Three-cut cluster pulling the last cluster of useEffect bodies + the
DnD-reorder visual feedback out of PlaylistDetail.tsx. 452 → 423 LOC
(−29).
usePlaylistRouteEffects — bundles two route-driven effects:
contextMenu reset (clears contextMenuSongId whenever playerStore's
context menu closes) and openEditMeta-from-route (consumes the
`openEditMeta` location.state flag, opens the meta modal, and
clears the flag with a navigate-replace so back-nav doesn't
re-trigger). Subscribes to playerStore directly.
useBulkPlPickerOutsideClick — the global mousedown listener that
closes the bulk-add-to-playlist picker when clicking outside the
picker wrapper. No-op when closed.
usePlaylistDnDReorder — owns dropTargetIdx state, the
container.addEventListener('psy-drop', …) wiring that calls
runPlaylistReorderDrop, and the handleRowMouseEnter
drag-over-visual helper. Subscribes to DragDropContext directly.
PlaylistDetail drops the direct runPlaylistReorderDrop import and
the contextMenuOpen store subscription — both now live in the hooks
that need them. Pure code move otherwise.
Four-cut cluster pulling the remaining handler bodies + memo island
out of PlaylistDetail.tsx. 507 → 437 LOC (−70).
runPlaylistLoad — async fetch + populate flow that the load
useEffect drives: getPlaylist → filterSongsToActiveLibrary →
setPlaylist + setSongs + setCustomCoverId from playlist.coverArt +
rebuild ratings + starredSongs from each song's stored userRating /
starred flag. Takes the six setters as deps. Page keeps the
useEffect that calls it on id / lastModified change.
usePlaylistPreview — startPreview callback (dispatches into
previewStore with the 'suggestions' source) plus the unmount-time
stopPreview cleanup. Returns just { startPreview }. No props
needed.
usePlaylistBulkPlayCallbacks — wraps playPlaylistAll /
shufflePlaylistAll / enqueuePlaylistAll utilities behind three
useCallback handles with the right deps array. Takes
{ songsLength, id, tracks, touchPlaylist, playTrack, enqueue }.
usePlaylistDerived — existingIds, tracks (songToTrack), sorted +
filtered displayedSongs via getDisplayedSongs, displayedTracks (with
the cheap aliasing optimization when displayedSongs === songs),
isFiltered. Subscribes to playerStore's userRatingOverrides +
starredOverrides directly so the page no longer threads them through
the memo deps.
PlaylistDetail drops these direct imports — they followed the new
modules: getPlaylist, filterSongsToActiveLibrary, songToTrack,
useMemo, getDisplayedSongs (now type-only),
playPlaylistAll/shuffle/enqueue from playlistBulkPlayActions.
Pure code move otherwise.
Four-cut cluster. 572 → 471 LOC (−101). Each handler on the page
collapses to a one-line delegate; pure code move otherwise.
runPlaylistSaveMeta — the meta save flow (updatePlaylistMeta then
optional uploadPlaylistCoverArt + getPlaylist refresh + cover toast,
or coverRemoved → null, then metaSaved toast + closes the modal).
Takes the deps separately from the opts object so the call-site
just passes through the modal's payload.
usePlaylistSongSearch — searchOpen + searchQuery driven debounced
search against subsonic. Owns the 350 ms timeout, the filter-out
of songs already in the playlist, and the searching state. Returns
{ searchResults, setSearchResults, searching } so addSong on the
page can still drop a just-added song out of the result list.
usePlaylistSongMutations — addSong / removeSong. removeSong is
trivial (filter + setSongs + savePlaylist with prevCount).
addSong preserves the .main-content scrollTop save / requestAnimationFrame
restore trick + drops the song out of both suggestions and search
results + fires the add toast with playlist name interpolation.
usePlaylistStarRating — handleRate (local override + playerStore
userRatingOverride + setRating API) + handleToggleStar (e.stopPropagation,
local set + playerStore starredOverride + star/unstar API). Reads
starredOverrides / setStarredOverride from playerStore directly so
the page doesn't have to thread them through.
PlaylistDetail drops these direct imports (now consumed in the
hooks/utility): updatePlaylistMeta, uploadPlaylistCoverArt, search,
setRating, star, unstar, showToast, useRef. Pure code move.
Three-cut cluster pulling tightly-scoped state + memo islands out of
PlaylistDetail.tsx as custom hooks. 659 → 565 LOC (−94). Each hook
returns the same names the page already used, so call sites stay
identical.
usePlaylistCovers(songs, customCoverId) — the 2×2 cover quad memo,
the four cover-quad URLs (with their stable cache keys to avoid the
buildCoverArtUrl-salt re-render loop documented in the comment),
the customCover fetch URL + cache key, and the blurred background
URL going through useCachedUrl. Returns coverQuadUrls /
customCoverFetchUrl / customCoverCacheKey / resolvedBgUrl. The page
no longer imports buildCoverArtUrl / coverArtCacheKey / useCachedUrl
directly.
usePlaylistSelection(songs, setSongs, savePlaylist) — selectedIds /
lastSelectedIdx state + toggleSelect (with shift-range support) +
allSelected + toggleAll + bulkRemove (savePlaylist + setSongs +
clears selection). Returns the same shape the tracklist props
already destructured. savePlaylist moves a few lines up in
PlaylistDetail so the hook can be called with it.
usePlaylistSuggestions(songs, playlist.id) — suggestions state +
loadSuggestions (genre-weighted random pull, top 10) + the auto-load
useEffect that fires on playlist change. Returns setSuggestions too
so addSong on the page can still filter the just-added song out of
the suggestions strip. The page drops the getRandomSongs import.
Pure code move otherwise — no behaviour change.
