A title's copy disappeared from a catalog item's library matches when the
server holding it went offline, and the screen then claimed the title was not
in the library at all.
The reverse-lookup fan-out only reaches online clients, so a registered
server that is offline lands in neither the succeeded, failed nor cancelled
set. The fold read that absence as "left the account" and dropped the
server's verified copies. Worse, the surviving wave looked complete, so it
was memoized for the rest of the profile session and never asked again. With
no server online at all, every server's copies were erased and the detail
screen asserted "Not in your library" over servers nobody had queried.
LibraryLookupResult now names the registered servers a wave could not even
reach, which needs a registered-server set on MultiServerManager because an
auth-rejected Plex server holds no client at all. The fold's rule is stated
positively -- only a server that answered may replace its own entry -- a wave
that skipped a server is never memoized past the TTL, and the detail screen
counts those servers as unchecked alongside the ones that failed.
A live library refresh or a sort change moved the highlight to a different
title, and Select then opened that one instead of the one the viewer chose.
Grid focus nodes are keyed by index while the cards are keyed by item, so a
merge that inserts an item before the focused slot leaves the highlight
pinned to the slot rather than the title: Flutter parks the node across the
rebuild and the replacement card re-attaches it. The library refresh already
compensated the scroll offset for that same index shift; focus is equally
index-pinned and was not compensated at all.
GridFocusNodeMixin.remapGridFocus carries the highlight with the item, and
the three grids that mutate content in place now use it: library browse, the
paginated card grid behind collections and playlists, and hub detail -- where
a user changing the sort was enough, no server push needed. Hub cards also
gain a key; they had none, so the element was silently updated with a
different item. The key is the global one, because aggregated hubs such as
Continue Watching can hold colliding per-server ids.
A promoted host started playing while other participants were still loading,
and could run at a different speed than the one it told the room.
The coordinator resolves the first epoch's readiness synchronously inside
attach, so seeding the known-peer roster afterwards had already missed it:
the fresh epoch saw an empty room, solo-started, and marked the first start
complete, after which the still-loading peers no longer gated anything. The
roster now seeds in _createCoordinator, so no coordinator exists without the
room it has to wait for.
Adopting the room's rate on promotion only set the value the coordinator
broadcasts. A guest that was paused when the rate changed never applied it to
its player -- position is aligned while stopped, rate is not -- so the new
host advertised one speed and ran another, and every guest kept correcting
against the difference. The host player is the room clock, so an adopted rate
is now applied to it.
A guest joining while the host handed the room to someone else could end up
following the host everyone else had just left behind.
join installs the client into the room and snapshots HostPeerID under
room.mu, then released the lock before enqueueing joined. The client is a
hostChanged recipient and a legal transfer target from the moment it is
installed, so a transfer committing in that gap enqueued the newer authority
first and the stale joined frame overwrote it on the client. transferHost
already enqueues while holding the lock for exactly this reason; joined
carries the same authority field and did not.
The admission frame is now built and enqueued while the room is still locked,
and enqueueFrame never blocks, so nothing waits on network I/O under the
lock. A beforeJoinRoomAck barrier alongside the two existing test hooks pins
the ordering: a transfer cannot commit while an admission is unpublished.
A host that handed the room to someone else and then lost its connection
could not get back in, and a guest promoted while it was offline came back
as a guest.
The relay names the host in every admission and every hostChanged, but the
client also kept its own _isHost flag beside that identity. Reconnect adopted
the relay's host id without recomputing the flag, so the two disagreed: a
demoted host still required the response to name itself and rejected its own
legitimate re-admission as an invalid response, retry after retry; a promoted
guest updated the session but left the transport a guest, so ending the
session sent leave instead of endSession and room re-creation stayed off.
The role is now derived from the host identity rather than stored next to it.
Before the relay has admitted us there is no authority yet, so the role is
the one we announced, which is what releasing a possibly-committed setup has
to go by. The identity assertion moves from mutable role state to message
semantics: a created response must name us, a joined response is adopted.
