Merge pull request #89 from saphid/feat/headset-telemetry-id

Frame Control 0.4.5: count headsets, lighter Comfort checks, sturdier APK installs

app/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "frame-control",
-  "version": "0.4.3",
+  "version": "0.4.5",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "frame-control",
-      "version": "0.4.3",
+      "version": "0.4.5",
       "license": "MIT",
       "devDependencies": {
         "electron": "^44.4.5",

app/package.json

@@ -1,7 +1,7 @@
 {
   "name": "frame-control",
   "productName": "Frame Control",
-  "version": "0.4.3",
+  "version": "0.4.5",
   "description": "Desktop app for managing a Valve Steam Frame over SSH",
   "private": true,
   "main": "main.js",

docs/privacy.md

@@ -33,6 +33,12 @@ computer, exactly as they were sent.
   yours.
 - Every event includes the app version, OS name (macOS, Windows or Linux),
   CPU architecture and Python version.
+- `frame_connected` carries `headset_id`: the first 16 hex digits of a
+  SHA-256 hash of the headset's `/etc/machine-id`, worked out on the headset,
+  so the machine-id itself never leaves it. It is the same for one headset on
+  every computer, which lets us count headsets rather than installs; it can't
+  be turned back into the machine-id or a serial number. `headsets_saved` is how
+  many headsets this copy of Frame Control has set up.
 
 ## Usage events
 
@@ -41,7 +47,7 @@ computer, exactly as they were sent.
 | `app_installed` | First run | |
 | `app_updated` | First run of a new version | `from_version` |
 | `app_opened` | At most once a day | |
-| `frame_connected` | The first time a SteamOS build is seen | `steamos_build`, `steamos_version` |
+| `frame_connected` | The first time a SteamOS build is seen on each headset | `steamos_build`, `steamos_version`, `headset_id`, `headsets_saved` |
 | `tab_viewed` | The first click on each tab in a session | `tab` |
 | `install_finished` | Any install finishes, working or not | `kind` (apk, flatpak, steam, title, web), `ok`, `seconds`, `error_category` (only when `ok` is false), `installer_code`, `xr_layer_missing`, and see below |
 | `update_offered`, `update_started`, `update_failed` | The update banner | `to_version`, `error_category` |
@@ -51,12 +57,15 @@ error becomes one category from this fixed list, plus Android's own
 `INSTALL_FAILED_…` code (`installer_code`) when there is one:
 `android_installer`, `apk_needs_newer_android`, `apk_wrong_abi`,
 `layer_missing`, `apk_repack_failed`, `tool_missing`, `apk_unreadable`,
-`cant_run_on_frame`, `steam_shortcut`, `frame_not_set_up`, `frame_auth`,
-`frame_unreachable`, `frame_disk_full`, `download_failed`, `flatpak`,
-`cancelled`, `lepton` or `other`. A VR APK that installed without Frame
-Control's OpenXR compatibility layer, because this copy of the app is missing
-it, carries `xr_layer_missing: true`; otherwise that field is left out. It
-names what was installed only when that's already public:
+`cant_run_on_frame`, `steam_shortcut`, `steam_not_running`,
+`steam_not_responding`, `steam_js_error`, `artwork_failed`, `copy_timeout`,
+`frame_not_set_up`, `frame_auth`, `frame_unreachable`, `frame_disk_full`,
+`copy_failed`, `download_failed`, `flatpak`, `cancelled`, `lepton`,
+`internal_error` (a mistake in Frame Control's own code) or `other`. A VR
+APK that installed without Frame Control's OpenXR compatibility layer,
+because this copy of the app is missing it, carries `xr_layer_missing: true`;
+otherwise that field is left out. It names what was installed only when
+that's already public:
 
 - F-Droid catalogue apps: `package`. Never the version, since a local build can reuse a
   catalogue app's package name

tests/test_comfort.py

@@ -256,3 +256,47 @@ class SensorTests(unittest.TestCase):
                 self.assertIsNone(status.activity_level())
         with patch.object(status, 'run', return_value='[{"operation":"status","activity_level":3}]'):
             self.assertEqual(status.activity_level(), 3)
+
+
+class DesktopBundleTests(unittest.TestCase):
+    """server.comfort runs the copy cached on the headset; the sources go only when it's missing."""
+
+    @unittest.skipUnless(os.name == "posix", "checks POSIX permissions")
+    def test_sources_are_sent_once_per_version(self):
+        import server
+        with tempfile.TemporaryDirectory() as tmp:
+            here, home = Path(tmp) / 'ui', Path(tmp) / 'frame'
+            here.mkdir()
+            home.mkdir()
+            for name in server.COMFORT_FILES:
+                (here / name).write_text('# ' + name + '\n')
+            (here / 'frame_comfort.py').write_text('import json, sys\nprint(json.dumps({"active": False, "got": sys.argv[1]}))\n')
+            sent = []
+
+            def ssh(remote, *, stdin, timeout):  # the headset: the script run under its own home
+                sent.append(stdin)
+                return subprocess.run([sys.executable, '-'], input=stdin, capture_output=True, text=True, check=True,
+                                      env={**os.environ, 'HOME': str(home)}).stdout
+
+            with patch.object(server, 'HERE', here), patch.object(server, 'ssh', side_effect=ssh):
+                first = server.comfort({'action': 'status'})
+                self.assertEqual(json.loads(first['got']), {'action': 'status'})
+                self.assertEqual(len(sent), 2)  # the check, then the sources
+                self.assertNotIn('# frame_steam.py', sent[0])
+                self.assertIn('# frame_steam.py', sent[1])
+                self.assertEqual(server.comfort({'action': 'status'}), first)
+                self.assertEqual(len(sent), 3)
+                self.assertNotIn('# frame_steam.py', sent[2])
+                cached, = (home / '.cache/frame-control/comfort').iterdir()
+                self.assertEqual(sorted(p.name for p in cached.iterdir()), sorted(server.COMFORT_FILES))
+                self.assertEqual(cached.stat().st_mode & 0o777, 0o700)
+                self.assertTrue(all(p.stat().st_mode & 0o777 == 0o600 for p in cached.iterdir()))
+                # A copy cut short is completed; a changed helper is a new version.
+                (cached / 'frame_status.py').unlink()
+                server.comfort({'action': 'status'})
+                self.assertEqual(len(sent), 5)
+                self.assertTrue((cached / 'frame_status.py').exists())
+                (here / 'frame_steam.py').write_text('# frame_steam.py, changed\n')
+                server.comfort({'action': 'status'})
+                self.assertEqual(len(sent), 7)
+                self.assertEqual(len(list((home / '.cache/frame-control/comfort').iterdir())), 2)

tests/test_comfort_ui.py

@@ -59,3 +59,32 @@ const log = ()=>{}, toast = ()=>{};
 '''
         result = subprocess.run(['node', '-e', setup + code + checks], capture_output=True, text=True)
         self.assertEqual(result.returncode, 0, result.stderr)
+
+    def test_poll_waits_while_the_frame_is_unreachable(self):
+        page = (ROOT / 'ui/index.html').read_text(encoding='utf-8')
+        start = page.index('async function pollComfort()')
+        code = page[start:page.index('$("comfortForm").onsubmit', start)]
+        js = r'''
+const assert = require('node:assert/strict');
+const el = {textContent: 'Checking session…'};
+const $ = () => el;
+let online = false, comfortBusy = false, calls = 0, timers = 0;
+const api = async () => { calls++; return {}; };
+const renderComfort = () => {};
+const setTimeout = () => { timers++; };
+''' + code + r'''
+(async () => {
+  await pollComfort();
+  assert.equal(calls, 0);  // no ssh attempt while the status check can't reach the Frame
+  assert.match(el.textContent, /when the Frame is connected/);
+  assert.equal(timers, 1);  // still scheduled, to resume when it's back
+  online = true;
+  await pollComfort();
+  assert.equal(calls, 1);
+  online = null;  // a status check that failed for another reason: keep polling
+  await pollComfort();
+  assert.equal(calls, 2);
+})().catch(e => { console.error(e); process.exitCode = 1; });
+'''
+        result = subprocess.run(['node', '-e', js], capture_output=True, text=True)
+        self.assertEqual(result.returncode, 0, result.stderr)

tests/test_frame_android_library.py

@@ -423,6 +423,19 @@ class InstallTests(unittest.TestCase):
         self.assertIn(('remove', '3346865537'), [c.args for c in api.call_args_list])
         self.assertTrue(any(c.args[0] == 'rm -rf Applications/Android/org.test.vr' for c in ssh.call_args_list))
 
+    def test_failed_update_removes_partial_apk_and_keeps_the_app(self):
+        with patch.object(android.frame_artwork, 'prepare', return_value=(self.images, [])), \
+                patch.object(android, 'read_meta', return_value=dict(self.existing)), \
+                patch.object(android, '_copy', side_effect=android.FrameError('copying x.apk to the Frame timed out')), \
+                patch.object(android, 'ssh', return_value='') as ssh, \
+                patch.object(android, 'shortcut_tool') as api:
+            with self.assertRaisesRegex(android.FrameError, 'timed out'):
+                android._install('x.apk', self.info, 'org.test.vr', False, None, None)
+        commands = [c.args[0] for c in ssh.call_args_list]
+        self.assertIn('rm -f Applications/Android/org.test.vr/app.apk.part', commands)
+        self.assertFalse(any(c.startswith('rm -rf') for c in commands))
+        api.assert_not_called()
+
     def test_remove_keeps_data_when_requested_and_survives_steam_failure(self):
         with patch.object(android, '_meta_or_fail', side_effect=lambda pkg: dict(self.existing)), \
                 patch.object(android, 'stop'), \
@@ -486,6 +499,49 @@ class InstallTests(unittest.TestCase):
         self.assertIn('podman stop -t 5 lepton-steamlaunch-2800000001', ssh.call_args.args[0])
 