Seven domain-eng splits peel ~200 LOC of read endpoints out of
`api/subsonic.ts`:
- `subsonicStreamUrl.ts` — `buildStreamUrl`, `coverArtCacheKey`,
`buildCoverArtUrl`, `buildDownloadUrl` (token-signed URL builders
for the four /rest endpoints we hand to the browser).
- `subsonicStarRating.ts` — `getStarred`, `star`, `unstar`,
`setRating`, `probeEntityRatingSupport`. `setRating` still triggers
the lazy `navidromeBrowse` cache invalidation; the same-folder
lazy import path is preserved.
- `subsonicSearch.ts` — `search`, `searchSongsPaged`.
- `subsonicScrobble.ts` — `scrobbleSong`, `reportNowPlaying`,
`getNowPlaying`.
- `subsonicAlbumInfo.ts` — `getAlbumInfo2`.
- `subsonicLyrics.ts` — `getLyricsBySongId`.
- `subsonicGenres.ts` — `getGenres`, `getAlbumsByGenre`.
63 external call sites migrated to direct imports. Four `vi.mock`
targets in the store-level tests pointed at `../api/subsonic` and
were updated to the new module paths.
Pure code-move. subsonic.ts: 762 → 561 LOC (−201).
Three domain-eng modules peel ~316 LOC of fetch/mapping out of
`api/subsonic.ts`:
- `subsonicLibrary.ts` — browse + random + per-song fetch
(`getMusicDirectory`, `getMusicIndexes`, `getMusicFolders`,
`getRandomAlbums`, `getAlbumList`, `getRandomSongs`,
`getRandomSongsFiltered`, `getSong`, `getAlbum`,
`filterSongsToActiveLibrary`, `similarSongsRequestCount`, plus
the private `albumIdsInActiveLibraryScope` cache).
- `subsonicArtists.ts` — artist endpoints (`getArtists`, `getArtist`,
`getArtistInfo`, `getTopSongs`, `getSimilarSongs2`,
`getSimilarSongs`). Uses Library's `filterSongsToActiveLibrary` and
`similarSongsRequestCount` for the per-library scoping fallback.
- `subsonicRatings.ts` — `parseSubsonicEntityStarRating` parser plus
`prefetchArtistUserRatings` and `prefetchAlbumUserRatings` workers
with the shared 7-min cache. Calls back into Library/Artists for
the per-id fetch.
51 external call sites migrated to direct imports. No re-export
shims in `subsonic.ts`. Statistics endpoints still in `subsonic.ts`
keep their `RATING_CACHE_TTL` constant locally (same 7-min window).
subsonic.ts: 1078 → 762 LOC (−316).
First Phase F slice. Splits the 1333-LOC `api/subsonic.ts` along its
two most obvious axes:
- `subsonicTypes.ts` — all ~24 exported interfaces + type aliases
(album/song/artist/playlist/directory/genre/now-playing/radio,
random-songs filters, three statistics shapes, search + starred
results, AlbumInfo, structured-lyrics types, etc.) plus the
`RADIO_PAGE_SIZE` constant.
- `subsonicClient.ts` — token-auth + `getClient` + `api<T>()` +
`libraryFilterParams` + `secureRandomSalt` / `getAuthParams` /
`SUBSONIC_CLIENT`. The credential-bearing API helpers
(`pingWithCredentials`, `apiWithCredentials`, `restBaseFromUrl`,
`probeInstantMixWithCredentials`) stay in `subsonic.ts` for now —
they could move into the client module in a follow-up.
66 external call sites migrated to direct imports from the new
modules (no re-export shims in `subsonic.ts`). Pure code-move;
contract test stays green.
subsonic.ts: 1333 → 1078 LOC (−255).
Migrates ~74 call sites away from the playerStore re-export shims that
were kept during M0–E.41 to avoid touching 30+ imports per PR. Now
that the bigger refactor work is done, each helper goes back to its
real home:
- `initAudioListeners`, `installQueueUndoHotkey`, `flushPlayQueuePosition`
→ from their own store modules
- `getPlaybackProgressSnapshot`, `subscribePlaybackProgress`,
`PlaybackProgressSnapshot` → from `playbackProgress`
- `resolveReplayGainDb`, `shuffleArray`, `songToTrack`
→ from `utils/*`
- `_resetQueueUndoStacksForTest`, `consumePendingQueueListScrollTop`,
`registerQueueListScrollTopReader` → from `queueUndo`
- `PlayerState`, `Track` types → from `playerStoreTypes`
Drops the corresponding 13 re-export stubs from `playerStore.ts` and
the now-unused imports. Also drops dead section banners + per-wrapper
comments above one-line action delegates. Trims one stale "(separate
PR)" note in `transportLightActions.ts` since that follow-up landed
in E.39.
`playerStore.ts`: 180 → 112 LOC (−68). Down from Phase E's starting
3732 LOC.
`bootstrap.test.ts` mock target updated from `../store/playerStore`
to `../store/queueUndoHotkey` to keep the spy reachable after the
import change.
* fix(orbit): event-driven host push on play/pause flips
Without this, the worst-case delay between "host hits pause" and "guest
stops" is two full polling windows (host's 2.5 s push tick + guest's
2.5 s read tick, plus network) — long enough for the guest to noticeably
run past the host. Subscribing to playerStore.isPlaying changes adds at
most one extra remote write per flip; non-flip state ticks still ride
the existing 2.5 s timer. The listener filters on isPlaying so the
per-second currentTime ticks don't trigger spurious pushes.
* fix(orbit): lock seekbar for guests — sync follows the host
Guests could drag/click/wheel the seekbar, which would jump the local
player and then snap back at the next host poll (2.5 s of inconsistent
UX) — or push the guest into a diverged state where Catch Up was the
only way back. The seekbar is host-controlled in Orbit; the guest input
path now reflects that.
- App.tsx exposes `data-orbit-role="host"|"guest"` on the root element
alongside the existing `data-orbit-active` marker.