The detail page's audio/subtitle preview could contradict playback on Jellyfin and Emby: a server default of -1 (subtitles off) previewed as a subtitle track, and a missing default with a container-default subtitle previewed as on where playback plays none (#1779 again).
The preview rebuilt a MediaSourceInfo from the item's version by hand, a third mapping beside the two the backends use for playback, and had already diverged twice. It now runs the ladder over the MediaSourceInfo the backend maps for playback, fetched from the metadata cache in the existing probe, and the hand-built mapping is deleted.
A catalog item that had a copy on server A lost it from the detail page when a later refresh could not reach A: partial results expire after the negative TTL and the fresh wave replaced the cache as a unit, so A failing while B answered empty left nothing, with no evidence A had removed anything.
The cache now holds per-server answers with one invariant: a server's answer is replaced only by that server. Servers that failed or were cancelled keep their last-known copies, servers named in no set have left the account, and the failed set still passes through so the outage stays visible and the wave is retried.
A library notification connection that timed out could still complete later and stay open after the owner was disposed. IOWebSocketChannel.connect applies connectTimeout with Future.timeout and drops the pending connect, so nothing could ever close a socket that finished late.
The channel factory now resolves only to established channels: it owns WebSocket.connect, applies its own deadline, closes a late upgrade, and returns a channel built from the resolved socket. The socket class no longer holds half-open channels, which also removes the connected-flag workaround around the library's close semantics.
The minified instrumentation variant, which the R8 reachability gate builds, no longer compiled: MpvLifecycleDeviceTest references MpvLifecycleTestActivity, which exists only in the debug source set. The test moves to androidTestDebug so it is compiled only when instrumentation targets debug.
A guest whose connection dropped during a host transfer could not get back in. The relay broadcasts hostChanged only to connected peers, and the guest's reconnect rejected the re-admission because the host differed from the one it had pinned, then left the room.
The pin predates transfers. The relay is the authority on host identity and verifies the reconnect token, so a valid re-admission naming a different host can only be a transfer: the client now adopts it and surfaces it through the same onHostChanged path.
On the relay, hostChanged is enqueued while the room lock is still held. Two transfers in quick succession run on different connections, and enqueueing after unlock could deliver the older authority change after the newer one; enqueueFrame never blocks, so holding the lock across it is safe.
Transferring the room to a 2.18.0 client left it with no host: that build speaks the same sync protocol version but does not handle the relay's hostChanged broadcast, so eligibility passed, the relay moved authority, the current host stepped down and the target never stepped up. A 2.18.0 bystander kept following the demoted host.
Join messages now advertise capabilities beside the version (cap: [hostTransfer]). A transfer requires the target and every same-version bystander to advertise it; peers on another version are already outside the room's sync and do not count. A version bump would have excluded every older peer from mixed rooms for a feature they may never use.
An older relay rejects transferHost as invalid_message; while a transfer is pending that is now a failed transfer, not a session error.
A host that stalled near the end of a file kept the whole room paused for good. Recovery demanded three times the stall in buffered headroom, and with ten seconds of media left a four-second stall asked for twelve that could never arrive; the known-cache branch re-checked every half second forever.
The two limits are now explicit: how much headroom to want (scaled by the stall, capped by what is left to buffer) and how long to wait for it (a deadline from the stall's end, shared with the no-cache branch). selfRecoveryMaxHoldMs was a headroom cap and is renamed to say so.
Selecting Home from the sidebar on TV left focus on the collapsed sidebar item: the content appeared, but the remote went nowhere. MainScreen hands a tab over while focus still sits on the sidebar item that selected it, and the guard added to lapse stale rail-focus claims read that as the user having navigated away.
Where focus sits cannot tell a live handoff from a stale claim. What distinguishes a stale claim is that focus moved after it was armed, so the claim now records the primary focus at arm time and lapses only once focus has changed and rests off-screen. A handoff made while hubs are still loading is honored when they land; a claim armed on a bare scope still lapses when the user moves to the sidebar.