 
+class CopyAndSshTests(unittest.TestCase):
+    """The commands behind an install fail as FrameErrors with readable words, and a large APK
+    gets the time it needs."""
+
+    def fake(self, stderr, code=1, stdout=b''):
+        """frame_host.run_ssh running a stand-in that writes these bytes, with the caller's options."""
+        script = (f'import sys; sys.stdout.buffer.write({stdout!r}); sys.stderr.buffer.write({stderr!r}); '
+                  f'sys.exit({code})')
+        return lambda argv, **kwargs: subprocess.run([sys.executable, '-c', script], **kwargs)
+
+    def test_large_apk_copy_timeout_grows_with_its_size(self):
+        seen = {}
+        with tempfile.NamedTemporaryFile(suffix='.apk') as f, \
+                patch.object(android.frame_host, 'run_ssh', side_effect=lambda argv, **kw: seen.update(kw)), \
+                patch.object(android.os.path, 'getsize', return_value=2 * 1024 ** 3):
+            android._copy(f.name, 'Applications/Android/x/app.apk.part')
+        # 2 GB at the slowest rate allowed: 4096 s, not the old fixed 600 (300 MB at 4 Mbit/s).
+        self.assertGreaterEqual(seen['timeout'], 2 * 1024 ** 3 / (512 * 1024))
+        with tempfile.NamedTemporaryFile(suffix='.apk') as f, \
+                patch.object(android.frame_host, 'run_ssh', side_effect=lambda argv, **kw: seen.update(kw)):
+            android._copy(f.name, 'x/launch.sh', timeout=120)
+        self.assertEqual(seen['timeout'], 120)  # a small file keeps the least time given
+
+    def test_copy_failure_in_another_encoding_is_a_frame_error(self):
+        # A message from a Windows code page, or a file name, isn't UTF-8.
+        with tempfile.NamedTemporaryFile(suffix='.apk') as f, \
+                patch.object(android.shutil, 'which', return_value=None), \
+                patch.object(android.frame_host, 'run_ssh', side_effect=self.fake(b'scp: Verbindung \x81ber\xff\n')):
+            with self.assertRaisesRegex(android.FrameError, r'\Acopying .* to the Frame failed: scp: Verbindung'):
+                android._copy(f.name, 'x/app.apk.part')
+
+    def test_ssh_failure_messages_are_text(self):
+        with patch.object(android.frame_host, 'run_ssh', side_effect=self.fake(b'mv: \xff cannot move\n')):
+            with self.assertRaisesRegex(android.FrameError, r'\Amv: .* cannot move\Z'):
+                android.ssh('mv a b')
+        with patch.object(android.frame_host, 'run_ssh', side_effect=self.fake(b'cat: write error: No space left on device\n')):
+            with self.assertRaises(android.FrameError) as caught:
+                android.ssh('cat > x', input=b'\x89PNG')  # bytes in, bytes out
+        self.assertEqual(str(caught.exception), 'cat: write error: No space left on device')
+        with patch.object(android.frame_host, 'run_ssh', side_effect=self.fake(b'', code=0, stdout=b'ok \xff\n')):
+            self.assertEqual(android.ssh('echo ok').strip(), 'ok �')
+
+
 class SteamAPITests(unittest.TestCase):
     def test_artwork_api_enums_and_safe_serialization(self):
         with patch.object(shortcuts, 'evaluate', return_value={'warnings': []}) as evaluate:

tests/test_frame_apk.py

@@ -93,6 +93,20 @@ class ApkInfo(unittest.TestCase):
             frame_apk.MAX_MANIFEST = limit
         self.assertLess(peak, 8 * 1024**2)
 
+    def test_damaged_download_is_an_apk_error(self):
+        # Corrupt deflate data used to escape as zlib.error, which the upload path didn't catch
+        # (a 500, and no install report) and frame_android.install reported as 'other'.
+        buf = io.BytesIO()
+        with zipfile.ZipFile(buf, 'w', zipfile.ZIP_DEFLATED) as z:
+            z.writestr('AndroidManifest.xml', manifest('com.example.bad', 0x7f010000, 0x7f010001, 21) * 20)
+        data = bytearray(buf.getvalue())
+        at = data.index(b'PK\x03\x04')
+        start = at + 30 + sum(struct.unpack_from('<HH', data, at + 26))  # past the name and extra field
+        for i in range(start, start + 40):
+            data[i] ^= 0x5a
+        with self.assertRaisesRegex(frame_apk.ApkError, 'could not read the APK manifest'):
+            self.read(bytes(data))
+
     def test_refuses_compression_android_cant_read(self):
         for method in (zipfile.ZIP_BZIP2, zipfile.ZIP_LZMA):
             buf = io.BytesIO()

tests/test_link.py

@@ -46,6 +46,17 @@ def explain(msg):
     return None
 
 
+class FailureDetail(unittest.TestCase):
+    def test_an_address_that_failed_says_why_in_a_few_words(self):
+        # #87: the banner put each address's detail after the headline, and with the whole
+        # message as the detail it said everything twice ("...may be answering here..").
+        self.assertEqual(fl.failure_detail({"message": "A long sentence.", "short": "SSH identity changed"}),
+                         "SSH identity changed")
+        self.assertEqual(fl.failure_detail({"message": "The connection to the Frame dropped. Try again."}),
+                         "The connection to the Frame dropped. Try again.")
+        self.assertEqual(fl.failure_detail(None), "SSH failed")
+
+
 class Probe(unittest.TestCase):
     def test_answers_refusals_and_unknown_names(self):
         with socket.socket() as srv:
@@ -200,7 +211,8 @@ class Connecting(unittest.TestCase):
         rows = {p["host"]: p for p in s["probes"]}
         self.assertEqual(rows["nothing.invalid"]["state"], "unresolved")
         self.assertEqual(rows["::1"]["state"], "sshfailed")
-        self.assertIn("different SSH identity", rows["::1"]["detail"])
+        # Its row says why in a few words: the banner's headline already has the whole sentence (#87).
+        self.assertEqual(rows["::1"]["detail"], "SSH identity changed")
         # Every ssh command was pointed at the winner, with the host key pinned per device.
         alias, opts = self.routes[-1]
         self.assertEqual(alias, "frame-t")
@@ -260,6 +272,8 @@ class Connecting(unittest.TestCase):
         self.assertEqual((s["phase"], s["error"]["stage"]), ("failed", "login"))
         self.assertIn("SSH key", s["error"]["message"])
         self.assertEqual(s["error"]["auth"], "denied")  # the page offers Set Up Connection
+        self.assertEqual(s["error"]["short"], "This computer's key refused")  # the header pill's words (#87)
+        self.assertEqual(s["probes"][0]["detail"], "This computer's key refused")
 
     def test_pinned_identity_is_checked_strictly(self):
         d = self.device("localhost")
@@ -349,6 +363,7 @@ class Connecting(unittest.TestCase):
                 self.assertEqual((s["phase"], s["error"]["stage"]), ("failed", "identity"))
                 self.assertEqual(s["error"]["hostkey"]["state"], "missing")
                 self.assertIn("missing or can't be read", s["error"]["message"])
+                self.assertEqual(s["error"]["short"], "Saved SSH identity missing")
                 view = next(x for x in fl.devices_view(self.link)["devices"] if x["id"] == d["id"])
                 self.assertTrue(view["identity_lost"])
                 scan = fl.devices_action(self.link, {"action": "identity-scan", "id": d["id"]}, None)["scan"]
@@ -634,6 +649,7 @@ class Connecting(unittest.TestCase):
         self.assertEqual((shown["state"], shown["type"], shown["host"]), ("changed", "ssh-ed25519", "localhost"))
         self.assertTrue(shown["fingerprint"].startswith("SHA256:"))
         self.assertIn("Devices tab", s["error"]["message"])
+        self.assertEqual(s["error"]["short"], "SSH identity changed")
         self.assertEqual(fd.known_hosts(d["id"]).read_text(), saved)  # never accepted on its own
         r = self.command(d)
         self.assertEqual(r.returncode, 255)

