close#2183
Reusing a room means whoever enters first becomes the host, and moving the
session to someone else's library meant everyone leaving and re-entering in
the right order. The host can now promote any connected guest from the
participant list (Watch Together screen and the in-player session sheet),
with playback, control mode, and reconnect identities carried across.
The relay owns the swap: a new `transferHost` message validates the sender
is the live host and the target a connected modern-protocol guest, then
swaps `HostPeerID` and the reconnect verifiers (each peer keeps its own
token), persists the room, and broadcasts `hostChanged` to every peer.
Clients flip roles in place: the controller swaps its role engine while
keeping the session, message queue, and player attachment. A promoted guest
seeds the coordinator with the room's last known intent (a paused room
stays paused) and the known-peer roster so the fresh epoch re-gates
instead of solo-starting; a demoted host falls back to a reconciler with a
fresh clock sync and asks the new authority for state. Guests re-pin the
host identity, reset their sequence, and re-converge their clocks.
A rejected transfer (`not_host` / `peer_not_found`) surfaces as a toast
instead of tearing the session down. Pre-transfer app builds in the room
ignore `hostChanged` and stop following the room at the next transfer;
targets on an old sync protocol are not offered the action.
Two relay setup tests failed on CI while the same commit had passed a
few minutes earlier. The harness rewrote every 10-second timer in the
zone to 10 ms, which also compressed the connect deadline of the
recovery that runs once setup retries are exhausted. On a loaded runner
that deadline expired after the relay had upgraded the socket but
before the client sent its first frame, so the recorded connection
carried no messages and shifted the expected sequence; when all three
recovery attempts were starved this way, the recovery never announced
at all.
A reply that never comes is safe to compress, a real loopback handshake
is not, and one zone hook matching on duration cannot tell them apart.
The service now exposes the setup acknowledgement, the release
handshake, and the release acknowledgement as separate budgets, each
still 10 seconds in production, and every test sets only the one it
means to expire. The blunt helper is gone; what remains collapses the
250 ms and 500 ms retry delays, which are waits rather than races.
Verified by reproducing both failure modes under CPU saturation, about
one run in ten, then 30 consecutive clean runs of the file under the
same load and a full suite at 6332 passing.
Non-English users saw English text in a dozen places and blank labels in
sixteen more.
The English came from sites that produce their copy away from the widget
that renders it, which is what the structural hardcoded-string check
cannot see: picture-in-picture refused with a raw literal instead of the
pipErrors.notSupported key that already existed; the two Jellyfin/Emby
auth throws missing display: rendered their developer message on the
add-server form; ServerParsingException.toString() fed its English into
the localized "Failed to load servers" wrapper; Watch Together
interpolated the whole PeerError, so a failed create read "Failed to
create session: PeerError(PeerErrorType.timeout): Timed out creating
session" and join printed its prefix twice; the hub and playlist
continuation footers rendered exception.toString(); shader rows showed an
English title over an already translated subtitle; the player queue fell
back to the raw Dart enum name; a Plex home user with no title showed
"Unknown"; a failed player start showed "Exception: Failed to initialize
player"; and the tvOS top-shelf header was hardcoded in an extension that
has no Flutter engine.
The blanks came from three recent features that added English keys
without translations. clean_translations.py filled all 21 siblings with
empty strings, so the Android TV resolution switch, every Jellyfin/Emby
recording-rule field, the demuxer row, the Companion Remote address
caption and the Seerr blocklist pill rendered nothing at all.
Two fixes are structural rather than key swaps. ContinuationStatusSliver
now takes an errorContext and calls a new non-logging
localizedLoadErrorText, so no future throw can leak through it. lib/mpv
stays free of user-facing copy: it raises a PlayerInitializationException
sentinel and a PlayerError.playerInitFailed cause tag that the player
screen resolves to localized text. The tvOS section title travels in the
shelf payload, additively, so an older cache still renders.
MemoryLogOutput extended LogOutput but was never wired as a Logger output (storage happens in the printer); DeletionNotifier/WatchStateNotifier's forServer/forItem filtered streams and WatchStateEvent.mediaType had no production consumers (tests now filter .stream directly); context.hiddenLibraries/profileSettings and toPlexUrl had zero call sites; MonoTokens.splashFactory was never read; OptimizedMediaImage's enableTranscoding/cacheKey chains, FocusableMediaCard.width/height/forceGridMode, TvBrowseRail's constant-zero gap functions, and media_image_helper's scaleFactor were never varied by any caller. The sha1 LRU stays (rebuild-hot artwork URLs); TvRailTrailing.none and the mono-theme copyWith stay live (lerp delegates to copyWith). Also carries the MusicPlayContext.id argument drops in the music screens and remaining sweep test updates.
HostCoordinatorCallbacks and GuestReconcilerCallbacks existed only to be unpacked by the controller through ten forwarding lambdas into its own flat fields. The engines now take individual nullable callbacks; the controller keeps its late-binding lambdas (the provider assigns the public fields after construction), and the test harnesses forward the callbacks they exercise.
Guests joining an "Anyone" room saw "Host controls playback" and a
locked room until the host actually started something. Control mode
only travelled inside the host's PlaybackState broadcast, and every
broadcast path requires an active media epoch, so an idle lobby had no
carrier at all: guests sat on the joinAsGuest hostOnly default. The v1
protocol's sessionConfig message covered this; the v3 rewrite lost it.
Carry the mode on the host's join messages instead: the directed join
reply every participant already sends to a new peer, and the host's
reconnect re-announce. The field is optional on the wire ('cm'), so
older clients ignore it and rooms with older hosts degrade to the
previous behavior. Guests apply it only from the relay-derived host
peer ID, never from a join's own spoofable isHost flag.
close#1950
An in-place source switch — audio, subtitle, version or quality — detaches
the host's player for the duration of the reload. `_broadcast` falls back to
a position of 0 when no player is attached, so any state published in that
window names 0:00 as the authoritative position and every guest hard-seeks
to the start of the item.
Heartbeats already suppress themselves while detached, which is why this
hides: the paths that leak the zero are the ones that answer on demand.
`onStateRequested`, `onPeerJoined` and `onReconnected` all broadcast
regardless of whether a player is attached, so a guest entering the player,
joining, or reconnecting mid-reload is the trigger.
Fall back to the last broadcast anchor instead. That field is only assigned
for untargeted broadcasts, so it holds the last position the room was
actually told, and the reload's own re-attach path already re-anchors from
it once the player comes back.
DD-3 gives video no exemption: a restricted vehicle must not play it at all. The
gate is read at the single point where media actually opens, so every path that
can start a picture - an explicit play, a gapless arm, a track or channel switch,
a frame-rate-match resume, a reload, and the queue navigation commands of the OS
media session - is covered by one check rather than by a guard at each call site.
A seek can also start playback with no play call, because mpv resumes when it
seeks off the end of a file, so a restricted seek is followed by a pause.
Watch Together needed the pause to be local. A vehicle stopping one peer is not a
room-wide intent: a guest's forced pause is swallowed by the attachment's ledger
rather than published, while a host's still pauses the room, because a host that
kept broadcasting a frozen anchor would stall or rewind every guest it was meant
to protect. The layer that owns a pause owns the resume for it, and one
acknowledgement is recorded per event, so a surplus cannot eat the user's next
real pause.
A room whose peers have all left is a code nobody is using, but the relay
kept it bound to the creator's reconnect capability and rejected every
other create with room_exists. The app compounded it: enterRoom only
promoted to host on room_not_found, so tapping a recent code landed the
user in the retained room as a guest of a host that was never coming
back, until the cleanup sweep finally dropped the room.
Create now replaces a room with no connected peers, and enterRoom hosts
the code when its probe join finds an empty room. An occupied room still
rejects create, including from its previous owner, and a host that is
merely disconnected still reclaims its peer ID through join with the
matching token.
The Dart suite spent 77% of its cost compiling one isolate per test file
while `flutter test` used half the cores, and every Maestro flow replayed
a full Jellyfin onboarding before its first real assertion.
- Add scripts/run_tests.sh, which runs `flutter test` with -j set to the
cores the process may actually use instead of the ncpu/2 default.
Measured on 8 cores: 190s -> 136s; -j 12 regresses to 165s, so it scales
to the core count rather than hard-coding one. CI and CONTRIBUTING use it.
A cgroup v2 quota, a cgroup v1 quota, and the cpuset/affinity nproc
reports can each be the binding limit independently, so the detector
takes the smallest; trusting whichever it found first would oversubscribe
4x on a container holding an 8-CPU quota while pinned to 2. Covered by
scripts/test_run_tests.py, which the ci_guard_checks.sh glob picks up.
- Add .maestro/subflows/ensure_onboarded.yaml: cold-start the app and only
onboard when no session is stored. Flows that just need a signed-in Home
use it; 02_onboarding_home, 08_logout, 09_download_offline_playback and
the profile regressions keep clearing state. 59s -> 16s per flow.
- Guard onboarding's two optional taps behind visibility checks. A missed
`optional: true` tap still runs the full element search, costing 3.0s
and 7.8s per onboarding to find nothing.
- Disable device animation scales in run_maestro.py, restored by the
existing cleanup path. CI's emulator got this from the runner flag;
physical devices never did.
- Shorten the watch_together setup-timeout replacement from 500ms to the
10ms the same file already proves sufficient, and shorten the retry
backoff at the one site that missed it: 8.04s -> 1.59s of execution.
- Make the LAN discovery waits deadline-based and resend the beacon while
polling. Loopback UDP drops datagrams under load, which timed out a
wait that could never be satisfied; this was the suite's one flaky test.
- Fix 08_logout, which searched for "Logout" and "Are you sure you want to
logout?" after both strings became "Log out". The flow had been failing
and aborting the suite before 09 ever ran.
flutter test 190s -> 131s. Maestro's Android suite 621s -> 385s across the
eight flows the baseline reached, and now runs all nine green.
Guest switch dispatch pre-marked its dedup key and fired-and-forgot, so
any failure (fetch error, reload busy with an auto-advance, navigation
race with the host exiting) silently stranded the guest on the old media.
A CurrentPlaybackDispatcher now marks a key handled only after the sink
reports success against the committed identity, with a serialized
in-flight slot, timeout, and generation reset; the reconciler re-offers
unattached media on every host heartbeat, making the heartbeat the retry
channel. Fetches that outlive their dispatch are re-validated against the
current snapshot so a stale switch can't override the live one.
hostExitedPlayer now rides the controller's ordered message queue with
host authentication instead of racing state handling in the provider.
Replaces the imperative play/pause/seek/positionSync message soup with a
single host-authored PlaybackState (seq-ordered, anchor-extrapolated,
phase machine: loading/waitingForPeers/paused/playing) that doubles as
the heartbeat, plus guest status reports and host-applied control
requests. Fixes the guest seek-back loop while the host loads (readiness
was keyed on a pre-load !buffering snapshot and heartbeats broadcast
frozen positions), adds real group buffering coordination (stall grace,
scheduled simultaneous resumes, 15s safety timeout), rate-nudge drift
correction with passthrough-aware seek fallback, session-scoped message
handling (no lost messages during episode-switch detach gaps), and an
expected-state ledger replacing the racy remote-action flag.
edde746