws_client, _unique("alice"))
+    room = ws_client.post("/api/rooms", json={"name": _unique("general")}).json()
+
+    bob = _register_ws(ws_client, _unique("bob"))
+    ws_client.post(f"/api/rooms/{room['id']}/join")
+    ws_client.post("/api/push/subscribe", json=_subscription_payload(_unique("bob")))
+
+    ws_client.post(
+        "/api/auth/login", json={"username_or_email": alice["username"], "password": "password123"}
+    )
+    with ws_client.websocket_connect("/ws/chat") as ws:
+        ws.send_json({"type": "join", "room_id": room["id"]})
+        assert ws.receive_json()["type"] == "joined"
+        ws.send_json({"type": "message", "room_id": room["id"], "content": "hello"})
+        assert ws.receive_json()["type"] == "message"
+        ws.send_json({"type": "join", "room_id": room["id"]})
+        assert ws.receive_json()["type"] == "joined"
+
+    assert len(calls) == 1
+    assert calls[0]["headers"] is None

frontend/src/lib/push.ts

@@ -1,5 +1,33 @@
 import { getVapidPublicKey, subscribePush, unsubscribePush } from '../api/push'
 
+// A stuck permission prompt or service-worker-ready wait would otherwise
+// leave TopBar's "Enable notifications" toggle permanently disabled with
+// no feedback at all -- confirmed live on a fresh Windows/Edge install
+// (never been used before): clicking it greyed the button out and never
+// showed the OS permission prompt, with no error and no way out short of
+// reloading. Most likely cause is Windows' own per-app notification
+// permission being off for Edge (Settings > System > Notifications), which
+// some Chromium versions handle by just never resolving the request
+// instead of rejecting it -- but whatever the cause, the UI should never
+// hang forever waiting on a browser API that may simply never settle.
+class PushTimeoutError extends Error {}
+
+function withTimeout<T>(promise: Promise<T>, ms: number, message: string): Promise<T> {
+  return new Promise((resolve, reject) => {
+    const timer = setTimeout(() => reject(new PushTimeoutError(message)), ms)
+    promise.then(
+      (value) => {
+        clearTimeout(timer)
+        resolve(value)
+      },
+      (err: unknown) => {
+        clearTimeout(timer)
+        reject(err)
+      },
+    )
+  })
+}
+
 function urlBase64ToUint8Array(base64String: string): Uint8Array<ArrayBuffer> {
   const padding = '='.repeat((4 - (base64String.length % 4)) % 4)
   const base64 = (base64String + padding).replace(/-/g, '+').replace(/_/g, '/')
@@ -27,7 +55,11 @@ export async function subscribeToPush(): Promise<void> {
     throw new Error('Push notifications are not supported in this browser')
   }
 
-  const permission = await Notification.requestPermission()
+  const permission = await withTimeout(
+    Notification.requestPermission(),
+    20_000,
+    "The browser never responded to the notification permission request. Check this browser's notification permission for this site, and your OS-level notification settings for the browser, then try again.",
+  )
   if (permission !== 'granted') {
     throw new Error('Notification permission was not granted')
   }
@@ -37,7 +69,11 @@ export async function subscribeToPush(): Promise<void> {
     throw new Error('Push notifications are not configured on the server')
   }
 
-  const registration = await navigator.serviceWorker.ready
+  const registration = await withTimeout(
+    navigator.serviceWorker.ready,
+    20_000,
+    'The browser never finished setting up its background service worker. Try reloading the page.',
+  )
   const subscription = await registration.pushManager.subscribe({
     userVisibleOnly: true,
     applicationServerKey: urlBase64ToUint8Array(public_key),