tests/test_link_ui.py

@@ -0,0 +1,136 @@
+"""The connection pill, the offline banner and the Devices tab's selection: the page's own
+functions from ui/index.html, run in Node against a minimal DOM (#87)."""
+import sandbox  # noqa: F401  (first: keeps tests off real data and services)
+import pathlib
+import re
+import shutil
+import subprocess
+import unittest
+
+ROOT = pathlib.Path(__file__).resolve().parents[1]
+PAGE = (ROOT / "ui/index.html").read_text(encoding="utf-8")
+
+
+def between(start, end):
+    return PAGE[PAGE.index(start):PAGE.index(end)]
+
+
+SETUP = r'''
+const assert = require('node:assert/strict');
+const elements = new Map();
+const $ = id => {
+  if (!elements.has(id)) elements.set(id, {id, textContent: '', innerHTML: '', title: '', hidden: true, value: ''});
+  return elements.get(id);
+};
+const esc = s => String(s).replace(/[&<>"']/g, c => ({'&': '&amp;', '<': '&lt;', '>': '&gt;', '"': '&quot;', "'": '&#39;'}[c]));
+const serverNow = () => 1000;
+const activeStage = s => (s.stages || []).find(x => x.state === 'active');
+const ago = () => 'a minute ago';
+const openTrust = () => {};
+const window = {};
+let online = false;
+const link = {live: true, s: null};
+const dv = {list: [], sel: null, data: null, found: 'x', editing: 'y'};
+let rendered = 0;
+const renderDevices = () => { rendered++; };
+let refuse = false;
+const act = async (label, fn) => refuse ? undefined : fn();
+const api = async () => ({ok: true});
+const STAGES = [{id: 'find', label: 'Finding the headset', state: 'done'},
+                {id: 'identity', label: "Checking the headset's identity", state: 'failed'}];
+const changed = {phase: 'failed', retry_at: 1012, device: {id: 'd1', name: 'Office Frame'}, network: {name: 'Home'},
+  stages: STAGES, probes: [{host: 'frame.local', detail: 'SSH identity changed'}],
+  error: {stage: 'identity', short: 'SSH identity changed', hostkey: {state: 'changed'},
+          message: 'This address answered with a different SSH identity from the one saved for Office Frame. '
+                 + 'If not, a different device may be answering here.'}};
+'''
+
+
+@unittest.skipUnless(shutil.which("node"), "Node runs the page's own functions")
+class LinkUI(unittest.TestCase):
+    def run_page(self, checks):
+        code = (between("function netLabel(s) {", "function renderConnDlg() {")
+                + between("async function useDevice(id) {", "async function devAction(")
+                + between("function deviceStatus(d) {", "function renderDevices() {"))
+        script = SETUP + code + "\n(async () => {\n" + checks + "\n})().catch(e => { console.error(e); process.exitCode = 1; });"
+        r = subprocess.run(["node", "-e", script], capture_output=True, text=True)
+        self.assertEqual(r.returncode, 0, r.stderr)
+
+    def test_the_pill_leads_with_the_reason(self):
+        # The header pill cut off "Offline: checking the headset's identity failed · retry in…":
+        # the reason comes first, in a few words, and only the countdown gets cut.
+        self.run_page(r'''
+  link.s = changed;
+  renderPill();
+  assert.match($('conn').innerHTML, /<b>Office Frame<\/b><span>SSH identity changed · retry in 12 s<\/span>/);
+  assert.equal($('conn').title, 'Office Frame is offline: SSH identity changed · retry in 12 s. Click for details.');
+  // Without a short reason: the stage that failed.
+  link.s = {...changed, retry_at: null, error: {stage: 'identity', message: 'Something.'}};
+  renderPill();
+  assert.match($('conn').innerHTML, /<span>Checking the headset&#39;s identity failed<\/span>/);
+''')
+
+    def test_the_banner_never_doubles_a_full_stop(self):
+        # "…may be answering here.." and "…(Developer Mode must be on).." (#87).
+        self.run_page(r'''
+  link.s = changed;
+  renderBanner();
+  assert.equal($('offDetail').textContent, 'Office Frame on Home. Tried frame.local: SSH identity changed. Trying again in 12 s.');
+  link.s = {...changed, probes: [
+    {host: 'frame.local', detail: "The Frame didn't accept this computer's SSH key. Set Up Connection pairs this computer with it again (Developer Mode must be on)."},
+    {host: '100.64.0.2', detail: 'No answer'}]};
+  renderBanner();
+  const text = $('offDetail').textContent;
+  assert.ok(!/\.\./.test(text), text);
+  assert.ok(text.includes("frame.local: the Frame didn't accept"), text);
+  assert.ok(text.includes('must be on); 100.64.0.2: no answer. Trying'), text);
+  assert.equal(asClause('SSH identity changed.'), 'SSH identity changed');  // abbreviations keep their case
+  assert.equal(asClause('Answered in 3 ms'), 'answered in 3 ms');
+''')
+
+    def test_the_devices_list_keeps_the_whole_message_within_reach(self):
+        self.run_page(r'''
+  link.s = changed;
+  const [dot, text, full] = deviceStatus({active: true, addresses: []});
+  assert.equal(dot, 'off');
+  assert.equal(text, 'SSH identity changed');
+  assert.equal(full, changed.error.message);  // the row's tooltip
+''')
+
+    def test_using_a_headset_selects_it(self):
+        # After "Use this headset" on Office Frame, the last headset's details stayed on show.
+        self.run_page(r'''
+  dv.list = [{id: 'd0', name: 'Living Room Frame', active: true}, {id: 'd1', name: 'Office Frame', active: false}];
+  dv.sel = 'd0';
+  await useDevice('d1');
+  assert.equal(dv.sel, 'd1');
+  assert.equal(dv.found, null);
+  assert.equal(dv.editing, null);
+  assert.equal(rendered, 1);
+  // Refused (an install is running): the selection stays, and the header shows the one in use.
+  dv.list = [{id: 'd0', name: 'Living Room Frame', active: true}, {id: 'd2', name: 'Spare', active: false}];
+  dv.sel = 'd0'; refuse = true;
+  await useDevice('d2');
+  assert.equal(dv.sel, 'd0');
+  assert.equal($('devSel').value, 'd0');
+''')
+
+
+class HeaderLayout(unittest.TestCase):
+    def test_the_devices_tab_explanation_wraps(self):
+        # It was cut to one line with no way to read the rest.
+        self.assertIn(".dev-problem .s { white-space: normal; }", PAGE)
+        self.assertEqual(len(re.findall(r'<div class="item dev-problem" role="alert">', PAGE)), 2)
+
+    def test_the_header_buttons_dont_give_way(self):
+        # Refresh overhung the right edge at 1280 px: the pill (and the headset picker) shrink, not the buttons.
+        header = between("<header>", "</header>")
+        end = header[header.index('<div class="hdr-end">'):]
+        for el in ('id="conn"', 'id="devSel"', 'id="battChip"', 'id="reportBtn"', 'id="refreshAll"'):
+            self.assertIn(el, end)
+        self.assertIn(".hdr-end > :not(.pill, .devsel) { flex: none; }", PAGE)
+        self.assertRegex(PAGE, r"\.hdr-end \{[^}]*min-width: 0;[^}]*margin-left: auto;")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_telemetry.py

@@ -138,7 +138,7 @@ class Gates(Base):
         tm.diagnostic("POST /api/android install", RuntimeError("boom"))
         tm.diagnostic("POST /api/comfort status", RuntimeError("ssh exited 255"))  # the command's own exit code?
         self.assertEqual([e["properties"]["error_category"] for e in self.queued()],
-                         ["frame_unreachable", "other", "other", "other"])
+                         ["frame_unreachable", "internal_error", "other", "other"])
 
     def test_page_events_are_checked(self):
         self.assertTrue(tm.page_event({"event": "tab_viewed", "properties": {"tab": "android", "extra": "x"}})["queued"])
@@ -164,6 +164,38 @@ class Lifecycle(Base):
         tm.frame_seen("20260922.1", "3.8")
         self.assertEqual(len(self.queued()), 1)
 
+    def test_frame_once_per_headset_with_saved_count(self):
+        tm.frame_seen("20260922.1", "3.8", "0123456789abcdef", 2)
+        tm.frame_seen("20260922.1", "3.8", "0123456789abcdef", 2)
+        tm.frame_seen("20260922.1", "3.8", "fedcba9876543210", 2)
+        props = [e["properties"] for e in self.queued()]
+        self.assertEqual([(p["headset_id"], p["headsets_saved"]) for p in props],
+                         [("0123456789abcdef", 2), ("fedcba9876543210", 2)])
+
+    def test_frame_many_headsets_are_not_sent_twice(self):
+        ids = ["%016x" % n for n in range(25)]
+        for _ in range(2):
+            for h in ids:
+                tm.frame_seen("20260922.1", "3.8", h, 25)
+        self.assertEqual(len(self.queued()), 25)
+
+    def test_frame_headset_id_must_be_the_hash(self):
+        tm.frame_seen("20260922.1", "3.8", "/etc/machine-id contents", None)
+        p = self.queued()[0]["properties"]
+        self.assertNotIn("headset_id", p)
+        self.assertNotIn("headsets_saved", p)
+
+
+class HeadsetId(unittest.TestCase):
+    def test_hash_of_machine_id_and_none_without_one(self):
+        import hashlib
+        import frame_status
+        with mock.patch.object(frame_status, "read", return_value="abc123"):
+            self.assertEqual(frame_status.headset_id(),
+                             hashlib.sha256(b"frame-control-headset:abc123").hexdigest()[:16])
+        with mock.patch.object(frame_status, "read", return_value=None):
+            self.assertIsNone(frame_status.headset_id())
+
 
 class Sending(Base):
     def serve(self, status=200):
@@ -263,6 +295,72 @@ class Scrub(unittest.TestCase):
         self.assertEqual(tm.categorize(FileNotFoundError(2, "No such file or directory", "ssh"))[0], "tool_missing")
         self.assertEqual(tm.categorize("[WinError 2] The system cannot find the file specified")[0], "tool_missing")
 