The subtitle "Render Resolution" setting was shown to every Android user
but only reached the ExoPlayer overlay; the mpv vo=mediacodec OSD plane
always rasterized at the full surface size. On a Fire TV Stick 4K Max
Gen 2 the edde746/plezy#2242 phone sign then cost up to 2 s of libass
render plus composite per frame at 1080p, showing late and lingering past
the cut.
Pass the fraction to the mpv core on initialize and give the OSD
SurfaceView a fixed buffer size below its view size; mpv rasterizes at
that size and the compositor scales the plane. At 1/2 the same sign
tracks its zoom at 12-24 updates/s and clears on the cut.
On TV, pressing Up on the Logs screen put focus on the back button and no
direction key could move it again, so the upload/copy actions were out of
reach (edde746/plezy#2242). The SelectionArea added in 8fe7e4bcf installs
always-enabled caret-movement actions that Android's default text-editing
shortcuts map plain arrow keys to; the keys were consumed as no-op
selection moves and never reached DirectionalFocusIntent.
Override those two intents above the region with an action that is
disabled for the collapsing (unshifted) variants and defers to the
region's own handler otherwise, so Shift+Arrow selection keeps working.
"Add to Watchlist" disappeared from library card menus after the first open, and
the detail-screen bookmark never appeared, for every Plex movie and show (#1873).
Plex Discover's `/library/metadata/matches` answers a `guid` lookup only when
paired with the numeric metadata `type`; a bare guid returns an empty container.
The empty result was cached as "no source can hold this item", hiding the entry
on the next open.
`PlexDiscoverClient.match` now takes the item kind and sends `type=1` (movie) or
`type=2` (show) alongside the guid; other kinds return null without a request.
Fixes#1873
Adding a Trakt, MAL or library title to the Plex watchlist failed with "no
rating key": /library/metadata/matches answers an empty container unless
`type` is sent, so every external-id resolution against Discover came back
empty. The client now passes the movie/show type with the guid.
The Explore library lookup spent up to eight search-index queries per
server per tap, plus one children fetch per candidate for sequels, and the
Trakt path added two alias/translation requests per detail open to reach
romaji and localized library titles.
Both backends index originalTitle on every copy of a foreign title whatever
language it is filed under, so the lookup now leads with the native title
and drops alternate titles as candidates: one query reaches the English,
romaji, localized and native copies alike. Plex's /library/all?guid= filter
is a literal prefix match with comma as OR, so every legacy-agent guid form
the external ids imply rides one indexed request alongside the Discover
guid, and the sequel gate asks for the season of every candidate in one
/library/all?type=3&show.id=…&season.index=N request. Worst case is five
concurrent requests per server, typically two or three, with no cloud call.
The Trakt alias/translation fetch is removed: the copies it reached are the
ones the native title reaches, and Trakt already sends original_title.
close#2098
With Steam Input enabled on Linux, every controller press acted twice:
Steam's desktop layout injects arrow/Enter/Escape keys through a uinput
keyboard while the physical controller stays readable over evdev, so
GamepadService synthesized the same key a second time. The duplicate
guard that fixed this on Windows was gated to Platform.isWindows and its
native key handler was never registered on Linux. Enable both on Linux.
close#1694
A Trakt item only carried its English title, so a Plex or Jellyfin library
filed under the romaji title (`Sousou no Frieren`), the user's language
(`Frieren: Tras finalizar el viaje`) or the native title stayed invisible to
the Explore library lookup.
The Trakt detail load now fetches /aliases and /translations/{app locale}
alongside people and related, picks one ASCII alias from the item's own
country plus the locale translation as alternate titles, and the matcher
reads original_title for every source. The lookup budget grows to four
title families, the matcher keys its cache on the candidate titles, and the
detail screen re-resolves when a detail load adds titles, not only ids.
close#2098
Explore showed "Not in your library" for titles the user owns. Two causes:
Plex's /library/all?title= filter is an ordered word-prefix substring match,
so `Oshi no Ko` never found a library titled `[Oshi no Ko]` and a romaji or
localized title was unreachable once the first candidate hit; and a server
that answered slower than the 10s header budget was treated as a dead
endpoint, dropped from the wave, cached as a negative for ten minutes and
cascaded through its stale LAN candidates.