- WaveformSeek's container gains a `.waveform-seek-container` class so
CSS can target it.
- Guest rule: `pointer-events: none` on children blocks click / drag /
wheel / hover; the parent keeps `cursor: not-allowed` + reduced opacity
so the disabled state is visually unambiguous.
Hosts and non-orbit users see no change.
* docs(changelog): credit PR #537 (orbit sync latency + guest seekbar)
* fix(orbit): make initial-sync seek visually stick on join
Reported: "I join the room, the waveform shows the host's live
position for a second or two, then snaps back to 0:00 and audio is
still playing from the start. Only after that snap can I press Catch
Up."
Two compounding causes in `useOrbitGuest`:
**1. Poll fires `applyMirror` before the engine is genuinely playing.**
`playTrack` flips `isPlaying` to `true` *synchronously* in its
optimistic store write, so the post-`playTrack` poll satisfied its
"engine ready" check before the Tauri `audio_play` had even
started producing sample. The `seek` inside `applyMirror` updates
the store position immediately (waveform jumps to host's live
position), but `audio_seek` is debounced and lands on a not-ready
engine — it silently no-ops. The engine then starts playing from
0, and its first progress events overwrite the optimistic seek
position, snapping the waveform back. Add `currentTime > 0.1` to
the poll's "engine genuinely playing" condition: once audio has
flowed past the cold-start barrier, the seek commits and the
seek-target guard correctly filters subsequent progress events.
**2. `applyMirror`'s play-state mirror raced its seek**, same shape
as the `onCatchUp` race fixed in #527. `player.seek` debounces
`audio_seek` via setTimeout(0) while `pause`/`resume` invoke
synchronously — pause arriving first leaves the engine paused at
the old position. Defer the play-state mirror by 200 ms so the
seek lands first.
* docs(changelog): add initial-sync seek-stickiness bullet
* fix(orbit): kick a fresh playTrack when engine is stuck mid-load
Follow-up to #525. Symptom: occasionally on join, the guest gets no
audio until the next host-driven track change, at which point a
fresh `playTrack` runs and audio plays normally.
Root cause: when the initial `syncToHost` poll hits its 5 s deadline
without the engine reporting `isPlaying === true` (slow Navidrome
cold-start), the next pull tick takes the cheap "track already loaded"
shortcut and calls `applyMirror`. `applyMirror` fires `seek` + `resume`
on an engine that is stuck in a "loaded but never started" limbo —
seek silently no-ops and `resume` can't kick a track that never began.
Guest is silent until something else triggers a fresh `playTrack`.
Tighten the shortcut: take the cheap path only when the engine is
already in the state the host expects (or playing while host is
paused, which is fine to align via pause). Otherwise fall through to a
fresh `playTrack` and the existing 5 s ready-poll, which re-initialises
the engine and lets audio actually start.
* docs(changelog): add engine-limbo follow-up bullet
* fix(orbit): re-sync when engine silently fell back to paused
The optimistic `isPlaying: true` `playTrack` writes synchronously
masks a `audio_play` failure: the post-playTrack poll sees the
optimistic flag, fires `applyMirror`, and the outer tick records
`lastAppliedRef = { ..., isPlaying: true }` as a successful sync.
But if the underlying `invoke('audio_play')` rejects later (network
blip / cold-start exhaustion), the catch handler flips the flag back
to `false` and schedules `next()`, which short-circuits to `audio_stop`
in an Orbit guest — leaving the player silent while `lastAppliedRef`
still claims we're playing. None of the divergence-detection branches
(track-change / play-pause-flip) match, so the guest never re-syncs.
Add a recovery check before the if/else-if chain: when the captured
`last` says we applied playing, the engine is currently not playing,
and the host is still playing the same track, reset
`lastAppliedRef.current = null`. The next iteration of the chain
re-runs initial-sync (with the 500 ms fast-poll cadence) which fires
a fresh `playTrack`. If `audio_play` succeeds the second time, audio
finally starts; if it keeps failing, we loop with no extra harm
(audio was already silent).
Adds an `engine-recovery` event scope to the diagnostics buffer so
future captures can see when this re-sync fired.
* docs(changelog): expand engine-recovery bullet to cover both cases
* fix(orbit): guest short-circuits queue-exhaustion fallback paths
When a guest's local queue runs out (single-track queue from `syncToHost`
empties on `audio:ended`), the player walks the standard fallback chain
in `next()`: radio top-up → infinite-queue → stop. The infinite-queue
branch builds a 6-track queue and calls `playTrack`, which trips
`orbitBulkGuard` and pops a "Add 6 tracks to the Orbit queue?" modal.
Hitting Cancel leaves playback frozen; "Add them all" injects unrelated
tracks into the host's shared queue.
In an active Orbit guest session the host owns the queue. Skip the
fallback paths entirely and just stop — the next `useOrbitGuest` pull
tick will sync to whatever the host advanced to.
Bonus side-effect: kills the deferred-promise race where a
`buildInfiniteQueueCandidates().then(...)` from a guest's track end
could resolve *after* a Catch Up replaced the queue and pop the modal
a second time against the now-current 1-track queue.
* fix(orbit): treat natural track-end as not-diverged in guest sync
When a guest's track ended naturally before the host advanced, the
divergence-detection branch read `player.isPlaying === false` and
classified it as the user manually paused — so it refused to load the
host's next track. The guest sat silent until they clicked Catch Up.
`handleAudioEnded` keeps `currentTrack` pinned to the just-ended track
and resets `currentTime` to 0, while a real manual pause leaves
`currentTime` somewhere mid-track. Use the 0-position discriminator to
classify natural-end as not-diverged so the host's new track loads.