+    def test_apk_install_failures_that_were_other(self):
+        """What frame_android.install can raise, worded as it is, that 0.4.x sent as 'other'
+        (or as frame_unreachable, when it was the Steam client or a slow copy)."""
+        tail = 'Traceback (most recent call last):\n  File "<stdin>", line 19, in target_ws\n'
+        cases = {
+            "steam_not_running": [
+                "SharedJSContext not found: is the Steam client running?",
+                tail + "urllib.error.URLError: <urlopen error [Errno 111] Connection refused>",
+                tail + "ConnectionRefusedError: [Errno 111] Connection refused",
+            ],
+            "steam_not_responding": [
+                "websocket handshake failed",
+                tail + "TimeoutError: timed out",
+                tail + "socket.timeout: timed out",
+                tail + '  File "<stdin>", line 45, in _read\nEOFError',
+            ],
+            "steam_js_error": ['JS error: {"exceptionId": 1, "text": "Uncaught", "exception": '
+                               '{"description": "Error: Steam artwork API unavailable; installation is incomplete"}}'],
+            "artwork_failed": [
+                "Steam artwork renderer did not produce every slot",
+                tail + "ValueError: incomplete artwork render",
+                tail + "ValueError: hero render is 0 bytes of png",
+                "could not prepare artwork: unknown artwork slot",
+                tail + "ValueError: all five Steam artwork slots are required",
+            ],
+            "copy_timeout": ["copying Big Game (timed out).apk to the Frame timed out",
+                             "copying launch.sh to the Frame timed out"],
+            "copy_failed": [
+                "copying game.apk to the Frame failed: rsync: connection unexpectedly closed (0 bytes received "
+                "so far) [sender]\nrsync error: error in rsync protocol data stream (code 12) at io.c(231)",
+                "copying game.apk to the Frame failed: ",
+            ],
+            "apk_unreadable": [
+                "x.apk: not a readable APK: [Errno 2] No such file or directory: 'x.apk'",
+                "x.apk: could not read the APK manifest: Error -3 while decompressing data: invalid block type",
+                "x.apk: the APK manifest has no package name",
+                "x.apk: lib/arm64-v8a/libx.so in the APK uses a compression Android does not",
+                "x.apk: AndroidManifest.xml in the APK is 40 MB, more than a real one (16 MB)",
+            ],
+        }
+        for category, messages in cases.items():
+            for message in messages:
+                self.assertEqual(tm.categorize(message)[0], category, message)
+        # Still where they were: ssh's own failures, and a full disk during a copy.
+        self.assertEqual(tm.categorize("timed out talking to frame")[0], "frame_unreachable")
+        self.assertEqual(tm.categorize("copying a.apk to the Frame failed: ssh: connect to host 10.0.0.2 "
+                                       "port 22: Connection refused")[0], "frame_unreachable")
+        self.assertEqual(tm.categorize("copying a.apk to the Frame failed: rsync: write failed on "
+                                       "\"app.apk.part\": No space left on device (28)")[0], "frame_disk_full")
+        self.assertEqual(tm.categorize("download failed: timed out")[0], "download_failed")
+
+    def test_python_errors_are_internal_whatever_they_say(self):
+        import json as _json
+        for error in (KeyError("paths"), TypeError("'NoneType' object is not subscriptable"),
+                      AttributeError("x"), _json.JSONDecodeError("Expecting value", "", 0),
+                      UnicodeDecodeError("utf-8", b"\xff", 0, 1, "invalid start byte"),
+                      KeyError("timed out")):
+            self.assertEqual(tm.categorize(error)[0], "internal_error", repr(error))
+        self.assertEqual(tm.categorize("KeyError: 'paths'")[0], "other")  # only the exception itself
+        self.assertEqual(tm.categorize(OSError(28, "No space left on device"))[0], "frame_disk_full")
+
+    def test_privacy_doc_lists_every_category(self):
+        doc = (ROOT / "docs" / "privacy.md").read_text()
+        for name in {n for n, _ in tm.CATEGORIES} | {"internal_error", "other"}:
+            self.assertIn(f"`{name}`", doc)
+
 
 class Compat(Base):
     def test_reports_are_shared_only_after_opting_in_without_file_names(self):
@@ -376,14 +474,17 @@ class LinkFailureProvenance(Base):
         c.close()
         return result
 
-    def test_routes_that_rewrap_a_link_failure_keep_its_mark(self):
-        # Review round 4: /api/android and /api/titles re-raise frame_android's FrameError as a
-        # Failure; after the status poll's first connection failure, theirs are held back too,
-        # also past the 10-minute window.
+    def serve(self):
         self.httpd = self.server.ThreadingHTTPServer(("127.0.0.1", 0), self.server.Handler)
         threading.Thread(target=self.httpd.serve_forever, daemon=True).start()
         self.addCleanup(self.httpd.server_close)
         self.addCleanup(self.httpd.shutdown)
+
+    def test_routes_that_rewrap_a_link_failure_keep_its_mark(self):
+        # Review round 4: /api/android and /api/titles re-raise frame_android's FrameError as a
+        # Failure; after the first connection failure, theirs are held back too, also past the
+        # 10-minute window.
+        self.serve()
         down = subprocess.CompletedProcess(["ssh"], 255, "", "ssh: connect to host 10.0.0.5 port 22: Connection timed out\r\n")
         meta = {"label": "Test App", "game_id": 5, "shortcut": None}
         with mock.patch.object(frame_host, "run_ssh", return_value=down), \
@@ -399,7 +500,34 @@ class LinkFailureProvenance(Base):
                 self.assertEqual(self.request("/api/android", {"action": "launch", "package": "org.test"})[0], 502)
         sent = [(e["properties"]["where"], e["properties"]["error_category"]) for e in self.queued()
                 if e["event"] == "$exception"]
-        self.assertEqual(sent, [("POST /api/comfort status", "frame_unreachable"), ("POST /api/android launch", "other")])
+        # The comfort poll's own failure isn't one (see the next test).
+        self.assertEqual(sent, [("POST /api/android launch", "frame_unreachable"), ("POST /api/android launch", "other")])
+
+    def test_the_comfort_poll_meeting_an_unreachable_frame_is_not_sent(self):
+        # 0.4.2: most reports were this poll (every 5 s) finding the headset asleep, away or
+        # refusing ssh. Like GET /api/status, it isn't reported; a session started or cancelled
+        # by hand still is, once, and a real failure of the poll is too.
+        self.serve()
+        for stderr in ("ssh: connect to host 10.0.0.5 port 22: Connection timed out\r\n",
+                       "banner exchange: Connection to UNKNOWN port -1: Connection refused\r\n",
+                       "ssh: Could not resolve hostname frame: nodename nor servname provided\r\n",
+                       "@    WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!     @\r\nHost key verification failed.\r\n",
+                       "No ED25519 host key is known for frame-control-f7b4e9ac and you have requested strict "
+                       "checking.\r\nHost key verification failed.\r\n",
+                       "steamos@192.168.1.9: Permission denied (publickey,password).\r\n"):
+            down = subprocess.CompletedProcess(["ssh"], 255, "", stderr)
+            with mock.patch.object(frame_host, "run_ssh", return_value=down):
+                self.assertGreaterEqual(self.request("/api/comfort", {"action": "status"})[0], 500, stderr)
+        self.assertEqual(self.queued(), [])
+        down = subprocess.CompletedProcess(["ssh"], 255, "", "ssh: connect to host 10.0.0.5 port 22: Host is down\r\n")
+        with mock.patch.object(frame_host, "run_ssh", return_value=down):
+            self.request("/api/comfort", {"action": "cancel"})
+        broken = subprocess.CompletedProcess(["ssh"], 1, "", "Traceback (most recent call last):\nOSError: boom\n")
+        with mock.patch.object(frame_host, "run_ssh", return_value=broken):
+            self.assertEqual(self.request("/api/comfort", {"action": "status"})[0], 502)
+        sent = [(e["properties"]["where"], e["properties"]["error_category"]) for e in self.queued()
+                if e["event"] == "$exception"]
+        self.assertEqual(sent, [("POST /api/comfort cancel", "frame_unreachable"), ("POST /api/comfort status", "other")])
 
     def test_download_failures_after_a_poll_failure_are_still_sent(self):
         # Review round 3: these file names, in the real checksum message through the web-link worker,

ui/frame_android.py

@@ -61,13 +61,20 @@ def ssh(cmd, input=None, timeout=120):
     try:
         # No inherited stdin (see server.ssh): Windows' ssh.exe would wait on it.
         feed = {'input': input} if input is not None else {'stdin': subprocess.DEVNULL}
+        # The Frame writes UTF-8, whatever this computer's code page (Windows would decode it as
+        # cp1252). errors='replace': stray bytes (a file name) mustn't turn a failed command, or
+        # one that worked, into a UnicodeDecodeError.
+        text = ({'text': True, 'encoding': 'utf-8', 'errors': 'replace'}
+                if isinstance(input, str) or input is None else {})
         p = frame_host.run_ssh(['ssh', *SSH_OPTS, FRAME, cmd], capture_output=True, **feed,
-                               timeout=timeout, text=isinstance(input, str) or input is None)
+                               timeout=timeout, **text)
     except subprocess.TimeoutExpired:
         raise frame_host.link_failure(FrameError(f'timed out talking to {FRAME}'))
     if p.returncode != 0:
-        error = FrameError((p.stderr or p.stdout or f'ssh exited {p.returncode}').strip()[-600:])
-        stderr = p.stderr if isinstance(p.stderr, str) else (p.stderr or b'').decode(errors='replace')
+        # Bytes input gives bytes output: decode it, so the message isn't "b'...'".
+        stderr, stdout = (v if isinstance(v, str) else (v or b'').decode(errors='replace')
+                          for v in (p.stderr, p.stdout))
+        error = FrameError((stderr or stdout or f'ssh exited {p.returncode}').strip()[-600:])
         raise frame_host.link_failure(error) if frame_host.ssh_link_failed(p.returncode, stderr) else error
     return p.stdout
 