Plex title candidates now go through /hubs/search with includeGuids, the
full-text index behind Plex's own search, which also covers originalTitle.
Both backends search every candidate concurrently and union the id-verified
copies instead of stopping at the first title that hit; the candidate cap
becomes two title families so a sequel can reach its alternate title. Lookup
requests run under a dedicated deadline with endpoint failover off. The
aggregation layer reports per-server failures, the matcher only memoizes
complete positive waves, and the detail screen shows "Couldn't check N
servers" instead of claiming absence.
close#2098
H.264 High 10 on Android TV boxes without a 10-bit hardware decoder
(Amlogic S905X4 class: onn 4K Pro, Homatics Box R) started black for over
ten seconds and then played choppy: mpv tried MediaCodec first, hit
"Could not initialize video chain", and only then fell back to software
decode on the GL vo, where the Mali-G31 driver has no
GL_EXT_texture_norm16 and pays an integer-texture conversion pass plus
lanczos scaling it cannot afford at 1080p.
Decide the decoder before mpv creates it. An on_preloaded hook now carries
both per-file policies (Dolby Vision P5 reshaping and this one): when the
track's codec-profile is High 10 (published at demux by the fork, see the
mpv-build pin) and MediaCodecList advertises no hardware AVCProfileHigh10,
hwdec is held at `no` and the session goes to the GL vo directly. The hold
parks any hwdec write from Dart and restores it for the next file, as the
DV P5 hold already did.
On a GL vo whose driver lacks norm16 (probed once through a pbuffer EGL
context), scale/cscale/dscale drop to bilinear and dither to off for the
session, restored when the vo goes back to the plane. Only options still at
their mpv defaults are touched, so a user's mpv.conf wins. Paired with the
fork's rg8-backed 16-bit plane emulation, the box goes from 13 drops/s and
a 13.7 s first frame to 0.7 drops/s and 2.5 s. Devices with norm16 (Shield,
Pixel 7) take neither the tier nor the emulation.
The native pin moves to edde746/mpv-build@0601b034da, which also picks up
the earlier android/linux/windows patch-series squash.
close#2065
The phone hero's chip rows were spaced 8px apart with 12/6 padding and a
16px gap to the action row, which read loose against the centred layout.
The scores pill and the tappable Rate chip also came out taller than the
plain text chips because their 16px icons pushed past the 13px label
line, so the row had a visible step in it.
Chip geometry now lives in one _HeroChips table: 4px spacing between
chips and rows, 10/5 padding, a 10px (6px on short heroes) gap to the
actions, and every chip sizes its content to a shared 20px box so icons
and text can never make one chip taller than its neighbours.
Landscape phones got a hero at 60% of a short viewport, which shrank the
logo to a sliver and let the overview run under the camera cutout. The
hero now has a floor of one full hero — status bar and back strip,
full-size logo, both chip rows and the action row — and the body sliver
sits in a SliverSafeArea so section text keeps the same horizontal inset
as the hero content above it. The hero's top padding also reserves the
back-button strip explicitly instead of relying on the logo budget.
On phones the movie/show hero hugged the left edge under a full-bleed
backdrop while the new collection page stacks its poster, title and
actions on the centre line, so the two pages read differently.
At widths under the mobile breakpoint the hero now centres the clear
logo (or title fallback), both chip rows and the action row; the hero
height, chip shedding, focus order and back button are unchanged. Wide
heroes keep the bottom-left column — a 400px logo centred in a
tablet-wide hero floats, and the wide collection header is left-aligned
too. TV is untouched.
_buildDetailLogoOrTitle gains an alignment and _buildDetailTitle a
textAlign, both defaulting to the previous values.
Collections opened as a bare poster grid under a plain app bar: no
collection artwork, no summary, no count, so they looked unrelated to
the movie/show and album pages around them (#1493).
The collection screen now leads with the same header the album and
artist pages use: the collection poster (square for music collections)
beside its title, item count, year span, content rating and collapsible
summary, with the Play / Shuffle / Download / Delete row underneath.
The collection's backdrop art washes the header region behind a scrim
that fades to the scaffold background before the grid starts; a
collection with only a poster gets a blurred copy of it instead so the
wash still reads as colour. A transparent title bar fades in once the
header scrolls away, and the circular back button matches the detail
pages. The year span is only shown once every page has loaded so a
partially fetched collection never reports a narrower range.
buildDetailScaffold gains optional `behind`/`above` layers so a detail
screen can paint art under, and chrome over, its scroll view without
re-implementing the overlay-sheet host and back handling.
Verified on Pixel 9a, 10" tablet and Android TV emulators: touch scroll
and pinned title, D-pad up from the grid lands on the action row and
scrolls the header back into view, D-pad back pops, Plex collection with
art and Jellyfin collection without art.
Snapping the window with Aero Snap (edge drag, Win+Arrow, Snap Layouts)
and quitting restored it where it sat before the snap. Windows keeps a
snapped window in the "arranged" state: GetWindowPlacement still reports
SW_SHOWNORMAL with the pre-snap rect in rcNormalPosition, so the saved
placement never reflected what was on screen.
Resolve IsWindowArranged from user32 at runtime (exported since Windows
10 1903, no header declaration) and, when the window is arranged, save
GetWindowRect converted to workspace coordinates instead. The same
correction covers the pre-fullscreen placement captured by the native
fullscreen toggle. Relaunch lands a normal window on the snapped rect;
there is no API to re-enter the snapped state.
close#1895
Keep package manager instructions out of the main download section until expanded. Add the community RPM repository as a setup link and place Moss last, below Nix.
Turning auto-skip off still surfaced a skip button on every intro and
credits marker, so a viewer who wants to watch episodes in full had no
way to silence it. The two auto-skip switches become per-marker
selectors — Off, Show button, Automatic — stored as skip_intro_mode and
skip_credits_mode. Off filters the marker before it becomes current, so
the player behaves as if the server had sent none: no prompt, no
countdown, no TV autofocus, and Back reaches the screen. Flipping a kind
to Off while its prompt is up drops the prompt immediately.
The legacy auto_skip_intro/auto_skip_credits booleans migrate on first
read: true becomes Automatic, false becomes Show button, matching what
each meant before.
close#2138
Watched posters always carried the corner checkmark, and viewers who
find it noisy had no way to turn it off. A new Appearance switch,
Show Watched Indicators (default on), gates the checkmark on every
surface that stamps watch state onto artwork: grid cards, hub rows,
folder tree rows, episode and playlist thumbnails. Progress bars and the
unwatched-count pill keep their own behavior.
close#1998
Apple TV shipped with passthrough opt-in while the E-AC-3/Atmos
sample-buffer renderer was unverified, so Atmos titles played as PCM
until the viewer found the switch. The renderer is now hardware-verified
on real receivers, so the pref defaults on for Apple TV the same way it
does for Android TV. An explicit stored value still wins.
close#1300
Playing a Plex live channel from outside the LAN always came back as a full
re-encode at the server's highest transcode tier (20 Mbps 1080p on the
reporter's server), no matter which quality the app was set to. The Plex live
path never sent a ceiling: a live source has no bitrate the server can check
against its remote-stream limit, so a remote session without a client cap
falls onto the server's own top tier. On the LAN the server remuxes anyway,
which hid the gap.
The live stream path now reads the same saved preset the library path and
the Jellyfin live path (#2198) already honor. Original is unchanged: a remux
(directStream=1) with no ceiling. A capped preset pins directStream=0 and
sends the bitrate limitation clause plus the videoResolution/videoQuality caps
so the encode lands at the chosen tier rather than the server's. The preset is
session state, so a recovery re-tune keeps the cap.
directPlay stays 0: a tuned Plex session is only reachable through the
transcoder's HLS output, so "no re-encode" on live means a remux, not direct
play. Capped sessions use the h264-only TS target because every codec in the
target becomes an encode output once directStream is off, and HEVC into TS is
the #1859 corruption; the broadcast-codec live target remains the Original
remux menu.
close#2072
A skip back of 10-20 s on a freshly tuned Plex Live TV channel landed on
or after the frame being shown, while the same skip worked once the viewer
had already time-shifted. Offset-less opens (initial tune, retry, channel
zap, subtitle switch at the edge) pinned stream position zero to wall clock
at open time, but the transcode starts behind real time by tuner ingest and
encoder start-up latency; offset opens were exact because the server
defines their origin.
Every /:/timeline response already carries the playback transcode as the
top-level TranscodeSession next to the CaptureBuffer wrapper; its timeStamp
is the epoch of stream position zero, which Plex's own client uses as the
playhead origin. The parser used to return only the capture window. It now
returns both, and each heartbeat re-anchors the clock on the playback
origin unless the response predates the current open or an open is still
calibrating. Offset-less opens seed a provisional anchor from the capture
edge instead of wall clock, and the live-edge threshold is widened to 15 s
so a live stream trailing the edge by normal latency still reads as live.
A disposal race could call blocking native destruction after initialization resumed on Android's main thread. Finish that teardown on IO, enforce the lifecycle threading invariant, and bound exit responsiveness and teardown latency in the device regression test.
Setting vo=null during disposal could reinitialize MediaCodec and deadlock vendor decoders while the surface was already being released. Terminate libmpv synchronously with its surface references intact, then release the Android surfaces, and cover repeated hardware-decoded teardown on devices.
The ExoPlayer carrier gate only asked whether the route could open a
192kHz/7.1 IEC 61937 track. TrueHD rides that same tuple, so a sink that
decodes Dolby but no DTS-HD passed the probe and was handed a DTS-HD
burst it discards: the AudioTrack initialises and drains, Android reports
no error, and neither the audio-recovery ladder nor the mpv fallback
fires, so the file plays with no audio at all.
Evaluate the gate per format. TrueHD keeps the transport-only probe, so
the #1804 routes are unchanged; DTS-HD additionally requires the route to
advertise ENCODING_DTS_HD, the pairing mpvSpdifCodecs already applies and
the reason mpv plays these files while ExoPlayer does not.
Declining the carrier is not a forced decode: the format falls through to
media3's raw path, which downgrades DTS-HD to the DTS core for receivers
that take the core but not the lossless stream.
Co-authored-by: claude-opus-5 (high effort, 1M context) <noreply@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
With the mpv backend on Android, every subtitle change cost two dropped
video frames on a Fire TV Cube (36 per minute with PGS SDH subtitles, 15
with srt, #2240) and accounts for the residual drops on the Shield in #2202,
whose sessions also had subtitles on. The diagnostic build showed the drops
in pairs: a frame handed over on time, then the video thread waking 90-100
ms late. vo_mediacodec rasterized subtitles and blitted them into the OSD
Surface on that thread, and locking the Surface waits CPU-side for the
compositor - dequeue plus the release fence, one or two vsyncs, 42-83 ms at
23.976 Hz. Kodi, VLC 4 and ExoPlayer all keep subtitle output off the
presentation thread.
Move the native pin to edde746/mpv-build@88c1b135e7: the whole OSD stage
runs on its own thread, fed with the video pts at every flip, so the video
thread never waits on the compositor or on libass. Subtitles latch on the
same vsync as the video in the common case and one later when the lock has
to wait. Shield on vo=mediacodec with ASS subtitles: no video-thread stage
over 10 ms and 0 drops across five 150 s runs, subtitles verified on
screen; DTS-HD MA 4K unchanged at 0 drops. The pin keeps the #2202
diagnostics and adds VO stage timing to them. Apple, Linux and Windows
assets are unchanged.
The smoothed passthrough clock from aeb06c2c9 (edde746/mpv-build 0105)
still let the Shield in #2202 drop a frame every 70 s or so on TrueHD and
DTS-HD MA. The average handles head jitter completely - on a Shield with
the raw head artificially held for 150 ms it went from 410 dropped frames
in 200 s to 0 - but it regresses on stalls: when the head freezes (device
underrun, rebuffer, a HAL that stops updating) while the wall clock runs,
the estimate runs past what was written, (written - head) wraps, and
getLatency reports 0 - the core sees the whole device buffer vanish in one
step and drops video. With the AO starved for 400 ms every 15 s that was 77
drops in 170 s against 8 with the raw clock, and a Box R 4K Plus hit the
wrap 361 times in four minutes of ordinary DTS-HD playback.
Move the native pin to edde746/mpv-build@a58f0c1336: the estimate is
clamped to the written position and its window restarted on overshoot, and
between reports it may deviate from "last report + elapsed" by at most 10%
of the elapsed time unless it is a second or more off - the two guards
media3's AudioTrackPositionTracker keeps over the same average. Starvation
run: 9 drops, no wrap; real HAL with pauses and seeks: 0 drops. The pin also
carries verbose-level diagnostics for this issue (per-drop timing in vo.c,
passthrough clock anomalies and a 10 s summary in ao_audiotrack) so the
next log from the reporter pins whatever remains. Apple, Linux and Windows
assets are unchanged (their content keys did not move).
Three findings from the #2239 investigation.
Discover armed "focus the browse rail when it has hubs" on its first
load and kept it armed indefinitely while the rail had nothing to show.
Hubs landing minutes later — reconnect, push refresh — then yanked the
remote off a sidebar item the user had since moved to. The claim now
lapses once focus sits on a control outside the screen; a bare content
scope (the startup state) still lets the rail take its initial focus.
On Apple TV a Menu press that reaches MainScreen at the home root is
consumed silently by design (the engine normally hands root Menu to
UIKit). That branch is what turned a focus slip into a "dead remote"
report, so it now logs passthrough state, picker state, and the primary
focus label.
SystemShelfService.debugReset awaited the mutation tail queued by the
previous widget test, whose FakeAsync zone ends without flushing the
microtasks that settle the chain, so any second test in a file using it
hung in setUp. It now drops the queue synchronously; the #2239 session
regression test moves back beside the profile-switch test it belongs
with.
On tvOS the profile picker shown after a background/standby resume lost
its focus highlight and the Siri Remote went dead until a force-quit.
Android TV hit the same shape with the player (#2034).
The picker and PIN dialog are pushed on the root navigator, above the
nested profile-session navigator. Nested routes still report
`isCurrent == true` under that cover, and Flutter leaves a covered
route's scope focusable, so any focus self-heal under MainScreen —
the sidebar reveal on an offline/online flip, a Libraries grid reload,
the TV browse rail — won the remote behind the picker. With menu
passthrough pinned off while the picker is up, Menu was swallowed too.
Restore the invariant once at the navigator boundary instead of at
every reclaim site: CoveredRouteFocusBoundary wraps the session in an
ExcludeFocus keyed on the root route's currency, so requests below a
cover are no-ops, and re-requests its own scope on uncover so focus
walks back to the leaf that had it (the covering route's pop culls the
excluded scope from the route scope's history before the exclusion
lifts, which is why Flutter's own restoration cannot). This makes
isRouteChainCurrent redundant; the player's guards return to plain
nested currency.
close#2239
A Seerr instance behind forward-auth (Authelia, Authentik, Cloudflare
Access) or HTTP Basic failed setup with "No Seerr instance at … (HTTP
200)": the probe followed the proxy's redirect to its login page and
decoded the HTML as "not JSON". A live session that later hit the same
wall took the proxy's 401 as Seerr's, re-authed through the wall, failed,
and unlinked a perfectly good stored session.
Seerr's API never redirects and always answers with JSON, so a 3xx or a
non-JSON 401/403 is the proxy talking. Seerr requests no longer follow
redirects, that shape maps to a SeerrProxyException with a message that
says what to do, the probe reports it in place of "no instance", and the
client neither re-auths nor unlinks on it.
close#1877
With ambient lighting enabled, ASS subtitles anchored to a corner (and PGS
bitmaps) rendered against the whole screen instead of the video: the effect
stretches mpv's frame to the window with video-aspect-override and lets the
shader composite the real picture inside it, and mpv places subtitles against
that stretched frame.
The pinned libmpv now carries --sub-video-rect-aspect (edde746/mpv-build
ec08018), which makes mpv derive its subtitle margins from the picture's real
aspect inside the displayed rect. AmbientLightingService hands it the video
aspect before overriding the frame, so a libmpv without the option refuses
cleanly with the frame untouched, and resets it on disable.
Verified on a Pixel 7 (2400x1080) with a 16:9 file whose OP credits are
\pos-anchored: subtitle extents are pixel-identical with ambient lighting on
and off, and mpv reports OSD borders l=240 r=240 for the overridden frame.
close#2120
On a phone in portrait the video controls header is too narrow: the clock crowds the back button, title, and track/chapter controls. Hide it there; landscape, tablets, desktop, and TV keep it.
Phones were locked to portrait outside the video player, and the mobile
shell had no answer for a wide, short viewport: car head units (which use
the mobile layout) and tablets in landscape spent their scarce height on
a bottom navigation bar and a 500dp hero.
In landscape the mobile shell now puts the bottom bar's destinations on a
leading Material NavigationRail (MobileNavigationRail) and keeps every
other mobile layout decision as it is. The rail keeps the bottom bar's
extras: the offline reconnect affordance as its leading action and the
long-press library quick picker on the Libraries destination. Labels
follow the existing nav-bar-labels setting but give way automatically
when the destinations cannot fit (a landscape phone is ~410dp tall), and
the rail scrolls as a last resort. The content beside it drops the leading
system inset the rail already absorbs.
Phones now allow every orientation; OrientationHelper.restoreDefaultOrientations
is context-free and the player exit path uses it instead of its own copy
of the phone lock. The Discover hero fills the viewport in mobile landscape
and compacts its logo and bottom offset when short, so its content stays
below the top bar.
The music mini-player learns a start inset (MiniPlayerInsetController.setNavInsets)
so it floats beside the rail in landscape and above the bar in portrait;
route suspension zeroes both.
Known follow-up: the media detail header is not yet laid out for a
landscape phone (logo runs under the status bar and back button).
On Android Automotive head units with a left- or right-positioned system
bar, the whole edge of the app rendered underneath it: Discover's title,
hero text and first card were clipped, and the player controls sat under
the bar. Play's car app quality AR-1 requires interactive UI to stay clear
of system bars.
The platform reports the bar as a systemBars inset and Flutter forwards it
as MediaQuery.padding.left/right, but the mobile screens only honour the
top and bottom insets. On the emulator this only showed after the first
video session (the player's edge-to-edge restore stops the DecorView from
fitting a navigation-bar-typed left bar); on Android 14+ CarSystemUI the
left bar is a status-bar-typed inset, so it is under the bar from launch.
Car bars are opaque and may be impossible to hide, so nothing is worth
drawing under them: FormFactorScale now wraps the automotive surface in a
horizontal SafeArea inside the scaled MediaQuery, consuming the insets
once for every route.
Reproduced and verified on the API 33 automotive image with the
com.android.systemui.rro.left overlay enabled for the foreground user.
Downloaded episodes never showed the partially-watched progress bar, so a
user resuming offline could not tell which episode they had started.
Offline playback already records progress: the tracker queues a progress
action and emits a WatchStateNotifier event, and DownloadProvider hydrates
those rows back into WatchStateStore on load. EpisodeCard consumed that
state via withFreshWatchState but then passed progressAvailable: false to
WatchedIndicator for offline cards, a guard left over from before offline
progress tracking existed. Downloaded movie cards (MediaCard) never had the
guard and already showed the bar.
Drop the guard and the now-unused WatchedIndicator.progressAvailable flag.
close#2236
A movie whose first chapter is titled "Opening Credits" or "Introduction"
got a Skip Intro button that skipped the whole chapter — several minutes
of picture. Detected intros are short; only Plezy's own chapter-title
fallback produced these. Cap chapter-derived intros at three minutes.
Server-supplied markers and credits chapters are unaffected.
close#2235
edde746