Confirmed via the captured guest log buffer:
18:43:08.598 [track-change] host: VJkV5… → 6i6RP… BUT guest diverged
(player.isPlaying=false ≠ last.isPlaying=true)
— guest stuck for ~33s until Catch Up was pressed.
* fix(orbit): Catch Up polls until engine is ready before seeking
The 400 ms blind setTimeout in `onCatchUp` was too short for an
HTTP-streamed cold-start on high-latency links. If the audio engine
wasn't ready by then, `seek(fraction)` silently no-oped and playback
started at 0:00, making Catch Up effectively useless on exactly the
slow links where it's needed. Captured log shows a Catch Up bringing
the guest to posSec=30, then 5 s later the guest was at posSec=6
(playback restarted from the head).
Replace with the same poll-until-ready pattern `syncToHost` already
uses: check every 100 ms, fire the seek as soon as the engine reports
playing, fall back to a blind apply at the 4 s deadline.
* docs(changelog): add orbit guest playback fixes entry
* fix(orbit): debounce Catch Up button + match bar item height
Two follow-on UX fixes after PR #525's three primary bugs landed:
1. **Debounce visibility.** Drift is computed from an asymmetric signal:
guest's `currentTime` updates in coarse ~5 s chunks, while host's
position is extrapolated linearly via `(nowMs - posAt)`. Even on a
perfectly-synced session the diff swings ±5 s every tick, so the
button flickered in and out continuously. Show only after drift has
stayed over the 3 s threshold for ≥ 3 s of wall clock — measurement
noise is filtered out, real sustained drift still surfaces in time.
2. **Match neighbour height.** The button was 32 px tall against 26 px
for the other action buttons (.orbit-bar__settings) so every flicker
shifted the entire bar height. Set `height: 26 px` and tighten the
padding/font so the layout is stable regardless of visibility.
* fix(orbit): tighten queue-extension lockout + reliable initial-sync seek
Two follow-on fixes after the 4-bug umbrella:
**1. Local queue-extension paths fully off during Orbit.**
Phase check broadened from `active` to cover `starting` / `joining` /
`active` so a fetch-then-join race can't pop the bulk-add modal *after*
the join. The proactive infinite-queue topper inside `next()` (which
fires when ≤ 2 auto-tracks remain ahead) is now also gated, plus each
async `.then()` callback in the radio + infinite-queue paths re-checks
at resolution time. A `playTrack(... 6-track queue ...)` after the user
joined Orbit was the path that re-triggered the "Add 5 tracks?" modal
on a freshly-joined guest.
**2. `syncToHost` only seeks once the engine reports playing.**
The previous 2 s deadline-fallback applied the seek even when the
engine hadn't started, where the seek silently no-ops and the track
plays from 0:00. Symptom: clicking Catch Up makes the song "jump 50 %
forward" — that's the seek finally landing because the engine is now
ready, the initial-sync seek had already failed silently. New deadline
is 5 s, and on timeout we return `false` so the outer pull tick keeps
`lastAppliedRef` null and the 500 ms fast-poll retries.
* fix(orbit): double-click play button + hide preview during session
Two cucadmuh-flagged gaps:
**1. Double-click on the inline play button now reaches the orbit-add
path.** The album-track row's onDoubleClick already routes to
`addTrackToOrbit` when in Orbit, but the inline play button stopped
propagation on click — so clicking it twice just fired the "double-
click to add" hint toast and never touched the orbit queue. Add an
onDoubleClick on the button itself that delegates to the parent's
`onDoubleClickSong`.
**2. Track preview is suppressed during an Orbit session.** Preview
shares the Rust audio engine with the shared playback, so starting
one as a guest yanks the host's track out from under everyone. A new
`[data-orbit-active]` attribute on `<html>` (set whenever role is
host/guest and phase is starting/joining/active) hides every
preview button via a single CSS rule, and `previewStore.startPreview`
short-circuits as a defensive guard for keyboard shortcuts and any
programmatic callers.
* feat(orbit): in-app diagnostics popover with copyable event log
Multiple users on Discord report Orbit guests stopping after the first
song with no errors anywhere — Settings → Debug → Export Logs is too
buried for non-technical reporters, and the relevant code branches
have no logging at all (silent fail). This adds a one-click "Copy log"
path right inside the Orbit session bar.
The new Activity-icon button next to Help opens a popover with:
- Live mini-display: role, host vs. guest track id + position, drift,
age of the host's last state write — all updating once a second.
- Scrolling event log textarea fed by an in-memory ring (200 events).
- Copy + Clear buttons. Copy formats `[ISO] [scope] body` lines and
drops them on the clipboard — paste straight into a Discord report.
Instrumentation lands at the previously-silent decision points:
- Guest pull tick: full snapshot of host vs. guest state on every read.
- Each branch of the divergence detection in `useOrbitGuest.ts` logs
which path it took and why (initial / track-change-followed /
track-change-diverged / play-pause-flip), making the
"stuck after first song" symptom diagnosable from the buffer alone.
- Host pushes log track id, isPlaying, queue length, guest count.
Events are also bridged to the existing `frontend_debug_log` Tauri
command when Settings → Logging is on Debug, so power users still get
the same data in `psysonic-logs-*.log` for offline triage.
i18n: full `orbit.diag.*` namespace in all eight locales. EN + DE are
native; ES / FR / NB / NL / RU / ZH are first-pass and may want a
polish from native speakers later.
* docs(changelog): add orbit diagnostics popover entry
* feat(selection): add useRangeSelection hook with Shift+Click range support
Reusable hook for multi-select state across pages that show grids of
items the user can pick. Tracks the selected ID set, a click anchor,
and exposes a `toggleSelect(id, { shiftKey })` callback:
- Plain click → toggles that item and moves the anchor to it.
- Shift-click on a second item → adds every item between the anchor
and the click target (inclusive) to the selection. The anchor moves
to the shift-clicked item so the next shift-click extends from
there.
Range expansion follows the items array passed to the hook, so the
caller controls the user-visible order (filtered + sorted list, not
the raw upstream array).
Implementation note: the anchor ref is snapshotted *before* the state
updater runs and written *after* it. React 18 strict mode invokes
state updater functions twice in dev to surface side effects, so any
ref mutation inside the updater would taint the second invocation and
the replay would miss the range branch.
* feat(selection): adopt Shift+Click range selection on grid pages
Wires `useRangeSelection` into the four pages that ship a multi-select
mode on top of card grids:
- Albums (passes the filtered/sorted `visibleAlbums` so range follows
the order the user actually sees)
- RandomAlbums
- NewReleases
- Playlists
`AlbumCard.onToggleSelect` is extended to forward `{ shiftKey }`, and
the card's onClick handler reads `e.shiftKey` from the React event
and threads it through. The Playlists grid uses an inline onClick on
the card div and was updated the same way.
User-visible behaviour: in selection mode, click an item then
shift-click a later item — every item between them gets selected.
Existing single-toggle behaviour is unchanged when no shift key is
held.
* docs: changelog entry for PR #484
Logs the Shift+Click range selection on grid pages under
v1.46.0 "## Changed".
* Enhance CachedImage and ArtistDetail components with improved image caching and priority handling
- Refactor CachedImage to utilize a priority system for image loading based on viewport visibility, improving performance during scrolling.
- Update useCachedUrl to accept an optional getPriority function for better cache management.
- Optimize ArtistDetail and Artists components by using useMemo for cover art URLs, reducing redundant calculations and improving rendering efficiency.
- Adjust image loading logic in CachedImage to ensure smoother transitions and avoid unnecessary fetch requests.
* perf(ui): unblock IDB cover art, stabilize mainstage rails and virtual lists
Let IndexedDB reads bypass the network concurrency slot so cached thumbnails
paint without queueing behind remote fetches; debounce disk eviction during
heavy scrolling.
Fix mainstage horizontal rails: dedupe album/song ids for React keys, widen
artwork budget overscan, avoid resetting the budget on list append, and raise
Home initial artwork budgets. CachedImage treats already-decoded images as
loaded; rail cards load cover images eagerly.
Refresh dynamic color extraction and extend virtual scrolling / scroll roots on
Albums, Artists, Playlists, and related surfaces.
Remove local agent-only commit instructions from the repository tree.
* perf(virtual): viewport-based overscan for main scroll lists
Drive TanStack Virtual overscan from measured scroll height so each list
renders about one screen of extra rows above and below the viewport for
snappier scrolling on Albums, Artists (list mode), and Tracks virtual song list.
Introduce useResizeClientHeight helpers (ID + ref) for ResizeObserver-based
clientHeight tracking.
* docs(changelog): note PR #468 UI cover cache, rails, and virtual lists
Add a coarse summary under 1.46.0 Changed for cover-art pipeline,
mainstage rails, viewport-based overscan, and library/chrome polish.
* feat(linux): optional native GDK for Nix gdk-session
Introduce PSYSONIC_ALLOW_NATIVE_GDK so main skips the default GDK_BACKEND=x11
pin when the Nix gdk-session wrapper sets the flag. Remove GDK_BACKEND from
the npm tauri:dev script so it does not override nix develop defaults.
* fix(ui): portal server switch menu above sidebar
Main column stacks below the sidebar (layout z-index), so an in-tree dropdown
could never win over the left nav. Render the menu via createPortal to
document.body with fixed coordinates, matching the library scope picker.
* feat(perf): add mainstage probe controls and cut WebKit repaint load
Add a dedicated performance probe surface for mainstage/home toggles and wire Linux CPU diagnostics to isolate expensive UI paths. Tune waveform drawing and Home artwork clipping/windowing so visible content loads immediately while reducing WebKit compositor pressure during playback.
* fix(perf): stop hero rotation when section is off-screen
Gate hero auto-rotation and backdrop crossfade by real viewport visibility using the actual scrolling ancestor. This prevents periodic 10-second CPU spikes from hidden hero updates while preserving normal behavior when the hero is visible.
* fix(perf): isolate player progress updates from mainstage diagnostics
Add probe toggles for PlayerBar waveform and live progress UI updates to confirm playback progress churn as the main CPU driver.
Restore Home artwork quality defaults and keep visual-degradation modes opt-in via debug flags only.
* fix(hero): resume background and autoplay after viewport return
Re-check hero visibility on focus/visibility changes and add a short recovery poll while off-screen so missed scroll/RAF events cannot leave hero animation paused.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(perf): decouple playback progress from mainstage compositing pressure
Throttle audio progress delivery and route live seekbar timing through a lightweight progress channel to cut focus-time WebKit CPU spikes.
Add focused diagnostics in Performance Probe and restore hero/waveform behavior so visuals remain stable while profiling.
* fix(debug): open performance probe with Ctrl+Shift+D
Replace logo-triggered opening with a keyboard shortcut and keep logo purely decorative to avoid accidental probe activation.
* docs(changelog): document experiment/performance probe and playback work
Add an [Unreleased] section for the performance probe, throttled audio
progress IPC, snapshot-based live UI updates, WaveformSeek scheduling
over the same canvas bar renderer, Hero/Home rail fixes, and Linux/Nix
GDK dev ergonomics.
* perf(linux): add WebKit probe, throttle progress IPC, snapshot playback UI
Ship Performance Probe (Ctrl+Shift+D), Rust-throttled audio:progress, a
playback progress snapshot channel with coarse Zustand timeline commits,
Linux /proc CPU readout for the probe, Hero and Home rail artwork fixes,
Tracks SongRail windowing parity, MPRIS cleanup, gated perf counters, and
WaveformSeek paused-seek correctness. Documented in CHANGELOG for PR #452.
* docs(changelog): fold perf work into 1.45.0 and refresh date
Drop the separate 1.45.1 heading; keep PR #452 notes under 1.45.0 Added and
set the section date to 2026-05-04. Restore the safety preface before the
versioned sections.
* docs(changelog): order 1.45.0 Added entries by PR number
Sort the 1.45.0 release notes so subsections follow ascending PR id (390
through 452), with PR #452 last.
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
The pending counter desynced from the actual approval list (state.queue
holds approved items as history, so the count never decreased after a
host approve). The host-pushed pendingApprovalCount workaround didn't
hold up under live testing either, so we're rolling the whole cap
feature back rather than ship something flaky.
What's gone:
- OrbitSettings.maxPending + state.pendingApprovalCount
- cap branch in applyOutboxSnapshotsToState (now back to mute-only)
- maxPending number input in settings popover
- pending counter chip in OrbitQueueHead
- 'cap-reached' branch in evaluateOrbitSuggestGate / OrbitSuggestGateReason
- cap-related toasts in ContextMenu / useOrbitSongRowBehavior
- cap-related i18n keys (suggestBlockedCap, settingMaxPending*, pendingCounter*)
- cap CSS (.orbit-queue-head__pending, .orbit-settings-pop__number)
What stays: per-guest suggestion mute (works correctly) and everything
that fed into both features (OrbitState.suggestionBlocked,
setOrbitSuggestionBlocked, evaluateOrbitSuggestGate, the participants
popover Mic/MicOff toggle, the suggestBlockedMuted toast).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The "X / Y pending" counter in the queue head and the guest-side
gate-check both used `state.queue.filter(non-host).length`, which is the
*history* count — items the host has already approved or declined still
sit in `state.queue` for attribution lookup, so the counter never
decreased. Reported as "3 / 4 pending" with no actual rows in the
approval list.
The merged / declined sets only exist in the host's local store, so the
guest can't filter them out itself. Solution: the host writes an
authoritative `pendingApprovalCount` into the state blob each tick;
guests (and the host's own UI) read it directly, with a fallback to the
old over-counting behaviour for any older client that doesn't write the
field.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two anti-spam knobs the host can dial during a live session:
1. Per-guest suggestion mute — Mic / MicOff toggle next to the
kick/ban buttons in the participants popover. Symmetric (re-enable
later). State lives in OrbitState.suggestionBlocked: string[]; the
guest reads it and disables its own Suggest controls so the user
sees a clear "muted" state instead of silent failures. Host-side
sweep also drops their outbox entries as a safety net.
2. Max pending approvals cap — number input in the session-settings
popover, default 0 (= unlimited so existing sessions are unaffected).
When set, the host sweep stops folding new outbox entries into the
approval list once the cap is reached. The OrbitQueueHead surfaces
"X / Y pending" so guests can see when they're getting close.
State changes are additive on the wire — both fields are optional, with
parseOrbitState defaulting them, so older clients keep working.
evaluateOrbitSuggestGate() centralises the guest-side allow/block check
shared between useOrbitSongRowBehavior and the ContextMenu Add-to-Session
items, plus suggestOrbitTrack as a defensive last line.
i18n: en + de + fr + nl + zh + nb + ru + es.
Also fixes an earlier dedupe-key collision: the (user, trackId) cache
keys were missing their separator (NULL byte slipped in during the
previous patch), so two tracks could share a key and one of them
silently overwrite the other. Restored the space separator.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Type `window.__psyHidden` once via global declaration; drop 6 `(window as any)` casts.
- `WindowVisibilityProvider` now ORs `window.__psyHidden` into the hidden state
(and into the initial state). On WebView2 `document.hidden` does not always
flip when `win.hide()` is called, so effects gated by `useWindowVisibility()`
can now actually skip creating intervals/rAF instead of relying solely on the
per-tick fallback check.
- Slow visible-poll from 200 ms to 500 ms (5 → 2 wakeups/s for visibility alone).
- Fix stray leading-space indent in `App.tsx` `window:close-requested` comment.
Co-authored-by: Psychotoxical <dev@psysonic.app>
* fix(windows): stop GPU rendering when windows are hidden
* perf(ui): tighten hidden-window rendering mitigation
Align implementation and documentation with what the injected pause/resume
scripts actually do, reduce main-thread and compositor wakeups while windows
are invisible, and make lifecycle cleanup reliable.
Rust (Tauri)
- Document PAUSE/RESUME as compositor-oriented hints: set __psyHidden, optional
--psy-anim-speed for CSS that opts in, and pause @keyframes via
animation-play-state only (not CSS transitions, not arbitrary timers or rAF).
- Walk descendants under #root instead of document-wide querySelectorAll('*')
to cut cost on large pages; skip the walk if #root is missing (flag still set).
- Call eval(PAUSE_RENDERING_JS) before hide() on all hide paths (tray menu,
tray icon toggle, mini open/close/show-main, native mini close) so the
webview sees __psyHidden while still fully schedulable.
- Shorten pause_rendering command rustdoc to match the script.
Frontend
- WindowVisibilityProvider: replace self-rescheduling useCallback with a single
effect that keeps one pending timeout id, uses a cancelled flag on unmount,
and syncs hiddenRef from document.hidden at mount (no orphaned timers after
tests/HMR).
- WaveformSeek (preview + animated seekbar): while hidden, poll with a 400 ms
timeout instead of requestAnimationFrame every frame; cancel both rAF and
timeout on teardown.
Why
- Accurate comments prevent future refactors from assuming “full JS idle”.
- Smaller DOM walks and fewer rAF wakeups reduce spikes when hiding the main
window or mini player on Windows WebView2 and elsewhere.
* perf(ui): pause global CSS + timers when Tauri hides the window
Builds on PR #273 (`7803d8e` + `0e07a73`). Adds a second HTML flag driven only by
Rust pause/resume inject so infinite animations still pause when WebView2 keeps
`document.hidden === false` after `win.hide()`, and stops a few periodic JS
timers while the window is invisible.
- `lib.rs`: set/remove `data-psy-native-hidden` on `<html>` in PAUSE/RESUME JS;
document why in rustdoc (rest of pause/resume unchanged from PR tip).
- `components.css`: same `animation-play-state: paused !important` rules as
`data-app-hidden` for `[data-psy-native-hidden="true"]`; refresh the comment.
- `App.tsx`: comment distinguishes tab visibility vs Tauri-native hide.
- `Hero.tsx`: do not run the 10 s carousel interval while hidden
(`useWindowVisibility` + `__psyHidden` in tick).
- `PlaybackScheduleBadge.tsx`, `playbackScheduleFormat.ts`: no 400/500 ms
intervals while hidden; skip ticks when `document.hidden` or `__psyHidden`.
- Drop the ~110 lines of dead full-screen overlay CSS that was never
referenced in JSX (.lucky-mix-overlay*, .lucky-mix-cube*, full-screen
@keyframes). Keep the .lucky-mix-pip* rules since the inline queue
variant re-uses them.
- Extract availability gate into hooks/useLuckyMixAvailable.ts (pure
predicate + hook). Replaces five duplicated inline checks in Sidebar,
MobileMoreOverlay, RandomLanding, Settings/SidebarCustomizer, and the
internal re-check in buildAndPlayLuckyMix.
- Add a cancel mechanic:
* luckyMixStore gets a cancelRequested flag + cancel() method
* LuckyMixCancelled sentinel is thrown from a bailIfCancelled() helper
sprinkled between await boundaries in the build loop
* catch-block swallows the sentinel silently (no toast, no error state)
* auto-cancel subscription on playerStore detects manual user track
changes (anything not in queuedIds) and flips cancelRequested so the
finished mix does not later overwrite the user's choice
* inline .queue-lucky-loading dice indicator is now a button — click
triggers explicit cancel, hover fades the dice to signal the action
- Restore the queue snapshot when the build fails before ever starting
playback, so the user does not land in an empty player after an error.
If playTrack already ran, leave current state alone (their track plus
whatever was enqueued is more useful than a stale pre-click queue).
- Rework locale labels to consistently carry the "Mix" suffix so the
entry reads well next to "Random Mix" / "Random Albums" in the
sidebar: DE Glücks-Mix, ES Mezcla Suerte, FR Mix Chance, NB Lykkemiks,
NL Geluksmix, ZH 好运混音. EN stays "Lucky Mix", RU stays
«Мне повезёт». Toast/settings/randomNavSplit copy aligned to the new
names. New luckyMix.cancelTooltip key added in all 8 locales.
AudioMuse gating stays — as Cuca confirmed, the feature's quality relies
on a correct similar-track selection, so the requirement is intentional.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Guest suggestions now auto-merge into the host's play queue at random
positions inside the upcoming range, so they actually surface alongside
host-picked tracks instead of piling up at the end. Per-item dedupe via
addedBy:addedAt:trackId keys survives reshuffles.
Guests get a read-only mirror of the session in place of their queue
panel — live-card for the host's current track, suggestion list with
submitter attribution, and a footer reminding them the host owns
playback. Track metadata is resolved lazily and cached locally.
New host-only settings popover (gear in the bar) with auto-approve and
auto-shuffle toggles plus a "Shuffle now" action. Settings live in the
OrbitState blob and push immediately to Navidrome; missing fields on
older blobs default to "both on".
15-min shuffle now touches the real playerStore queue via a new
shuffleUpcomingQueue action — previous behaviour only reshuffled the
guest-facing suggestion history. A manual shuffle is available from the
settings popover so the interval can be verified without waiting.
Start-modal reworked into a single step: share-link is visible and
copy-able from the start, auto-copied on Start if the host hasn't
already hit Copy. Link carries a live name-derived slug (parser strips
it, SID is authoritative). LAN/remote-server hint above the facts.
Random session-name suggester pulls from three pattern pools for ~10k
unique combinations.
All user-visible Orbit strings moved to a dedicated orbit i18n
namespace (en + de); other locales fall back to en via i18next.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add scheduled pause and resume timers in the player store, cleared on stop
and track changes. Long-press opens a compact preset modal anchored above
the transport row; one-tap presets plus custom minutes. Portaled countdown
badge on the play button; clear the long-press click guard when the modal
closes so the first play/pause click works after scheduling.
* feat(servers): magic string invites and duplicate-aware labels
Add psysonic1- share payloads (URL, Subsonic credentials, optional server
name). Login and add-server forms decode on input and keep magic field
below standard fields. Navidrome User Management generates strings after
PUT password updates where required. Resolve duplicate display names as
username@host in chrome, settings, and login.
* feat(servers): tighten magic-string import and admin share flow
After a decoded magic string, username is read-only and the password field
shows a fixed-length mask so length is not inferred. Password reveal stays
disabled until saved-server quick connect (login) or reopening the add-
server form. Navidrome admin share UI persists the password via API before
copy and adds a plaintext-handling disclaimer. Locales updated.
* feat(settings): save new Navidrome user and copy magic string
Add a non-admin "Save and get magic string" flow: create the user, assign
libraries when needed, then encode credentials to the clipboard. Reuse
the plaintext-handling disclaimer without the password-update hint meant
for edits. Strings added for all locales.
* fix(settings): tighten Navidrome add-user validation and submit feedback
Require username, display name, and a non-empty password (trimmed) for new
users; gate Save and “save and get magic string” on explicit checks with
toast feedback. Show red borders only after a failed submit, and clear
them when the user fills the fields. When editing, keep an empty password
as “unchanged” and validate identity fields with a dedicated message.
Strings: userMgmtValidationMissingIdentity across locales.
The existing module-level Map<songId, CachedLyrics> only survived
within a session. After every app restart the lyrics fetch chain
(server → lrclib → netease) ran from scratch for every track the user
played, even ones whose lyrics were already resolved 5 minutes ago
before the previous session ended.
Add an IndexedDB layer that's only consulted on RAM miss:
- L1 (RAM): unchanged, sync, hit on tab switch / track repeat
- L2 (IndexedDB, async): hit after restart, hydrates RAM
- Network: unchanged fallback chain
Key format `${serverId}:${songId}` so the same songId on two different
Subsonic servers cannot collide.
TTLs: 90 days for resolved lyrics (they rarely change once shipped),
7 days for `notFound` entries (gives the user / admin a chance to add
lyrics later without an indefinite negative cache).
YouLyPlus mode skips L2 — a persisted entry from the standard pipeline
wouldn't have word-level sync, which is the whole point of lyricsplus.
Co-authored-by: Psychotoxical <dev@psysonic.app>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two independent user-facing fixes rolled together.
── Lyrics: YouLyPlus provider (issue #172) ──
Adds word-by-word synced (karaoke) lyrics as an alternative lyrics mode.
Backed by the public lyricsplus aggregator (Apple Music / Spotify /
Musixmatch / QQ Music) — no API keys on our side, subscription costs are
borne by the backend operator. Five mirrors are tried on network failure.
Settings → Lyrics now exposes a mode radio (Standard vs YouLyPlus) plus
a static-only toggle. YouLyPlus misses silently fall back to the existing
server + LRCLIB + Netease pipeline so obscure tracks still resolve. The
drag-to-reorder source list is hidden in YouLyPlus mode since the
external aggregator manages its own source priority.
Rendering: per-word spans on each line, active word highlighted with
accent-tinted glow. Subscribed imperatively via usePlayerStore.subscribe
so 500 ms progress ticks do not re-render the whole lyrics block. The
Fullscreen Player reuses the same pattern; the 5-line rail layout is
unchanged. white-space: pre on .fs-lyric-word preserves the trailing
spaces the API embeds in each syllabus token (flex context would
otherwise collapse them).
i18n: new keys in all 8 locales (de, en, fr, nl, nb, ru, es, zh).
── macOS mic-prompt suppression ──
Ships a vendored cpal 0.15.3 at patches/cpal-0.15.3/ wired via
[patch.crates-io]. The only change is in src/host/coreaudio/macos/mod.rs:
audio_unit_from_device now unconditionally uses IOType::DefaultOutput for
output streams instead of conditionally choosing HalOutput.
AUHAL (HalOutput) is classified as input-capable by macOS TCC and
triggers the microphone-permission dialog at first launch / after each
update even for playback-only apps. Previous attempts (removing
NSMicrophoneUsageDescription, setting com.apple.security.device.audio-input
false in Entitlements.plist) did not suppress the prompt because TCC
fires at AudioUnit instantiation, not at plist level. The upstream fix
in cpal PR #1070 / cpal 0.17 only helps when the pinned device equals
the system default; always-DefaultOutput covers all cases.
Tradeoff: per-device output selection is a no-op on macOS — the stream
always follows the system default. Settings.tsx surfaces this via
audioOutputDeviceMacNotice (hides the CustomSelect on macOS) and skips
audio_list_devices + device-watcher effects there. Matches the behaviour
of Apple Music / Spotify on macOS.
Also removes the now-obsolete com.apple.security.device.audio-input
entitlement (sandbox is disabled anyway, so it was ignored).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lyrics Sources:
- Replace lyricsServerFirst + enableNeteaselyrics toggles with a new
Settings section — three sources (Server, LRCLIB, Netease) each
individually toggled and drag-reorderable via psy-drop DnD
- useLyrics.ts iterates sources in user-defined order, skipping disabled
ones; embedded SYLT from local files still wins unconditionally
- onRehydrateStorage migration from legacy fields on first load
ReplayGain Pre-Gain:
- New authStore fields: replayGainPreGainDb (0…+6 dB) and
replayGainFallbackDb (-6…0 dB, untagged / radio)
- audio.rs compute_gain applies pre_gain_db and fallback_db
- Settings sliders under ReplayGain mode selector
- Radio HTML5 volume scaled by fallbackDb factor on playRadio
i18n(ru): incorporate PR #148 translation improvements (kilyabin)
All 7 locales updated for new keys.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Filter lines matching 作词/作曲/编曲/制作人/etc. that Netease
embeds as timestamped LRC entries at the start of the song.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Netease is queried only when server and LRCLIB both return nothing,
preserving the existing lyrics chain completely. Off by default.
- Rust command `fetch_netease_lyrics` proxies Netease API (CORS bypass)
- `src/api/netease.ts` TypeScript wrapper via invoke
- `authStore.enableNeteaselyrics` toggle (default: false)
- `useLyrics`: Netease fires last in fallback chain when enabled
- Settings toggle in the Lyrics section
- `lyricsSourceNetease` label in LyricsPane
- i18n: all 7 languages (en, de, fr, nl, nb, ru, zh)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Parse type and serverVersion from Subsonic ping; persist per-server identity and
run a background Instant Mix probe (random songs + getSimilarSongs) for
Navidrome 0.60+. Hide the AudioMuse server toggle when the probe returns no
similar tracks; clear prefs when the server is no longer eligible.
Artist pages fall back to Last.fm similar artists when AudioMuse is enabled but
the server returns none. Instant Mix failures set a per-server issue flag with a
settings warning and toast. Settings description links the official plugin repo
via i18n Trans.
Frank Stellmacher
cucadmuh
Maxim Isaev
Frank Stellmacher
Ivan Pelipenko
Psychotoxical