@@ -136,11 +143,19 @@ def check_installable(info):
 
 
 _install_lock = threading.Lock()  # installs are rare; one at a time avoids every race
+# The slowest transfer a copy is allowed before it times out, in bytes a second (about 4 Mbit/s,
+# a slow or busy Wi-Fi link). A fixed 10 minutes failed any APK over about 300 MB at that rate.
+COPY_MIN_RATE = 512 * 1024
 
 
 def _copy(src, dest, executable=False, timeout=600):
-    """Copy a local file to the Frame: rsync where installed (not on Windows), else scp."""
+    """Copy a local file to the Frame: rsync where installed (not on Windows), else scp.
+    timeout is the least time allowed; a large file gets longer (COPY_MIN_RATE)."""
     name = os.path.basename(src)
+    try:
+        timeout = max(timeout, os.path.getsize(src) / COPY_MIN_RATE)
+    except OSError:
+        pass  # rsync or scp says what's wrong with the file
     rsync = None if frame_host.WINDOWS else shutil.which('rsync')  # see server.push_file
     if rsync:
         cmd = ['rsync', '-a', *(['--chmod=u+x'] if executable else []),
@@ -148,7 +163,8 @@ def _copy(src, dest, executable=False, timeout=600):
     else:
         cmd = ['scp', *SSH_OPTS, src, f'{FRAME}:{dest}']
     try:
-        frame_host.run_ssh(cmd, check=True, capture_output=True, stdin=subprocess.DEVNULL, text=True, timeout=timeout)
+        frame_host.run_ssh(cmd, check=True, capture_output=True, stdin=subprocess.DEVNULL, text=True,
+                           errors='replace', timeout=timeout)
     except subprocess.TimeoutExpired:
         raise FrameError(f'copying {name} to the Frame timed out')
     except subprocess.CalledProcessError as e:
@@ -279,6 +295,13 @@ def _install(apk_path, info, pkg, flatscreen, name, source, artwork=None):
                 ssh(f'rm -rf {d}', timeout=30)
             except FrameError:
                 pass
+        elif not ok:
+            # An update that failed: the app still works, but a partial copy of the new APK
+            # (scp leaves one; it can be gigabytes) mustn't stay beside it.
+            try:
+                ssh(f'rm -f {d}/app.apk.part', timeout=30)
+            except FrameError:
+                pass
 
 
 

ui/frame_apk.py

@@ -7,6 +7,7 @@ the resource table in resources.arsc.
 """
 import struct
 import zipfile
+import zlib
 
 # android: attribute resource ids; names can be stripped by shrinkers, ids can't.
 ATTR = {0x01010001: 'label', 0x01010002: 'icon', 0x01010003: 'name',
@@ -221,7 +222,9 @@ def apk_info(path):
             res = Resources(_read(z, 'resources.arsc', MAX_ARSC) if 'resources.arsc' in names else b'')
             from frame_apk_vr import detection
             vr_info = detection(manifest_data, names)
-        except (struct.error, IndexError, zipfile.BadZipFile) as e:
+        # A damaged download: corrupt deflate data (zlib.error), a member cut short (EOFError),
+        # undecodable text (UnicodeDecodeError) or an encrypted entry (RuntimeError).
+        except (struct.error, IndexError, zipfile.BadZipFile, zlib.error, EOFError, ValueError, RuntimeError) as e:
             raise ApkError(f'could not read the APK manifest: {e}')
         tags = {}
         for tag, attrs in elements:

ui/frame_link.py

@@ -238,6 +238,13 @@ def scrub_failure(text, hosts=()):
     return frame_telemetry.scrub(hide_operands(hide_hosts(str(text or ""), hosts)), 600)
 
 
+def failure_detail(error):
+    """What an address that answered but failed SSH shows: the short reason when there is one
+    (the whole message is the banner's headline already), else the message."""
+    error = error or {}
+    return error.get("short") or error.get("message") or "SSH failed"
+
+
 def failure_category(message, raw=""):
     """The telemetry error category (frame_unreachable, frame_auth, ...) for a failure: a fixed
     word, never its text."""
@@ -810,7 +817,9 @@ class Link:
         return "Connecting"
 
     def fail(self, sid, message, raw="", **more):
-        """A stage failed. `more`: what the page can offer to fix it (hostkey, auth)."""
+        """A stage failed. `more`: what the page can offer to fix it (hostkey, auth), and `short`:
+        the reason in a few words, for the places a whole sentence doesn't fit (the header pill,
+        the address that failed)."""
         self.stage(sid, "failed", message)
         with self.cond:
             self.state["error"] = dict({"stage": sid, "message": message, "raw": raw}, **more)
@@ -896,7 +905,7 @@ class Link:
                         self.learn(device, a["host"], net, None)
                         return True
                     with self.cond:
-                        why_not = (self.state["error"] or {}).get("message") or "SSH failed"
+                        why_not = failure_detail(self.state["error"])
                     self.probe_update(i, state="sshfailed", detail=why_not)
                     if outcome != "next":
                         return False
@@ -946,7 +955,7 @@ class Link:
                 self.learn(device, a["host"], net, results[pick]["rtt_ms"])
                 return True
             with self.cond:
-                why_not = (self.state["error"] or {}).get("message") or "SSH failed"
+                why_not = failure_detail(self.state["error"])
             self.probe_update(pick, state="sshfailed", detail=why_not)
             if outcome != "next":
                 return False
@@ -1269,6 +1278,7 @@ class Link:
                            "If SteamOS was reinstalled or reset, or this is a new headset, check its new key and trust "
                            "it on the Devices tab. If not, a different device may be answering here.")
             self.fail("identity", message, text,
+                      short="Saved SSH identity missing" if lost else "SSH identity changed",
                       hostkey={"state": "missing" if lost else "changed", "type": shown[0] if shown else None,
                                "fingerprint": shown[1] if shown else None, "host": host})
             return "next"
@@ -1279,7 +1289,8 @@ class Link:
         if re.search(r"Permission denied", text) and not forward:
             # Not something a retry fixes: the headset needs this computer's key (Set Up Connection).
             self.fail("login", self.explain(text) or "The headset didn't accept this computer's key. Run Set Up "
-                      "Connection: it pairs this computer with the headset again.", text, auth="denied")
+                      "Connection: it pairs this computer with the headset again.", text, auth="denied",
+                      short="This computer's key refused")
             return "stop"
         if step == "connected":
             self.fail("login", "Logged in, but the shared SSH connection didn't start.", text)

ui/frame_status.py

@@ -6,6 +6,7 @@ error. Sensor paths verified on SteamOS 0.3.0; OpenVR timing on 0.4.1
 """
 import ctypes as C
 import glob
+import hashlib
 import math
 import json
 import os
@@ -121,7 +122,16 @@ def os_release():
         k, _, v = line.partition("=")
         out[k] = v.strip('"')
     return {"version": out.get("VERSION_ID"), "build": out.get("BUILD_ID"),
-            "variant": out.get("VARIANT_ID")}
+            "variant": out.get("VARIANT_ID"), "headset": headset_id()}
+
+
+def headset_id():
+    """A one-way hash of the headset's /etc/machine-id: the same for this headset on every
+    computer, and the machine-id itself never leaves the headset. None without one."""
+    mid = read("/etc/machine-id")
+    if not mid:
+        return None
+    return hashlib.sha256(("frame-control-headset:" + mid).encode()).hexdigest()[:16]
 
 
 def volume():

ui/frame_telemetry.py

@@ -50,6 +50,7 @@ SETTINGS = STATE / 'settings.json'
 OUTBOX = STATE / 'outbox.jsonl'
 SENT = STATE / 'sent.jsonl'
 SENT_KEEP = 200
+FRAMES_SEEN_KEEP = 100  # build/headset pairs already reported, enough for several headsets' updates
 OUTBOX_MAX = 2000  # events kept while offline; the oldest go first
 FLUSH_EVERY = 60
 REPEAT_WINDOW = 600  # the same diagnostic error is sent at most once in this many seconds
@@ -266,9 +267,23 @@ CATEGORIES = [
     ('layer_missing', re.compile(r'OpenXR compatibility layer')),
     ('apk_repack_failed', re.compile(r'could not prepare the APK for the Frame')),
     ('tool_missing', re.compile(r"\[WinError 2\]|No such file or directory: '(?:ssh|scp|rsync|adb)")),
-    ('apk_unreadable', re.compile(r'(?i)not a zip|bad apk|AndroidManifest|ApkError|unexpected package name')),
+    ('apk_unreadable', re.compile(r'(?i)not a zip|bad apk|AndroidManifest|ApkError|unexpected package name|'
+                                  r'not a readable APK|not an APK|APK manifest|in the APK (?:uses a compression|is )')),
     ('cant_run_on_frame', re.compile(r"can't run on the Frame")),
     ('steam_shortcut', re.compile(r'(?i)steam did not return a shortcut|shortcut list|no Steam shortcut')),
+    # The Frame-side scripts that talk to the Steam client's UI (frame/android/steam_shortcuts.py), which
+    # fail with their own words or a Python traceback, its last lines kept. Before frame_unreachable:
+    # "Connection refused" and "timed out" here are Steam's debug port on the Frame, not ssh.
+    ('steam_not_running', re.compile(r'SharedJSContext not found|'
+                                     r'URLError: <urlopen error \[Errno 111\] Connection refused>|'
+                                     r'ConnectionRefusedError: \[Errno 111\]')),
+    ('steam_not_responding', re.compile(r'websocket handshake failed|(?:socket\.timeout|TimeoutError): timed out|'
+                                        r'(?:\A|\n)EOFError\Z')),
+    ('steam_js_error', re.compile(r'JS error: ')),
+    ('artwork_failed', re.compile(r'incomplete artwork render|artwork renderer did not produce|'
+                                  r'render is \d+ bytes of|could not prepare artwork|all five Steam artwork slots')),
+    # Copying the APK (or its launcher) to the Frame: rsync or scp ran out of time.
+    ('copy_timeout', re.compile(r'to the Frame timed out\Z')),
     ('frame_not_set_up', re.compile(r'(?i)Could not resolve hostname|no "?frame"? (?:SSH )?alias')),
     ('frame_auth', re.compile(r'(?i)Permission denied|Host key verification failed|'
                               r'REMOTE HOST IDENTIFICATION HAS CHANGED|host key is known for')),
@@ -280,6 +295,8 @@ CATEGORIES = [
                                      r'Network is unreachable|Host is down|asleep|kex_exchange|banner exchange|'
                                      r'ssh: connect to host |mux_client_|client_loop: ')),
     ('frame_disk_full', re.compile(r'(?i)No space left|disk full|ENOSPC')),
+    # rsync or scp failed for another reason ("connection unexpectedly closed", or no words at all).
+    ('copy_failed', re.compile(r'\Acopying .+ to the Frame failed: ')),
     ('download_failed', re.compile(r'(?i)HTTP (?:Error )?\d{3}|URLError|download|certificate verify failed')),
     ('flatpak', re.compile(r'(?i)flatpak|flathub')),
     ('cancelled', re.compile(r'(?i)cancel')),
@@ -287,8 +304,16 @@ CATEGORIES = [
 ]
 
 
+# Python's own errors for a mistake in Frame Control's code, whatever their message says.
+BUG_ERRORS = (KeyError, IndexError, TypeError, AttributeError, NameError, AssertionError,
+              ZeroDivisionError, RecursionError, UnicodeError, json.JSONDecodeError)
+
+
 def categorize(message):
-    """(category, detail): a fixed category name, plus an Android installer code when there is one."""
+    """(category, detail): a fixed category name, plus an Android installer code when there is one.
+    message may be the exception itself: one of BUG_ERRORS is internal_error."""
+    if isinstance(message, BUG_ERRORS):
+        return 'internal_error', None
     text = str(message or '')
     for name, pattern in CATEGORIES:
         m = pattern.search(text)
@@ -366,16 +391,24 @@ def app_started():
         _save(s)
 
 
-def frame_seen(build, version):
-    """The Frame's SteamOS build, once per build (public build numbers)."""
-    key = f'{build}/{version}'
+def frame_seen(build, version, headset=None, saved=None):
+    """The Frame's SteamOS build (public build numbers), once per build for each headset.
+    headset: the anonymous hash from frame_status.headset_id; saved: how many headsets
+    this copy of the app has set up."""
+    headset = headset if isinstance(headset, str) and re.fullmatch(r'[0-9a-f]{16}', headset) else None
+    key = f'{build}/{version}' + (f'/{headset}' if headset else '')
     with _lock:
         s = settings()
         if not build or key in s['frames_seen']:
             return
-        s['frames_seen'] = (s['frames_seen'] + [key])[-20:]
+        s['frames_seen'] = (s['frames_seen'] + [key])[-FRAMES_SEEN_KEEP:]
         _save(s)
-    capture('frame_connected', {'steamos_build': str(build)[:40], 'steamos_version': str(version or '')[:40]})
+    p = {'steamos_build': str(build)[:40], 'steamos_version': str(version or '')[:40]}
+    if headset:
+        p['headset_id'] = headset
+    if isinstance(saved, int) and saved >= 0:
+        p['headsets_saved'] = saved
+    capture('frame_connected', p)
 
 
 def install_finished(kind, ok, seconds=None, error=None, diagnose=True, **props):

ui/index.html

@@ -51,6 +51,10 @@
   nav a:hover { color: var(--bright); }
   nav a.on { color: var(--blue); border-bottom-color: var(--blue); }
   .spacer { flex: 1; }
+  /* The connection pill and the buttons after it, pushed to the right: closer together than the tabs, and the
+     pill (not the buttons) gives way when the window is narrow, so nothing is pushed past the right edge. */
+  .hdr-end { display: flex; align-items: center; gap: 10px; min-width: 0; flex: 0 1 auto; margin-left: auto; }
+  .hdr-end > :not(.pill, .devsel) { flex: none; }
   .chip { display: inline-flex; align-items: center; gap: 8px; height: 32px; padding: 0 12px; border-radius: 3px;
           background: var(--btn); color: var(--text); font-size: 13px; white-space: nowrap; }
   .dot { width: 8px; height: 8px; border-radius: 50%; background: var(--dim); flex: none; }
@@ -60,6 +64,7 @@
   main { max-width: 1480px; margin: 0 auto; padding: 22px calc(28px + env(safe-area-inset-right)) 30px calc(28px + env(safe-area-inset-left)); display: flex; flex-direction: column; gap: 26px; }
   .grid-top { display: grid; grid-template-columns: minmax(0, 1.7fr) minmax(330px, 1fr); gap: 22px; }
   .grid-3 { display: grid; grid-template-columns: repeat(auto-fit, minmax(300px, 1fr)); gap: 22px; }
+  @media (max-width: 1366px) { header { gap: 20px; } nav a { padding: 0 11px; } }
   @media (max-width: 1150px) { .grid-top, .grid-3 { grid-template-columns: 1fr; } header { gap: 16px; padding: env(safe-area-inset-top) calc(18px + env(safe-area-inset-right)) 0 calc(18px + env(safe-area-inset-left)); } nav a { padding: 0 10px; } }
   @media (max-width: 820px) { .brand b { display: none; } }
   nav svg { display: none; }
@@ -74,7 +79,7 @@
   .page { display: flex; flex-direction: column; gap: 26px; }
   .page:not(.on) { display: none; }
   nav kbd { font: inherit; font-size: 10.5px; color: var(--dim); margin-left: 7px; letter-spacing: 0; }
-  @media (max-width: 1150px) { nav kbd { display: none; } }
+  @media (max-width: 1366px) { nav kbd { display: none; } }  /* the tabs' titles still name the key */
 
   /* ---- offline banner ---- */
   .banner { display: flex; align-items: center; gap: 14px; padding: 12px 16px; border-radius: 4px;
@@ -87,14 +92,14 @@
   .wait::before { content: ""; width: 7px; height: 7px; border-radius: 50%; background: var(--dim); flex: none; }
 
   /* ---- connection pill, its details dialog, and the Devices tab (frame_link.py) ---- */
-  .pill { height: 40px; padding: 0 12px; gap: 9px; max-width: 420px; min-width: 190px; flex: 0 1 auto; background: var(--btn); }
+  .pill { height: 40px; padding: 0 12px; gap: 9px; max-width: 420px; min-width: 150px; flex: 0 1 auto; background: var(--btn); }
   .pill .dot { flex: none; }
   .pill .dot.wait { background: var(--warn); box-shadow: 0 0 6px rgba(217,162,58,.7); animation: pulse 1s ease-in-out infinite; }
   @keyframes pulse { 50% { opacity: .35; } }
   .pill .pt { display: flex; flex-direction: column; align-items: flex-start; min-width: 0; line-height: 1.2; text-align: left; }
   .pill .pt b { font-weight: 500; font-size: 13px; color: var(--bright); max-width: 100%; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
   .pill .pt span { font-size: 11.5px; color: var(--muted); max-width: 100%; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
-  .devsel { background: var(--btn); color: var(--text); border: 0; border-radius: 3px; height: 40px; padding: 0 8px; font: inherit; font-size: 13px; max-width: 160px; }
+  .devsel { background: var(--btn); color: var(--text); border: 0; border-radius: 3px; height: 40px; padding: 0 8px; font: inherit; font-size: 13px; max-width: 160px; min-width: 90px; }
   .dlg { background: #1e2329; color: var(--text); border: 1px solid rgba(255,255,255,.1); border-radius: 4px;
          padding: 22px; width: min(640px, 94vw); box-shadow: 0 20px 60px rgba(0,0,0,.6); }
   .dlg::backdrop { background: rgba(0,0,0,.55); }
@@ -132,7 +137,7 @@
   .dev-panel select, .dlg select { background: rgba(0,0,0,.28); color: var(--text); border: 1px solid transparent; border-radius: 3px;
                                    padding: 8px 8px; font: inherit; font-size: 13px; }
   .addr .t { overflow-wrap: anywhere; }
-  .addr .s { white-space: normal; }
+  .addr .s, .dev-problem .s { white-space: normal; }
   .addr-edit { display: grid; grid-template-columns: 2fr 1fr 1.4fr auto auto; gap: 8px; align-items: center; width: 100%; }
   .addr-add { display: grid; grid-template-columns: 2fr 1fr 1.4fr auto; gap: 8px; margin-top: 12px; }
   @media (max-width: 700px) { .dev-form, .addr-edit, .addr-add { grid-template-columns: 1fr; } }
@@ -655,7 +660,7 @@
     <a href="#tools" title="Files, clipboard, Linux apps, remote and power (4)"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M14.7 6.3a4 4 0 0 0-5.2 5.2L3.5 17.5a2.1 2.1 0 0 0 3 3l6-6a4 4 0 0 0 5.2-5.2l-2.5 2.5-2.5-2.5z"/></svg>Tools<kbd>4</kbd></a>
     <a href="#devices" class="desk-only" title="Headsets, their addresses and the connection (5)"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M5 12.5a10 10 0 0 1 14 0M8.5 16a5 5 0 0 1 7 0"/><circle cx="12" cy="19.5" r="1"/><path d="M2 9a15 15 0 0 1 20 0"/></svg>Devices<kbd>5</kbd></a>
   </nav>
-  <div class="spacer"></div>
+  <div class="hdr-end">
   <button class="pill" id="conn" role="status" aria-live="polite" title="Connection details"><span class="dot wait"></span><span class="pt"><b>Connecting…</b><span></span></span></button>
   <select class="devsel desk-only" id="devSel" aria-label="Headset" title="Switch headset" hidden></select>
   <span class="chip" id="battChip" title="Battery">—</span>
@@ -665,6 +670,7 @@
   <button id="refreshAll" title="Refresh (R)" aria-label="Refresh">
     <svg width="15" height="15" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2.4" aria-hidden="true"><path d="M21 12a9 9 0 1 1-3-6.7"/><path d="M21 3v6h-6"/></svg>
   </button>
+  </div>
 </header>
 
 <main>
@@ -677,7 +683,8 @@
   </div>
   <div class="notice" id="privacyNotice" hidden>
     <div class="grow">Frame Control sends anonymous usage statistics: that it was installed and opened, its version,
-      your operating system, which tabs you use, and whether installs on the Frame worked. Never file names, paths,
+      your operating system, which tabs you use, whether installs on the Frame worked, and how many headsets you use
+      (a one-way hash of each headset's machine ID, the same on every computer, so they're counted once). Never file names, paths,
       addresses or anything you've typed, and it isn't linked to you. You can also share whether Android apps
       worked and the details of errors, which helps fix problems faster.</div>
     <button class="small" id="noticeSettings">Privacy settings</button>
@@ -1142,7 +1149,8 @@
       <div class="shelf-head"><h2>Privacy &amp; updates</h2><span class="spacer"></span><span class="sub" id="appVersion"></span></div>
       <label class="popt"><input type="checkbox" id="tUsage"><b>Anonymous usage statistics</b>
         <span class="sub">Installs, opens, version, operating system, tabs used, and whether installs on the Frame
-          worked (with an error category, never the message). F-Droid package names only.</span></label>
+          worked (with an error category, never the message). F-Droid package names only. Each headset's SteamOS
+          build, a one-way hash of its machine ID (to count headsets, not just installs), and how many you've set up.</span></label>
       <label class="popt"><input type="checkbox" id="tCompat"><b>Share compatibility results</b>
         <span class="sub">Your APK reports and tests (package, version, result, your notes) go to the shared
           compatibility database, so the verdicts get better for everyone.</span></label>
@@ -1587,7 +1595,7 @@ $("vrUtilities").onclick = async e => {
   await loadUtilities();
 };
 
-// The Frame owns the clock. Poll independently of status and the selected tab.
+// The Frame owns the clock. Poll independently of the selected tab, from the first status check on.
 let comfortBusy = false, comfortSeen = new Set(), comfortSession = null, comfortHydrated = false;
 async function localNotification(message, request = false) {
   try {
@@ -1646,7 +1654,12 @@ function renderComfort(s) {
   if ($("comfortEvents").textContent !== history) $("comfortEvents").textContent = history;
 }
 async function pollComfort() {
-  if (!comfortBusy) {
+  // Not while the Frame can't be reached: each poll would be another ssh attempt waiting to
+  // time out, and the banner already says why. Status checks bring it back (setOnline).
+  if (online === false) {
+    const message = "Session status appears when the Frame is connected.";
+    if ($("comfortStatus").textContent !== message) $("comfortStatus").textContent = message;
+  } else if (!comfortBusy) {
     comfortBusy = true;
     try { renderComfort(await api("/api/comfort", {action: "status"})); }
     catch (e) {
@@ -1680,7 +1693,6 @@ $("castBtn").onclick = () => {
   // Fullscreen is a direct user gesture; iPhone falls back to its inline viewer.
   $("viewer").requestFullscreen?.().catch(() => {});
 };
-pollComfort();
 
 // ---- battery ----
 function battery(b, power) {
@@ -4495,7 +4507,7 @@ if (window.frameApp && window.frameApp.onInstallLink) {
 }
 
 setView("headset");
-refresh().then(() => loadShots());  // after status, so app names resolve
+refresh().then(() => { loadShots(); pollComfort(); });  // after status, so app names resolve and comfort knows if the Frame answers
 document.addEventListener("visibilitychange", () => { if (!document.hidden && online === false) refresh(); });
 </script>
 <script>
@@ -4524,6 +4536,20 @@ function netLabel(s) {
 }
 function viaText(v) { return v ? v.host + (v.ip && v.ip !== v.host ? ` (${v.ip})` : "") : ""; }
 
+// Why the last attempt failed, in a few words: the server's short reason, else the stage that failed.
+function failureReason(s) {
+  const err = s.error || {};
+  if (err.short) return err.short;
+  const st = (s.stages || []).find(x => x.id === err.stage);
+  return st ? `${st.label} failed` : "Not connected";
+}
+// A sentence or a label as a clause inside another sentence: no closing full stop (the sentence
+// adds its own) and a lower-case first letter, unless the first word is an abbreviation ("SSH").
+function asClause(text) {
+  const t = String(text || "").trim().replace(/[.\s]+$/, "");
+  return /^[A-Z][a-z]/.test(t) ? t[0].toLowerCase() + t.slice(1) : t;
+}
+
 // The pill in the header: always there, and says what's happening right now.
 function renderPill() {
   const s = link.s;
@@ -4535,9 +4561,8 @@ function renderPill() {
     line = `${netLabel(s)} → ${viaText(s.via)}` + (s.via && s.via.rtt_ms != null ? ` · ${s.via.rtt_ms} ms` : "");
   } else if (s.phase === "failed") {
     dot = "off";
-    const left = s.retry_at ? Math.max(0, Math.ceil(s.retry_at - serverNow())) : null;
-    const st = (s.stages || []).find(x => x.id === (s.error || {}).stage);
-    line = `Offline: ${st ? st.label.toLowerCase() : "not connected"} failed` + (left != null ? ` · retry in ${left} s` : "");
+    // The reason first, in a few words, so a narrow pill cuts the countdown rather than the reason.
+    line = failureReason(s) + (s.retry_at ? ` · retry in ${Math.max(0, Math.ceil(s.retry_at - serverNow()))} s` : "");
   } else {
     const st = activeStage(s);
     const probing = (s.probes || []).find(p => p.state === "trying" || p.state === "resolving");
@@ -4545,7 +4570,7 @@ function renderPill() {
               : "Connecting…";
   }
   $("conn").innerHTML = `<span class="dot ${dot}"></span><span class="pt"><b>${esc(dev)}</b><span>${esc(line)}</span></span>`;
-  $("conn").title = `${dev}: ${line}. Click for details.`;
+  $("conn").title = `${dev}${s.phase === "failed" ? " is offline" : ""}: ${line}. Click for details.`;
   return true;
 }
 window.connPill = renderPill;
@@ -4562,13 +4587,13 @@ function renderBanner() {
   if (s.phase === "failed" && s.error) {
     $("offline").hidden = false;
     $("offMsg").textContent = s.error.message;
-    const tried = (s.probes || []).map(p => `${p.host}: ${p.detail.toLowerCase()}`).join("; ");
+    const tried = (s.probes || []).map(p => `${p.host}: ${asClause(p.detail)}`).join("; ");
     const left = s.retry_at ? Math.max(0, Math.ceil(s.retry_at - serverNow())) : null;
     $("offDetail").textContent = `${s.device ? s.device.name : "Frame"} on ${netLabel(s)}. ${tried ? "Tried " + tried + ". " : ""}` +
       (left != null ? `Trying again in ${left} s.` : "");
   } else if (s.phase === "connecting" && online === false) {
     const st = activeStage(s);
-    $("offDetail").textContent = `Trying again: ${st ? st.label.toLowerCase() + (st.detail ? " (" + st.detail + ")" : "") : "connecting"}…`;
+    $("offDetail").textContent = `Trying again: ${st ? asClause(st.label) + (st.detail ? " (" + st.detail + ")" : "") : "connecting"}…`;
   }
 }
 window.connBanner = renderBanner;
@@ -4746,7 +4771,10 @@ async function useDevice(id) {
   if (!res) {  // refused (an install is running): show the headset still in use
     const current = dv.list.find(x => x.active);
     if (current) $("devSel").value = current.id;
+    return;
   }
+  // The Devices tab shows the headset now in use, not the one it was showing before.
+  if (dv.sel !== id) { dv.sel = id; dv.found = null; dv.editing = null; renderDevices(); }
 }
 async function devAction(label, body, btn) {
   const res = await act(label, () => api("/api/devices", body), btn);
@@ -4758,7 +4786,8 @@ function deviceStatus(d) {
   const s = link.s;
   if (d.active && s) {
     return s.phase === "connected" ? ["on", `Connected via ${viaText(s.via)}`]
-         : s.phase === "failed" ? ["off", s.error ? s.error.message : "Offline"] : ["", "Connecting…"];
+         : s.phase === "failed" ? ["off", s.error ? s.error.short || s.error.message : "Offline", s.error && s.error.message]
+         : ["", "Connecting…"];
   }
   const last = Math.max(0, ...d.addresses.map(a => a.last_ok || 0));
   return ["", d.transient ? "Not set up in Frame Control" : last ? `Last connected ${ago(last)}` : "Not connected yet"];
@@ -4768,10 +4797,10 @@ function renderDevices() {
   if (!dv.data) return;
   $("devCount").textContent = dv.list.length;
   $("devList").innerHTML = dv.list.map(d => {
-    const [dot, text] = deviceStatus(d);
+    const [dot, text, full] = deviceStatus(d);
     return `<div class="item dev-item${d.id === dv.sel ? " sel" : ""}" data-dev="${esc(d.id)}">
       <span class="dot ${dot}"></span><div class="grow"><div class="t">${esc(d.name)} ${d.active ? '<span class="tag">In use</span>' : ""}</div>
-      <div class="s">ssh ${esc(d.alias)} · ${esc(text)}</div></div>
+      <div class="s"${full ? ` title="${esc(full)}"` : ""}>ssh ${esc(d.alias)} · ${esc(text)}</div></div>
       ${d.active ? "" : `<button class="small" data-use="${esc(d.id)}">Use this headset</button>`}</div>`;
   }).join("") || `<div class="sub">No headsets yet. Add one to get started.</div>`;
   $("devList").querySelectorAll("[data-dev]").forEach(el => el.onclick = e => {
@@ -4822,10 +4851,10 @@ function renderDetail() {
       <button class="small" data-mv="1" title="Try later"${i < d.addresses.length - 1 ? "" : " disabled"}>↓</button>
       <button class="small" data-ed>Edit</button><button class="small danger" data-rm>Remove</button></div>`).join("");
   const err = d.active && link.s && link.s.phase === "failed" ? link.s.error : null;
-  const problem = err && err.hostkey ? `<div class="item" role="alert"><div class="grow"><div class="t">${err.hostkey.state === "missing"
+  const problem = err && err.hostkey ? `<div class="item dev-problem" role="alert"><div class="grow"><div class="t">${err.hostkey.state === "missing"
         ? "Its saved SSH identity is missing" : "Its SSH identity has changed"}</div>
         <div class="s">${esc(err.message)}</div></div><button class="small action" id="dvTrustNow">Check the new key…</button></div>`
-    : err && err.auth === "denied" ? `<div class="item" role="alert"><div class="grow"><div class="t">The headset didn't accept this computer's key</div>
+    : err && err.auth === "denied" ? `<div class="item dev-problem" role="alert"><div class="grow"><div class="t">The headset didn't accept this computer's key</div>
         <div class="s">Set Up Connection pairs this computer with it again. On the headset, Developer Mode must be on.</div></div>
         <button class="small action" id="dvPair">Set Up Connection…</button></div>` : "";
   el.innerHTML = `<div class="shelf-head"><h2>${esc(d.name)}</h2>${d.active ? '<span class="count">In use</span>' : ""}<span class="spacer"></span>

ui/server.py

@@ -403,11 +403,38 @@ def status(_body):
     s = json.loads(ssh("python3 -", stdin=(HERE / "frame_status.py").read_text(), timeout=20))
     osr = s.get("os") if isinstance(s, dict) else None
     if isinstance(osr, dict):
-        frame_telemetry.frame_seen(osr.get("build"), osr.get("version"))
+        # The headset's hash is for usage statistics only; the page and MCP clients don't get it.
+        frame_telemetry.frame_seen(osr.get("build"), osr.get("version"), osr.pop("headset", None),
+                                   len(LINK.reg.devices()) if LINK else None)
         frame_report.frame.update(build=osr.get("build"), version=osr.get("version"))
     return s
 
 
+COMFORT_FILES = ("frame_comfort.py", "frame_status.py", "frame_steam.py")
+COMFORT_MISSING = {"comfort": "missing"}  # what COMFORT_RUN prints when the cached copy isn't complete
+# Runs frame_comfort.py from ~/.cache/frame-control/comfort/<version>, first writing any of
+# `files` that aren't there (none: just check they all are). Private, and each file written
+# whole or not at all, so a copy cut short is finished by the next call.
+COMFORT_RUN = """import json, os, pathlib, subprocess, sys
+os.umask(0o077)
+files = %r
+root = pathlib.Path.home() / '.cache/frame-control/comfort' / %r
+if files:
+    root.mkdir(parents=True, exist_ok=True)
+for name, source in files.items():
+    path = root / name
+    if not path.exists():
+        tmp = root / (name + '.' + str(os.getpid()))
+        tmp.write_text(source)
+        tmp.replace(path)
+if not all((root / name).is_file() for name in %r):
+    print('{"comfort": "missing"}')
+    sys.exit()
+r = subprocess.run([sys.executable, str(root / 'frame_comfort.py'), %r], capture_output=True, text=True)
+print(r.stdout, end='')
+"""
+
+
 def comfort(body):
     try:
         frame_comfort.validate(body)
@@ -415,25 +442,14 @@ def comfort(body):
         raise Failure(str(e), 400)
     # Content-addressed, user-only helper bundle. Desktop and phone use the same
     # on-headset state/lock; no listener, service registration or third-party app.
-    import hashlib
-    files = {name: (HERE / name).read_text() for name in
-             ("frame_comfort.py", "frame_status.py", "frame_steam.py")}
+    # The page polls this every few seconds, so run the copy already on the headset and
+    # send the sources only when that version isn't there yet.
+    files = {name: (HERE / name).read_text() for name in COMFORT_FILES}
     version = hashlib.sha256(json.dumps(files, sort_keys=True).encode()).hexdigest()[:16]
-    script = """import json, os, pathlib, subprocess, sys
-os.umask(0o077)
-files = %r
-root = pathlib.Path.home() / '.cache/frame-control/comfort' / %r
-root.mkdir(parents=True, exist_ok=True)
-for name, source in files.items():
-    path = root / name
-    if not path.exists():
-        tmp = root / (name + '.' + str(os.getpid()))
-        tmp.write_text(source)
-        tmp.replace(path)
-r = subprocess.run([sys.executable, str(root / 'frame_comfort.py'), %r], capture_output=True, text=True)
-print(r.stdout, end='')
-""" % (files, version, json.dumps(body))
-    out = json.loads(ssh("python3 -", stdin=script, timeout=65))
+    out = json.loads(ssh("python3 -", stdin=COMFORT_RUN % ({}, version, COMFORT_FILES, json.dumps(body)), timeout=65))
+    if out == COMFORT_MISSING:
+        out = json.loads(ssh("python3 -", stdin=COMFORT_RUN % (files, version, COMFORT_FILES, json.dumps(body)),
+                             timeout=65))
     if out.get("error") and "active" not in out:
         raise Failure(out["error"], 409)
     return out
@@ -2280,6 +2296,10 @@ def connection_state():
 
 # ---- HTTP ------------------------------------------------------------------
 
+# POST reads the page repeats on a timer, as (path, action).
+POLLS = {("/api/comfort", "status")}
+
+
 def action_of(body):
     """The action a request asked for, for diagnostics: a short word, never user data."""
     a = body.get("action") if isinstance(body, dict) else None
@@ -2533,7 +2553,10 @@ class Handler(BaseHTTPRequestHandler):
         except ClientGone:
             raise
         except Failure as e:
-            if e.status >= 500:
+            # A poll that couldn't reach the headset is the headset asleep or away, which the page
+            # already shows; like GET /api/status, not an error to report.
+            polled = (path, action_of(body)) in POLLS and frame_telemetry.link_failed(e)
+            if e.status >= 500 and not polled:
                 frame_telemetry.diagnostic(f"POST {path} {action_of(body)}", e)
             self.send_error_json(str(e), e.status, e.apk)
         except (ValueError, TypeError) as e: