The lazy source-build commit (1ddde12) introduced a race: when the
user taps play on a new track while a previous queue's
_fillRemainingSources is still working, the stale fill keeps
calling _player.addAudioSource() on the new player state — appending
old-playlist tracks into the new queue and confusing the player into
the "locked to one song" symptom.
Fix: queue-generation counter. setQueueFromTracks bumps
_queueGeneration first thing; the background fill captures its gen
at start and aborts before any further player mutation if a newer
queue has taken over. The previous play() never gets to mutate the
new player state.
Also resets _suppressIndexUpdates at the start of every
setQueueFromTracks (defensive — covers the case where a prior
backward-fill bailed on its gen check before reaching `finally`)
and only releases the flag in finally if we're still the active
gen.
Symptoms this should resolve:
- "locked to one song" after rapid play taps
- Late `play() returned 73189ms` lines indicating a previous
hung play() call finally resolving and stepping on current state
- Player stuck in odd processingState after queue swaps
Same pattern as album + artist detail. PlaylistDetailScreen accepts
optional Playlist seed via go_router extra. While
playlistDetailProvider is still resolving, render the header
(description if any + track count) plus a "loading tracks…" inline
spinner instead of an opaque full-screen CircularProgressIndicator.
The body fills in when tracks arrive via drift watch re-emit.
Routing reads s.extra as Playlist. PlaylistCard +
PlaylistsListScreen pass the playlist via extra. Surfaces with no
seed available (deep links etc.) fall back to the original full-
screen spinner.
Track row rendering already uses ListView.builder so visible row
count was never the bottleneck — the wait was for the detail fetch
itself. Header-on-tap is the win that makes the screen feel snappy.
Two unrelated wins as a single batch.
Flutter — lazy source building in setQueueFromTracks:
Today: Future.wait builds all N AudioSource objects (drift queries +
LockCaching ctor) before the player can call setAudioSources →
play(). Measured at 83ms for a 25-track playlist, on top of the
~285ms initial-source preload.
New flow: build only the initial source, hand it to setAudioSources
([initial], initialIndex: 0) so play() can start, then background-
fill the rest. Forward direction (skipNext targets) added via
addAudioSource. Backward direction (skipPrev) inserted at index 0..
initialIndex-1 with _suppressIndexUpdates true so the unavoidable
currentIndex shifts don't push the wrong MediaItem onto the stream.
Saves the up-front source-build wait — tap-to-audio for long queues
should drop by ~80-100ms even on cache hits.
Server — Cache-Control on the three byte-serving endpoints:
- /api/albums/{id}/cover: max-age=86400, must-revalidate. Covers
change rarely (re-scan, MBID enrichment); a day of cache is safe
and skips conditional GETs for the bulk of a session.
- /api/playlists/{id}/cover: max-age=300, must-revalidate. Collages
recompute when contents change; short enough for edits to feel
fresh, long enough to skip repeat fetches during a session.
- /api/tracks/{id}/stream: max-age=31536000, immutable. Track bytes
are immutable for a given id (scanner re-indexes by file_path; new
files get new ids). LockCachingAudioSource on the Flutter side
already disk-caches, but proper headers let it skip even the
conditional 304 on repeat plays.
Logs were spitting "RenderFlex overflowed by 1.00 pixels on the
bottom" from album_card.dart whenever the library Albums tab or
artist detail album grid rendered. Cell height was computed as
cover + gap + title + artist + 4px slack, which assumes pixel-perfect
14sp/12sp line heights — Flutter's actual rendering with ascent/
descent + line-height multipliers wants one more pixel.
Bump slack from 4 to 8 in both grids. AlbumCard layout unchanged;
the warnings stop.
Otherwise the log shape is healthy now: prefetcher fires once per
library page emit (expected, one batch per pagination), cache misses
fetch sequentially with clean drift-write → re-emit cycles, and
album taps are single-round-trip cold-fetches followed by drift hits.
No more cycles of duplicate fetches.
Same fix shape as albums: drift's CachedPlaylistAdapter.toRef was
returning coverUrl: '' because the cache table doesn't persist the
server-derived URL. Set it to /api/playlists/<id>/cover in the
adapter — handleGetPlaylistCover serves the cached collage from
disk, so the URL is deterministic and the round-trip through drift
no longer drops it.
PlaylistCard + PlaylistsListScreen pass the existing queue_music
icon as ServerImage's fallback, so when the server hasn't built a
collage yet (system playlists with no tracks at build time), the
endpoint 404s and the icon shows over the slate background instead
of an empty box.
Tap→audio measured at ~370ms — that path's fast. Real symptom: a few
seconds of audio, then silence with no event surfaced to Flutter.
ExoPlayer is failing/completing somewhere and we have no log to act
on.
Three changes, all log-only:
- Add onError to playbackEventStream so stream failures (404, range-
request bugs, decoder errors, network drops) print instead of
silently halting playback.
- Subscribe to playerStateStream and log playing + processingState
on every transition. Silent stops will now show as a state shift
to completed / idle / buffering with no resumption.
- Subscribe to processingStateStream separately to catch fine-grained
state transitions ExoPlayer reports between source advances.
After hot-restart, tap-then-go-quiet should produce a sequence we
can read — most likely either "processingState=completed" partway
through (server returning premature EOS or wrong Content-Length) or
a thrown error from ExoPlayer's source-loading path.
CI fixes:
- artist_detail_screen.dart: drop unnecessary foundation import (debugPrint
comes from material) and unused metadata_prefetcher import.
Playback timing visibility (so we can stop guessing where the lag
lives):
- playTracks now logs serverUrl / token / configure / setQueue /
play() returned, each stage in milliseconds. The next time you
tap play, we'll see exactly where the seconds go.
- setQueueFromTracks adds two more measurements: total source-build
time across all tracks, and setAudioSources duration.
Small concrete win:
- audio_handler caches the application cache dir path on first use
(already cached in _maybeRegisterStreamCache; now also used in
_buildAudioSource for the LockCachingAudioSource path). One less
platform channel hit per track on cache-miss queue builds.
Once we see real numbers we can decide whether the fix is to build
sources lazily (initial source first → play → background-add the
rest), pre-warm the audio handler at app start so playTracks skips
serverUrl + token reads entirely, or something else.
The prefetcher + alwaysRefresh combination was creating a feedback
loop visible in the logs as repeated `metadataPrefetcher: warming N
albums` cycles, each kicking N parallel getAlbum fetches that then
triggered drift writes that triggered re-emits that re-ran the
prefetcher. Tap-to-play was queueing behind 14+ in-flight cache
fetches.
Three structural fixes:
1. Prefetcher hard-dedupes per session via _warmedArtists Set.
Re-rendering a screen no longer re-fires fetches for ids we've
already seen.
2. Prefetcher only warms artistProvider, not albumProvider. Albums
carry track lists; pre-warming N albums fans out N parallel
"fetch tracks" round trips for content the user may never visit.
Artist rows are single-row lookups — cheap. Album detail loads
on tap (still fast: server-side perf work makes it ~one round
trip).
3. Drop alwaysRefresh from albumProvider, artistProvider,
artistAlbumsProvider, artistTracksProvider. Each was kicking one
silent background refresh per first cache hit. With the prefetcher
creating many subscriptions in parallel, that meant every
prewarmed id triggered an extra fetch even when drift was already
populated. playlistsListProvider keeps alwaysRefresh — system
playlists genuinely rotate UUIDs and need the catch-up. Pull-to-
refresh remains the explicit invalidation path everywhere else.
Removed the warmAlbums calls from the library Albums tab and artist
detail album grid (the storm sources).
Net effect: cold app boot warms ~12-15 artist rows once, period.
Tapping a tile still fetches its detail on demand (one round trip,
fast). User-initiated playback isn't queued behind cache work.
MetadataPrefetcher gains warmAlbums(ids) / warmArtists(ids) public
methods so callers can fan out drift-cache warm-ups for whatever
collection just landed.
Wired into:
- Library Artists tab — warms first 8 artists from the page on every
data emit (initial + paginate + refresh).
- Library Albums tab — same for first 8 albums.
- Artist detail album grid — warms first 8 albums from the artist's
album list as soon as it loads, so tapping into any of them is a
drift hit.
Hard-cap of 8 per call (same as the home prefetch). Set spans across
calls aren't deduped at this layer because providers themselves
short-circuit on cached values.
Also instrument the artist-detail play button: try/catch around the
artistTracksProvider read + playTracks call, snackbar on
empty-tracks or thrown-error so silent failures stop being silent.
The current behavior was an early return on tracks.isEmpty with no
visible feedback.
Across-the-board sluggishness was every cold-cache tap doing one
network round trip while the user waits. SWR helps on re-visits but
the first time you tap a tile from home you eat the latency.
MetadataPrefetcher listens to homeProvider. When /api/home returns,
it fires fire-and-forget reads on albumProvider + artistProvider for
the top-N items in each home section (recently added, rediscover,
most played, last played). Each provider read triggers the existing
cold-cache path, which writes to drift. By the time the user
actually taps a tile, it's already a drift hit and the detail
screen renders instantly.
Cap N=8 per section (covers what's visible on a typical phone
without scrolling). Set spans dedupe across sections so popular
artists don't get fetched five times. Errors are swallowed — a
failed prefetch is silent and the tile falls back to its on-tap
fetch behavior.
Wired alongside the existing audio prefetcher in app.dart's
postFrameCallback.
Detail screen showed empty rows for system playlists because the
fetch batch wrote cachedPlaylists, cachedPlaylistTracks (positions),
cachedArtists, and cachedAlbums — but never cachedTracks themselves.
The detail screen's LEFT OUTER JOIN cachedTracks then returned null
on every row, so trackId was null and titles came back empty.
PlaylistTrack on the wire carries enough to populate cachedTracks
(id, title, albumId, artistId, durationSec). Adds a third dedup map
in fetchAndPopulate, batched with the existing artist + album writes.
track_number / disc_number aren't on the wire so they default to 0;
the detail screen doesn't surface them.
Reusing cachedTracks across albumProvider + playlistDetailProvider
also means tapping a playlist's track to play it now finds the row
in drift instead of triggering another fetch.
The 404 on tapping the For-You tile traced to stale drift rows.
BuildSystemPlaylists rotates UUIDs on every rebuild, so old For-You
/ Songs-Like ids accumulate in cachedPlaylists. The list provider's
fetchAndPopulate was only doing insertOrReplace, which adds new rows
but never removes the obsolete ones — so the home tile renders 8
playlists when the server only knows 4, and 4 of those tiles 404 on
tap.
Two fixes:
playlistsListProvider.fetchAndPopulate now reconciles. After
fetching the fresh list, deleteWhere any user-owned drift row whose
id isn't in the fresh response, then upsert the fresh set in the
same batch. Public-from-others rows are left alone — they're not
keyed by ownership and we don't want to drop someone else's public
playlist just because the current user's response didn't enumerate
it. Operates inside the existing batch so it's atomic.
playlistDetailProvider.fetchAndPopulate now treats a DioException
404 as "this row is stale": delete the cachedPlaylists + any
cachedPlaylistTracks rows for this id, return false so the UI yields
emptyDetail. The next render of the home row sees the row gone and
the tile disappears, completing the cleanup.
Side note: every system-playlist rebuild discards drift rows for the
just-evicted UUIDs and writes the new ones. That cycle's been
silently churning since system playlists shipped — this is the
first time the cleanup actually runs.
handleGetHome itself is well-architected (5 sections in parallel via
goroutines, latency-bound by the slowest single query). The cold-
start lag is two of those queries doing wider scans than necessary.
ListLastPlayedArtistsForUser was iterating FROM artists a with a
LATERAL play_events join per row — O(total_artists in library) plan
even for users who've only played a handful. Inverted: aggregate the
user's plays by artist_id first via the play_events → tracks join
(uses play_events_user_track_idx + tracks pkey), then attach the
artist row and lateral cover/count subqueries only for the artists
that actually appear. Cost now bounded by play history, not library
size.
ListMostPlayedTracksForUser was joining tracks/albums/artists for
every play_event row before grouping — O(total plays) work for
joins. Pre-aggregated play_events into a CTE keyed by track_id +
count(*), then joined to tracks/albums/artists only for the
distinct-tracks survivors. Order-by uses the pre-computed count.
No handler or generated-Go signature changes — both queries return
the same rowset shape, just much faster on libraries where total
artists/plays >> distinct-played-artists/distinct-played-tracks.
Two new sqlc queries replace three sequential per-album round trips
that were dominating detail-screen latency.
GetAlbumWithArtist: handleGetAlbum was doing GetAlbumByID then
GetArtistByID — separate round trips for one logical lookup. The new
query joins albums + artists with sqlc.embed and returns both in one
SELECT. Detail-page DB cost: 3 trips → 2.
ListAlbumsByArtistWithTrackCount: handleGetArtist was loading the
artist's album list, then issuing one CountTracksByAlbum per album to
populate track_count. On a 30-album artist that's 32 sequential
queries — each ~5ms over a local DB, ~30ms over a remote one. The
new query embeds the album row + a correlated count(*) subquery, so
every album's track count comes back in one SELECT regardless of
album count. Detail-page DB cost: 1 + N → 1 + 1.
Together these account for the bulk of cold-cache navigation latency
on the Flutter client. Combined with the existing SWR + nav
hydration on the client side, detail screens should render their
header instantly and the body within one round trip instead of
N+constant.
Closes the gap where LockCachingAudioSource wrote files to disk but
never told AudioCacheManager about them — meaning evict() couldn't
reclaim stream-cached files when usage exceeded the cap, only
explicitly-pinned downloads.
Wire just_audio's bufferedPositionStream as the "download complete"
signal: when bufferedPosition reaches duration (with 200ms slack for
header bytes), look up the on-disk file at the LockCaching path,
read its size, and insert an audio_cache_index row via the new
AudioCacheManager.registerStreamCache(). Source defaults to
incidental so stream-cached tracks are first to be evicted under
pressure.
Dedupe via _streamCacheRegistered Set so we don't hit drift on every
~200ms buffered-position emit. Cache the application cache dir path
on first use for the same reason.
Eviction now sees the full set of files on disk; usageBytes() (which
already walks the dir) and evict() (which reads the index) are
finally consistent for stream-cached tracks. Pinned tracks keep
their existing manual-download flow unchanged.
The Storage card has been showing "0 B" for users who only stream
(never explicitly pin or download an album). usageBytes() summed
SUM(size_bytes) from audio_cache_index — but the streaming path
through audio_handler writes files via LockCachingAudioSource without
ever inserting an index row, so the index undercounts (often to
zero) for normal use.
Walk the cache directory instead. Catches everything on disk:
manually pinned tracks (registered in the index), stream-cached
tracks (LockCaching), partial downloads. Falls back gracefully when
a file is racing against concurrent writes / deletes.
Eviction still operates on the index (it needs the source/recency
metadata to pick eviction order). Stream-cached files aren't subject
to eviction today — separate problem; addressed when we wire a
download-complete hook from LockCaching back into the index.
Tapping the queue button from /now-playing crashed with
NavigatorState._debugCheckDuplicatedPageKeys. Root cause: when I
moved /now-playing out of the ShellRoute earlier, /queue was left
inside the shell. Pushing /queue while /now-playing was on top
made go_router try to mount a second ShellRoute instance under the
existing one — both shells got the same page key.
Move /queue out of the ShellRoute too, sibling to /now-playing.
Shell stays mounted (with whatever child it had) underneath both
top-level routes; pop from /queue returns to /now-playing or the
shell's previous child as appropriate.
Move shuffle / repeat / queue out from below the play controls and
combine with the like + kebab into a single row sitting just above
the seek bar. Title row drops back to title-only (truly centered now,
no Stack needed since nothing competes for the row's right edge).
Layout order is now: art → title → artist → album → [shuffle, repeat,
queue, like, kebab] → seek → prev/play/next.
The "queue" icon in the top-right of the AppBar stays for now —
redundant with the new row but matches what users have already
muscle-memoried for opening the queue from any player state.
Both providers were FutureProvider<Paged<T>> returning a single page
of 50, so once the user scrolled past 50 items the list just ended.
Replace with AsyncNotifier<Paged<T>> that exposes loadMore(): fetches
the next page using items.length as the offset and appends to the
existing list. Idempotent guard via _loadingMore flag — concurrent
scroll events near the bottom collapse to a single fetch.
Both tabs now wrap the GridView in NotificationListener<ScrollNotification>
that fires loadMore() when within 800px of maxScrollExtent. The
lookahead is enough that the next page lands before the user hits
the visible bottom on a typical phone scroll.
Pull-to-refresh updated to invalidate + re-await the provider so a
manual refresh always starts from the first page.
Update installs were failing at the system scanner step because every
release APK was signed with a different signature: build.gradle.kts
fell back to signingConfigs.getByName("debug"), and the debug
keystore is generated per-machine — so every CI runner had a fresh
random key. Android refuses to upgrade an installed app when the
signature doesn't match, hence the "App not installed" failure
after the scan.
Two-part fix:
1. build.gradle.kts now defines a real "release" signing config when
ANDROID_KEYSTORE_PATH is exported (with the matching password +
alias env vars). Without those, falls through to the debug
keystore so local `flutter run --release` still works.
2. flutter.yml decodes ANDROID_KEYSTORE_B64 (CI secret, base64 of the
real release keystore) into a tmp path before the release-APK
build step, then exports ANDROID_KEYSTORE_PATH + the three
password/alias secrets as env vars. CI now hard-errors if the
keystore secret is missing on a tagged build — we don't want
another silent debug-keystore release.
OPERATOR ACTION REQUIRED before the next tag:
1. Generate a release keystore (one-time):
keytool -genkey -v -keystore minstrel-release.keystore \
-alias minstrel -keyalg RSA -keysize 2048 -validity 10000
2. Base64 it:
base64 -w0 minstrel-release.keystore > keystore.b64
3. In Forgejo, add four repo secrets:
ANDROID_KEYSTORE_B64 = contents of keystore.b64
ANDROID_KEY_ALIAS = minstrel
ANDROID_STORE_PASSWORD = the password you set
ANDROID_KEY_PASSWORD = same password (or different alias pwd)
4. Keep minstrel-release.keystore offline and backed up — losing it
means you can never sign an upgrade for any existing install.
5. Existing installs from prior debug-keyed releases must be
uninstalled before installing the first key-signed release. This
is a one-time transition; future tagged builds will upgrade
cleanly.
- artist_detail_screen.dart missed an `import '../models/artist.dart'`
for the ArtistRef seed parameter; analyze flagged undefined_class.
- radio.dart + player_provider.dart doc comments wrapped URL
parameters in <...> which lint reads as HTML. Switched to backticks.
Full player title is now centered absolutely via a Stack: title with
horizontal padding equal to the actions cluster width sits at the
optical center, while LikeButton + TrackActionsButton are pinned to
the right edge with Positioned. Fixed-height SizedBox(32) keeps the
row stable when the title wraps to a single ellipsized line.
Three issues, all related to the player surface:
1. Player UI didn't update on track change. audio_handler's
_onCurrentIndexChanged only kicked off the cover load — it never
pushed the new MediaItem onto the mediaItem stream. Title/artist/
cover stayed pinned to whatever setQueueFromTracks(initialIndex:)
set on first play. Now the listener pushes queue[idx] when the
index changes.
2. Player kebab "Go to artist" 404'd while the same item from
MostPlayed worked. Same TrackActionsSheet for both, but the
player's _trackRefFromMediaItem was hardcoding artistId: ''
because audio_handler's _toMediaItem never stashed it in extras.
Stash artist_id alongside album_id; player_bar +
now_playing_screen read it back. Both kebabs now navigate.
3. "Start radio" didn't exist on Flutter even though the server has
/api/radio?seed_track=<id>. New RadioApi (lib/api/endpoints/
radio.dart) wraps the endpoint; PlayerActions.startRadio(trackId)
fetches + plays the result via the existing playTracks path.
New menu item between "Add to playlist" and the divider above
"Go to album", calls startRadio with a snackbar error fallback.
System playlists now show in the home row (good!) but tapping them
landed on an empty playlist. Same root cause as the earlier album
bug: playlistsListProvider wrote the playlist *row* to drift but
never the tracks. playlistDetailProvider only triggered cold-fetch
when the row was missing, so it yielded with an empty track list.
Refactor playlistDetailProvider to mirror albumProvider's approach:
- fetchAttempted guard (one-shot per subscription).
- Detect "playlist row exists but trackRows empty" and trigger the
same fetchAndPopulate the cold-cache path uses. Drift watch
re-emits with populated tracks.
- 10s timeout on the API fetch + diagnostic prints so any failure
surfaces in logs.
While here, fix two adjacent issues:
- Wipe + re-insert this playlist's track positions on every fetch
so server-side deletions actually propagate. Without the wipe,
removed tracks would linger in drift forever.
- Write the artist + album rows referenced by the fetched tracks
into cachedArtists / cachedAlbums (deduped). Without this, the
joined artistName/albumTitle columns in the playlist track rows
surface empty.
Cover art for system playlists is a separate issue — server emits
coverUrl but it may be empty for system mixes; PlaylistCard falls
back to the queue_music icon. Will tackle in a follow-up.
Artists tab 404: client called /api/library/artists, but the server
mounts the artists list at /api/artists (handleListArtists). Albums
sit at /api/library/albums for historical reasons — the paths aren't
symmetric. Switch listArtists() to /api/artists with sort=alpha.
Albums tab grid: matched the responsive 3-up layout we built for
artist detail. LayoutBuilder computes cellW from available width;
AlbumCard sized to the cell with titleMaxLines: 2; mainAxisExtent
matches actual content height (cover + 2-line title + artist line +
fudge). No more wide-aspect cells with empty space below the card.
Also wired `extra: ref` on the artists/albums grids and the Liked
tab so detail-screen nav hydration kicks in here too — taps from
library screens get the same instant header that home tiles do.
Three changes addressing the cold-start spinner + stale-on-revisit pain.
A. SWR on remaining cacheFirst providers
- artistProvider, artistAlbumsProvider, artistTracksProvider all gain
alwaysRefresh: true. Cache hit still renders instantly; one
background refresh per subscription keeps the row from going stale
forever.
- albumProvider (inline async*) and playlistDetailProvider (inline
async*) now keep a `revalidated` flag and kick a one-shot
background fetch on the first complete cache hit. Same effect as
cacheFirst's alwaysRefresh, just inline.
- Three providers gained the connectivity timeout that
album/artist/playlist already had.
B. Navigation hydration
- AlbumDetailScreen accepts `AlbumRef? seed`; ArtistDetailScreen
accepts `ArtistRef? seed`. When the live provider is still loading,
the seed populates cover/title/artist immediately so the page
isn't blank.
- Routing wires `extra: AlbumRef|ArtistRef` from go_router into
the seed parameter.
- Call sites updated: home (Recently added, Rediscover albums +
artists, Last played), artist detail album grid. Where a ref isn't
available (track actions sheet), the screen falls back to the
spinner — no regression.
C. Cold-start home skeleton
- Replace the full-screen CircularProgressIndicator on /home with a
layout-preserving skeleton: 5 section titles + 6 grey card-shaped
placeholders per row. The page feels populated immediately;
sections fill in independently as data arrives via the per-section
providers.
- Drops the unused DelayedLoading import.
Net effect: re-visits to detail screens render instantly (cache hit
+ silent refresh); first visit from a tile shows the seed header
immediately while tracks load; cold-start home shows a layout
skeleton instead of a 30s blank spinner.
Web UI shows system playlists; Flutter shows placeholders. Wire shape,
adapters, drift schema, and filter constants all line up on inspection,
so adding instrumentation to pinpoint where the system rows fall out:
- Log what /api/playlists?kind=all actually returns (owned/public
counts, including how many of owned are system).
- Log what cacheFirst sees on each drift emit: total rows, filtered,
how many are system, how many landed in owned vs pub, plus the
user.id used for the owned-vs-pub split.
Also add the 3s connectivity timeout that albumProvider/artistProvider
got — keeps the alwaysRefresh path from blocking on a stalled
connectivity stream.
Three log lines from one home-screen visit will tell us:
- Does the server emit system rows? (wire log: system=N)
- Do they land in drift? (drift log: filtered system=N)
- Do they survive the user-id filter? (drift log: owned system=N)
Two test failures from the comparator rewrite:
1. "different unparseable strings → newer" was useful — branch-name
builds (e.g. PackageInfo reports 'main' while server reports
'dev') should still surface the banner so the operator sees that
something is misaligned. Restore that fallback: if both sides
parse to all-zero components (no numeric structure on either),
compare as strings.
2. "honors prerelease ordering" tested pub_semver behavior we no
longer use. Replaced with tests that cover what the new
comparator actually does: zero-padding for length mismatch,
leading-zero normalization, 4-part date+build comparisons, and
month-rollover correctness without leading zeros.
flutter pub get rejects "2026.05.11.0+1" — it requires strict
MAJOR.MINOR.PATCH+BUILD with no leading zeros. Use 2026.5.11+1
as the local default; CI's --build-name="${TAG#v}" still injects
the full 4-part tag for release builds, so production APKs report
the tag verbatim while local dev builds get a sensible recent value.
Update banner showing on identical versions:
- pubspec.yaml was stuck at the placeholder 0.1.0+1, so
PackageInfo.version returned "0.1.0" while the server reported the
actual release tag (e.g. "2026.05.10.1"). Comparison correctly said
"newer" → banner always showed.
- Bump pubspec to 2026.05.11.0+1 so the local default matches the
release cadence even before CI overrides it.
- Update flutter.yml release step to pass --build-name="${TAG#v}" so
every tagged APK reports the tag as its PackageInfo.version. Future
releases stop drifting from pubspec.
- Rewrite isVersionNewer to do component-wise int comparison with
zero-padding: pub_semver.Version.parse rejects 4-part date versions
like "2026.05.10.1", at which point the old code fell back to
string inequality and treated "2026.05.10" as newer than itself
vs "2026.05.10.0". Drop the pub_semver import (no longer used).
Lock-screen play/pause not responding:
- PlaybackState only listed MediaAction.seek in systemActions, which
on Android 13+ means tapping the lock-screen play/pause button
doesn't route back to the AudioHandler. Add play, pause,
skipToNext, skipToPrevious to the set.
- Add androidCompactActionIndices: [0, 1, 2] so the compact
notification view explicitly maps the three buttons.
Album art being smaller than the lock-screen frame is upstream of
this commit — the cover-cache writes whatever pixel dimensions the
server returns. If the server's /api/albums/<id>/cover returns small
thumbnails for these albums, the lock screen renders them at that
size. Worth a separate look at the server cover-emit path.
artistAlbumsProvider populates cachedAlbums (album rows only) without
ever fetching their tracks. When the user taps from the artist grid
into an album, albumProvider sees a drift hit on the album row, never
triggers cold-cache, and yields with an empty track list — visible as
the album header rendering above an empty body.
Refactor albumProvider's cold-cache logic into a helper closure that
either branch (no album row OR album row + no tracks) can call. Add a
fetchAttempted guard so a server response that legitimately returns
zero tracks doesn't loop forever.
Decision flow:
- albumRows empty: cold-fetch (existing path), yield empty on
failure/offline
- albumRows hit but trackRows empty AND not yet attempted: same
cold-fetch — drift watch re-emits with the populated tracks
- albumRows hit, trackRows hit (or attempted already): yield as-is
Track row (album detail):
- Vertical padding 10→6 and only render the artist line when non-empty,
so tracks without a populated artist row don't reserve a blank line.
- Number column 28→22 — visually tighter without making 3-digit
numbers cramp.
Mini player missing artist:
- Symptom was an empty `artist` field on MediaItem after tapping a
track in album detail; layout change wasn't the cause. Real cause:
albumProvider's cold-cache only wrote cachedAlbums + cachedTracks,
never cachedArtists. The drift JOINs that build the AlbumRef + each
TrackRef returned null for the artist row, so artistName fell back
to '' and the audio handler stamped MediaItem.artist=''.
- Cold-cache now collects every distinct (artistId, artistName) pair
off the album header + each track, inserts them into cachedArtists
alongside the album/tracks. Subsequent JOINs populate artistName,
the audio handler stamps it on MediaItem, and the mini player picks
it up via the existing artistName.isNotEmpty render.
- Existing cached albums won't get artist names until they're
re-fetched (cache invalidation TBD). Future cold-cache hits will.
Artist detail album grid:
- AlbumCard gains showArtist (default true). Pass false in the artist
detail grid since the page header already names the artist. cellH
trimmed by the 16dp artist line.
AlbumCard gains optional `width` (default 140 keeps horizontal lists
unchanged) and `titleMaxLines` (default 1) so callers can let the
title wrap to a second line. Cover sizes to width - 16 instead of a
hardcoded 124, so the card scales down cleanly when a narrower cell
width is passed in.
Artist detail grid: switch to 3 columns. LayoutBuilder computes the
cell width from the available width so AlbumCard sizes to the cell
exactly (no overflow). cellH = cover + gap + 2-line title + artist
line + small fudge — tight enough that there's no visible gap below
the card, tall enough to fit a wrapped title.
Drift cache intentionally drops cover_url (server-derived). The
adapters comment says "REST cold-cache fallback briefly shows the real
values before drift takes over" — but once drift takes over, covers
are empty forever. Fix per source:
- Album cover: server constructs the URL deterministically as
/api/albums/<id>/cover (internal/api/convert.go:69). Mirror that
in CachedAlbumAdapter.toRef so AlbumRef.coverUrl is non-empty
whether the row came from a fresh fetch or a drift hit. Restores
cover art on the artist detail album grid (and any other surface
reading albums from drift).
- Artist cover: server picks a representative album and reuses its
cover (convert.go:98). Drift doesn't store the pointer, so derive
client-side via the artist's first loaded album. New _ArtistAvatar
prefers a non-empty server-emitted coverUrl and falls back to
/api/albums/<firstAlbumId>/cover, then slate while the album list
is still loading.
Album grid spacing was off because childAspectRatio: 0.8 inflated
each cell taller than the AlbumCard's actual ~160dp footprint,
leaving a visible gap below every card. Switch to mainAxisExtent: 168
with explicit 8dp main/cross spacing — cells now match the card and
sit on a clean grid.
Move the like + kebab buttons out of the title row and place them as
siblings to the title/artist column. Row's default crossAxisAlignment
centers them against the row's full 48dp height (set by the album
art), so visually they sit at the vertical center of the title+artist
block instead of being pinned to the title baseline.
Width stays 32dp each — horizontal footprint matches what we had.
Bumped icon size 18→20 since they have more vertical space to occupy.
Album detail screen rendered the cover slot as a bare slate Container
— no actual image. Wire ServerImage at /api/albums/<id>/cover (96dp,
rounded). Same for artist detail: ClipOval around ServerImage of the
artist's coverUrl, falling back to slate when empty.
Artist navigation hang: artistProvider goes through cacheFirst, which
had no logging — so we couldn't see whether the cold-cache fetch was
running. Add an optional `tag` parameter that, when set, debugPrints
each step (drift hit/miss, connectivity, fetch start/done/error).
Wire tags into artistProvider and artistAlbumsProvider, plus add the
3s connectivity timeout there too as belt-and-suspenders.
Next test pass should show cacheFirst[artist(<id>)]: lines that
pinpoint where the artist screen is sticking.
Two changes to surface where detail-screen spinners hang:
1. Connectivity().checkConnectivity() goes over a platform channel
that on some Android builds stalls. Wrap with a 2s timeout and
default to "online" if it times out — being wrong about
connectivity costs at most one failed fetch, but being stuck
spins the UI forever.
2. albumProvider's cold-cache flow now debugPrints at every step:
drift miss → connectivity result → API call → drift write → ack
the re-emit. Also adds a 10s timeout on the API fetch and a 3s
timeout on the connectivity await so a hang surfaces as an
error-yielded empty AlbumRef instead of an infinite spinner.
Once the operator's next test produces logs we can pinpoint which
step is hanging. The diagnostic prints stay until the issue's root
cause is confirmed; will be removed in a follow-up.
Root cause for "only the first tile loads, others spin forever":
Connectivity().onConnectivityChanged only emits on connectivity
*changes*, not on subscription. Cold-cache code paths in albumProvider,
artistProvider, playlistsProvider, likes etc. all gate their fetches
on `await ref.read(connectivityProvider.future)`. Until the OS
happened to report the first connectivity flip, that future never
resolved — so any detail screen for an album/artist/playlist not
already in the drift cache hung indefinitely at the connectivity
check.
The "first tile" appeared to work because by the time the user
tapped it, a connectivity event had landed (or that album was already
cached from a prior session). Subsequent tiles whose data wasn't yet
cached blocked.
Switch the provider to async* and seed it with an immediate
checkConnectivity() before forwarding the change stream. .future
now resolves on first read regardless of whether the OS has reported
a change yet.
Image with width+height but no fit paints the source at its intrinsic
resolution inside the box. The cover-cache thumbnails are smaller
than 48dp, so the art rendered as a tiny inset in the slate box
instead of filling it. Cover stretches/crops uniformly to fill.
Why the seek bar appeared frozen: it was reading
PlaybackState.updatePosition, which audio_service only updates on
event transitions (play/pause/buffer/seek). Between events it sits
unchanged, so the bar only jumped at intervals.
Expose just_audio's positionStream (~200ms cadence) from the audio
handler, wrap as positionProvider, and read that in both the mini bar
and the full player. Now the bar advances continuously while playing.
While we're here: spread the full player's vertical layout per
operator — gap to seek 20→32, seek-to-primary 8→24, primary-to-
secondary 16→24, plus 24 below to match. The full player had visible
slack above the controls; this redistributes it.
Title (22pt parchment) → artist (14pt ash) → album (12pt ash). Album
is conditional — drops out if MediaItem.album is empty rather than
leaving a blank line. Tightened the gap to the seek bar (24→20) to
absorb the new line on small screens.
You correctly diagnosed: the mini bar was still visible underneath
the full player, and that's also why the bottom controls overflowed
by 45px (the shell's mini bar was eating the bottom 88dp).
Move /now-playing OUT of the ShellRoute and up to a top-level GoRoute.
Now pushing /now-playing unmounts the shell entirely — no mini bar
underneath, no fight for the bottom strip. The slide-up transition
still feels right because the shell stays painted for the duration of
the animation (so during the transition both are briefly visible, as
you noted is fine).
Drop the hideMini conditional in _ShellWithPlayerBar — no longer
needed since the shell isn't even built when the full player is on
top. Recovering the 88dp also resolves the bottom-controls overflow
without needing to shrink the layout.
Mini player simplifies to track-info + prev/play/next. The
shuffle/repeat/queue cluster moves to the full player, where it has
the room to breathe.
Full player (NowPlayingScreen) now has:
- Real album art (FileImage when artUri is file://, ServerImage from
/api/albums/<id>/cover otherwise — same auth-aware loader as the
rest of the app, with errorBuilder so a failed fetch falls back to
a slate placeholder instead of a broken-image glyph).
- Title row with inline like + kebab.
- Full-width seek with start/end timestamps below.
- 36/72/36 prev/play/next.
- Secondary row: shuffle, repeat, queue.
Mini ↔ Full transition mechanics:
- Tap mini → push /now-playing with custom slide-up transition
(CustomTransitionPage, 280ms easeOutCubic).
- Vertical drag-up flick on mini (>200 px/s) → same push.
- System back / leading expand-more icon → Navigator.pop, which the
CustomTransitionPage replays as a slide-down (240ms easeInCubic).
- Drag-down on full player past 80px OR a >500 px/s downward flick
→ Navigator.pop. Buttons and the seek slider stay tappable since
Flutter's gesture arena gives priority to the more-specific child
recognizers.
- Mini player hides while /now-playing is the active route — the
full player IS the player UI in that state, no need to fight over
visual space.
User reports only the first card in 'Recently added' navigates and the
HitTestBehavior.opaque fix in d703fc2 didn't fully resolve it.
Switch all three card widgets (album, artist, playlist) from
GestureDetector to Material(transparent) + InkWell. This is the more
idiomatic Flutter pattern for tap-to-navigate cards:
- InkWell registers via the Material ripple system instead of the
raw gesture arena, sidestepping any subtle deferToChild edge cases
inside horizontal ListViews.
- Visible ripple feedback on tap means a missed tap is now diagnosable
visually — if the ripple shows but nothing navigates, the issue is
on the route side; if no ripple, the tap never landed.
Material color is transparent so the existing dark page background
shows through unchanged.
You correctly diagnosed the symptom: the AppBar gets taller when the
banner is present. Why: the shell is [Banner, Expanded(child),
PlayerBar] and the banner uses SafeArea(bottom:false) to clear the
status bar. But MediaQuery.padding is screen-relative — the child's
Scaffold/AppBar reads MediaQuery.padding.top and applies its own
status-bar inset on top of the banner, doubling the gap.
Watch shouldShowUpdateBannerProvider in the shell. When it returns
non-null (banner visible), wrap the child in MediaQuery.removePadding
removeTop: true so the AppBar mounts directly under the banner. When
the banner is hidden, child gets its normal top inset.
The 401s on /api/albums/<id>/cover were the root cause of "only the
first tile is navigable" — failed images leaving slate placeholders
that combined with the deferToChild hit-test (fixed in d703fc2) to
silently swallow taps.
ServerImage:
- sessionTokenProvider is a FutureProvider; .value is null after a
hot restart until secure-storage resolves. Old code fired
Image.network with headers:null at that moment, the network image
cache locked in the 401 response, and never retried. Switch to
AsyncValue.when so the widget waits for the token before mounting
Image.network with the auth header attached.
- Add errorBuilder so a single failed image renders the parent
fallback instead of leaking a stack trace + broken-icon glyph.
Player bar:
- media.artUri is set by AlbumCoverCache as Uri.file(path). Wrapping
it in NetworkImage attempts HTTP on a file:// URI and fails. Use
FileImage when the scheme is file, else NetworkImage.
Net effect: failed image loads no longer leak errors or block other
widgets from rendering / receiving taps.
Three issues from the screenshot:
1. Title invisible: LikeButton + TrackActionsButton were raw IconButtons
with the default 48dp tap target, eating ~96dp of the title row.
Wrap both in 32×32 SizedBox at the call site so the title actually
gets enough room to render (still ellipsizes if needed).
2. Layout per operator's revised spec: track info gets a 2:1 share of
the row vs the controls column. The controls column now stacks the
shuffle/repeat/queue sub-row above the prev/play/next sub-row, both
right-aligned.
3. Right-edge 2px overflow: previous play controls were 40+48+40=128dp
in a flex-1 slot of ~110dp. Resize to 32+40+32=104dp matching the
shuffle/repeat/queue row above (3×32=96dp), with the play button
slightly larger to keep visual hierarchy.
Drop the volume provider import since the slider is gone (handled by
phone hardware buttons).
Same root cause as the playlist card fix in f65650f — GestureDetector
defers to children by default, so taps over the cover image area
(ServerImage / SVG fallback) didn't reliably propagate. Mark the
detector opaque so the entire card surface is tappable, not just the
text below the cover.
Player bar: drop the volume slider — phone hardware controls volume.
Right cluster collapses to a single row of shuffle/repeat/queue.
Playlist card: GestureDetector defers to children by default, so taps
over the cover image area didn't always register. Add HitTestBehavior.opaque
so the whole card surface is tappable.
Update banner: was 3-4× the height of its text — IconButton's default
48dp tap target plus generous vertical padding dominated. Tighten
padding (8→4), shrink dismiss button to 32×32 with iconSize 16, slim
the action button to 28dp height with shrinkWrap tap target.
Home screen: ClampingScrollPhysics + 140dp bottom padding (already
applied in working tree) so scroll stops just above the mini-player.
The previous commit message said I added data-testid="player-bar-compact"
+ "player-bar-desktop" to PlayerBar.svelte but the edits actually
dropped silently — only the test file changes landed. Without the
testids, the test file's within(getByTestId(...)) calls all fail.
This adds the actual attributes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two bundled fixes:
### Web: PlayerBar test failures
PlayerBar renders both compact (md:hidden) and desktop (hidden md:flex)
blocks; jsdom doesn't apply CSS media queries so both DOM trees are
present in tests. screen.getByRole/getByText found duplicates →
14 test failures.
Added data-testid="player-bar-compact" + "player-bar-desktop"; tests
scope queries via within(getByTestId(...)). Compact is the focus of
#358, so most tests scope there. The "Up next" subgroup explicitly
scopes to desktop since that copy was dropped from compact.
### Flutter: system playlists not loading
cacheFirst was too conservative — when drift had user-created
playlists from sync but no system playlists (For-You / Discover),
fetchAndPopulate never fired (drift wasn't empty). Result: home
tile row showed user playlists but never the system ones.
Added cacheFirst alwaysRefresh option = stale-while-revalidate.
playlistsListProvider opts in: yields cache immediately, then kicks
off REST refresh in background. Drift watch() picks up the new rows
and re-emits.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the kebab-with-popover PlayerOverflowMenu approach with all
player controls visible inline. Operator's stated intent was "two-row
layout with all controls visible," not "kebab hides shuffle/repeat/
queue/volume" — restructured to match.
### Compact view (below md)
┌─────────┬───────────┬───────────────────┐
│ art │ ♥ ⋮ │ 🔀🔁 ☰ │ ← short sub-row
│ title │ │ │
│ artist │ ⏮ ⏯ ⏭ │ volume slider │ ← play row at full size
├─────────┴───────────┴───────────────────┤
│ 0:42 ━━━━━●━━━━━━━━━━━━━━━━━━━━━━ 3:21 │
└─────────────────────────────────────────┘
- Column 1: cover + title/artist
- Column 2: like+kebab on top sub-row, play controls (40/48/40) below
- Column 3: shuffle/repeat/queue on top, volume slider below
- Bottom: full-width seek slider with time labels
### Desktop view (md+)
Existing 3-column layout preserved; only change is the redundant
PlayerOverflowMenu kebab (which had been mounting at all widths)
is no longer there. Volume + shuffle + repeat + queue stay inline
in the right cluster as before.
### Cleanup
- Deleted PlayerOverflowMenu.svelte + its test (no callers remain)
- The TrackMenu kebab (per-track actions) stays — it's a varying
list of actions that doesn't fit inline
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Came up debugging the in-app update flow — wasn't obvious which image
the container was actually running without exec'ing in. New flow:
### Server
- internal/server/version.go: new `var ServerVersion = "dev"`,
overridden via -ldflags at build time.
- /healthz response gains a "version" key alongside the existing
"status" + "min_client_version". Backward-compat: existing clients
ignore unknown JSON fields. Endpoint stays unauthenticated.
### Build
- Dockerfile: new `ARG MINSTREL_VERSION=dev`, threaded into the
go build -ldflags so the binary's ServerVersion is stamped at
link time. Default "dev" preserves local `docker build` ergonomics.
- .forgejo/workflows/release.yml: tags step also emits a `version`
output (the git tag for tag pushes, "main" for branch pushes);
build step passes it as `--build-arg MINSTREL_VERSION=...`.
### Web
- web/src/lib/components/ServerVersion.svelte: small understated
text ("Server v2026.05.10.2") that fetches /healthz on mount.
Renders nothing on parse failure or pre-version images.
- Mounted at the bottom of Settings.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Server's clientVersionResponse marshals to {version, apk_url,
size_bytes} but the component was reading body.apkUrl / body.sizeBytes
(camelCase). The !body.apkUrl guard was always true → early return →
download button never rendered even when /api/client/version returned
200 with valid data.
Verified directly: /api/client/version returns
{"version":"v2026.05.10.1","apk_url":"/api/client/apk","size_bytes":65301242}
on the deployed v2026.05.10.1 image; the UI just wasn't reading those
keys.
Map wire (snake_case) → component (camelCase) explicitly via a
WireInfo type so the boundary is auditable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When /api/client/version returns 404 (no APK in /app/client/, e.g.
v2026.05.10.0 which had a failed CI APK-attach step), the previous
shape rendered the section heading + intro paragraph with an empty
hole where the button should be. Confusing — looks like a broken
button.
Moved the section wrapper, heading, and intro paragraph INSIDE
MobileAppDownload so the whole block collapses together when info
is null. Settings page just mounts <MobileAppDownload />; nothing
visible if no APK is available.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both prior secrets did related work (one for registry push, one for
release asset I/O) and required separate Forgejo PATs to manage.
Replacing with a single CI_TOKEN keyed off a single PAT with the
union of scopes (write:repository + write:package).
- flutter.yml: APK attach step → CI_TOKEN
- release.yml: docker login → CI_TOKEN; APK fetch step → CI_TOKEN
No behavior change — same calls, single secret name. Operator needs
to set CI_TOKEN in repo settings (Actions → Secrets) with a PAT that
has write:repository + write:package; the prior REGISTRY_TOKEN /
RELEASE_TOKEN secrets can be removed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes the bandwidth-abuse vector on the in-app update flow. Both
endpoints now sit inside the authed.Group; APK additionally gets a
60s/user rate limit to suppress accidental hammering or scripted
abuse.
### Server
- internal/api/client_assets.go:
- clientAPKAllowDownload(): in-memory map[userID]time.Time under
a mutex. Returns 0 (allow) or wait duration (block).
- handleClientAPK reads user from context, checks the limit,
returns 429 + Retry-After header if blocked.
- testResetClientAPKRateLimit() lets tests start clean.
- internal/api/api.go: routes moved from the root /api group into
the authed.Group block (alongside /quarantine, /requests, etc.).
- Tests: added TestClientAPK_401WhenUnauthenticated and
TestClientAPK_RateLimit_429OnRapidSecondCall (also verifies
different user gets a fresh slot). Existing tests updated to use
authedRequest() helper.
### Web
- MobileAppDownload.svelte: switched from bare fetch (no credentials)
to api.get<>() which carries the session cookie. 404 / 401 /
network errors all silently hide the download row.
- Removed the login-page mount entirely — pre-auth surfaces should
never show this. Settings → Mobile app section keeps it for
logged-in users.
Flutter unaffected: dio's Bearer interceptor already attaches the
token, and the polling only fires once the post-login shell mounts
the banner widget.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes the discoverability gap on the in-app update flow — the
/api/client/apk endpoint exists but had no web UI surface to find
it. User asked to be able to "visit the site on my phone and
download the apk from there."
- web/src/lib/components/MobileAppDownload.svelte — fetches
/api/client/version once on mount; renders a download link with
version + size when 200; renders nothing on 404 (no APK bundled,
graceful degradation in dev environments and pre-CI-wiring images).
- Mounted on the login page (below the Register link) so the link
is discoverable without authentication. The /api/client/* endpoints
are themselves unauthed, so the flow works end-to-end for any
visitor on a phone.
- Also mounted in Settings → Mobile app section for logged-in users
who want to grab the matching APK for sideloading on a different
device.
Browser handles the download via the server's existing
Content-Disposition: attachment; filename="minstrel.apk" header.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
pub_semver is permissive with leading zeros — '2026.05.10' parses as
2026.5.10, so date-style tags get strict semver ordering, not the
unparseable-fallback path. The test was asserting "any difference =
newer" for date strings, which is incorrect; the actual behavior
(and the right behavior) is proper ordering.
Split the test group into two:
- date-style (parses as semver): newer/older/equal assertions
- truly unparseable (e.g. branch names like 'main', 'dev'): the
fallback's "any difference = newer" semantics still apply
Implementation in client_update_provider.dart unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The previous fix-forward intended this change but the Edit didn't
make it into the commit (only the Go side landed). Re-applying the
StateProvider → NotifierProvider conversion so flutter analyze stops
failing on the undefined-function for StateProvider.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- flutter analyze: Riverpod 3 dropped StateProvider; replaced
_dismissedVersionsProvider with a small Notifier<Set<String>>
+ NotifierProvider, mutating via an `add(version)` method.
Public API (shouldShowUpdateBannerProvider, dismissUpdateProvider)
unchanged.
- golangci errcheck: defer f.Close() now wrapped in func() { _ = f.Close() }();
os.Setenv calls in test helper switched to t.Setenv (cleaner — auto-restores
on test cleanup, no manual Unsetenv needed).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Final phase of the in-app update flow. release.yml on tag pushes
fetches the APK that flutter.yml is attaching to the same release,
drops it into client/ in the build context, and the Dockerfile's
COPY client/ /app/client/ bakes it into the image.
### How it sequences
flutter.yml and release.yml both trigger on tag pushes and run in
parallel on different runners (flutter-ci vs go-ci). flutter.yml
typically finishes APK build + release attachment in 2-5 min.
release.yml polls the release page for up to 15 min for the APK to
appear, then proceeds. If the APK never lands (flutter.yml failure,
network hiccup), release.yml emits a warning and ships the image
without the bundled APK — server returns 404 from /api/client/version,
banner stays hidden, manual download from the release page still
works. Graceful degradation, not a blocker.
### Repo shape
- client/.gitkeep + client/README.md so the directory exists in git
and the COPY in the Dockerfile always finds something to copy
- .gitignore excludes client/minstrel.apk + .version so accidental
commits don't bloat the repo
- Dockerfile: COPY --chown=minstrel:minstrel client/ /app/client/
### Why polling vs cross-workflow trigger
Forgejo Actions' workflow_run support varies by runner version; the
polling approach is universally compatible. If/when we standardize on
a runner that handles workflow_run cleanly, the polling step can
become a `needs:` dependency.
Closes#397.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes the operator-facing half of #397. Soft banner mounts at the
top of the shell when the server has a newer APK bundled at
/api/client/version; tap Install → APK downloads to cache → Android
PackageInstaller intent fires → user taps Install in the system
dialog → app restarts on the new version.
### Dart side
- lib/update/update_info.dart — wire shape for /api/client/version
- lib/update/client_update_provider.dart — Riverpod AsyncNotifier
polling on app start + every 24h. Compares semver via pub_semver
with non-semver string-equality fallback. Server 404 = "no update
channel" = silent. Companion shouldShowUpdateBannerProvider gates
on a per-version dismissed-set (in-memory; resets on restart).
- lib/update/installer.dart — UpdateInstaller wraps dio.download +
the MethodChannel call to MainActivity.kt.
- lib/update/update_banner.dart — Material banner with idle /
downloading (with progress) / error stages. Mounted in
_ShellWithPlayerBar above the route content; SizedBox.shrink()
when no update available so non-shell routes (login, server-url)
see no banner anyway.
- isVersionNewer() extracted as a pure function and tested across
semver, prerelease ordering, leading-v normalization, and
non-semver fallback (date-tag style).
### Android side
- AndroidManifest.xml — REQUEST_INSTALL_PACKAGES permission;
FileProvider with authorities ${applicationId}.fileprovider.
- res/xml/file_paths.xml — exposes the cache directory so the
downloaded APK can be served via content:// URI (file:// is
blocked since Android 7).
- MainActivity.kt — MethodChannel handler builds the FileProvider
URI and fires Intent.ACTION_VIEW with FLAG_GRANT_READ_URI_PERMISSION
so the system installer can read across the process boundary.
Phase 3 (CI sequencing — bake APK + version file into /app/client/
during release.yml's tag flow) is the remaining piece. Without it
the server returns 404 from /api/client/version and the banner
silently does nothing — graceful degradation, but no actual updates.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase 1 of the in-app update flow — server side. Endpoints serve the
bundled Android APK + sidecar version file from /app/client/.
Returns 404 gracefully when files aren't present, so dev environments
and pre-CI-wiring images degrade cleanly to "no update available."
- internal/api/client_assets.go: handleClientVersion + handleClientAPK.
Both unauthenticated (matches /healthz) so install flow doesn't
depend on a live session. APK served with proper
application/vnd.android.package-archive Content-Type +
http.ServeContent so Range requests work for resumable downloads
on flaky networks.
- Path resolves to /app/client/ by default; MINSTREL_CLIENT_APK_DIR
env var overrides for dev.
- Dockerfile creates /app/client/ + commented COPY hooks for the CI
sequencing phase.
- Tests cover all four states: missing apk, apk-but-no-version,
both present (200 with correct shape), apk stream (200 with
correct Content-Type + body bytes).
Phases 2 (Flutter client provider + banner + install intent + Android
manifest changes) and 3 (CI sequencing to bake the APK into the image)
land in follow-up commits.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two fixes in one commit because they're entangled — the systemVariant
work would have been theater otherwise.
## The wire-format bug
/api/library/sync was emitting PascalCase JSON for artist / album /
track / playlist upserts (raw json.Marshal of sqlc-generated structs
with no JSON tags — sqlc.yaml: emit_json_tags=false). Flutter's
sync_controller _*FromJson reads snake_case keys, so all metadata
sync rows landed in drift with empty strings / zero ints.
The like_track / like_album / like_artist / playlist_track entities
work because they're hand-built `map[string]string` payloads with
snake_case keys — they sidestepped the bug. The 4 raw-marshal
entities did not.
Existing sync test caught zero of this — it asserts on len(upserts)
not field shape.
Fix: server-side view structs in library_sync_views.go with proper
JSON tags + pgtype-flattening (UUID → 8-4-4-4-12 hex string,
Date → "2006-01-02"). Mirrors the playlistRowView pattern from
/api/playlists. New library_sync_views_test.go pins the wire keys
so future field-name drift breaks loud.
## systemVariant column (closes#357 plan C v1 limitation)
playlistSyncView now carries `system_variant` server → wire.
Flutter drift schema bumped from 1 → 2 with onUpgrade adding the
`systemVariant TEXT NULL` column to cached_playlists. Cursor reset
to 0 in the migration so existing rows refresh with the new field
on the next sync.
playlistsListProvider now filters locally by systemVariant:
- kind='user' → systemVariant IS NULL (the add-to-playlist sheet's intent)
- kind='system' → systemVariant IS NOT NULL
- kind='all' → no filter
Closes the documented v1 limitation where the add-to-playlist sheet
showed system playlists alongside user-created ones.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes the cleanest remaining #356 gap — Discover submits Lidarr
requests but had no surface for users to view or cancel their own.
Admin had it; user didn't.
- RequestsApi at lib/api/endpoints/requests.dart — listMine() +
cancel(id). Same /api/requests endpoint the web /requests page
uses; identical AdminRequest wire shape so the existing model
is reused (just augmented with matched_*_id fields for the
"Listen" CTA on completed rows).
- MyRequestsController mirrors the web's auto-poll (#369 piece
already shipped server-side / web-side): 12s refresh while any
row is mid-ingest (status='approved'), stops on settle. Riverpod
Timer in build() that's cancelled in onDispose.
- RequestsScreen with kind/status pills, ingest progress copy,
Cancel (with confirm dialog) and Listen CTAs per row state.
- Route /requests under the shell. Entry point in settings between
Profile and Appearance — "My requests".
- Widget tests cover empty / pending / completed / rejected states +
the Cancel-confirm dialog.
Out of scope this slice: Discover-screen "View your requests" CTA
after submitting a new request. Reasonable follow-up but doesn't
block the management loop.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Half of #369 — the auto-poll piece. Inbox deferred (separate task).
Both /requests (user) and /admin/requests (admin, 'approved' tab only)
now refetch every 12s while at least one row has status='approved'.
Stops automatically when all rows settle to pending/completed/rejected.
TanStack Query's default refetchIntervalInBackground=false handles
the visibility behavior — polling pauses when the tab is hidden and
resumes on focus.
Predicate hasInFlightRequest() is exported + tested so the polling
logic is auditable independent of TanStack Query's internals.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phases 1+2 of the background-play resilience work. Net effect on
desktop: gap-free track transitions on slow networks, automatic
recovery from transient stalls / network errors mid-track.
- preload="metadata" → "auto" on the main audio element so the
browser pre-buffers more aggressively.
- attachStallRetry() — listens for stalled / waiting / network-error
events. Schedules a reload-from-current-position after delayMs
(3s default) if the buffer hasn't recovered. MEDIA_ERR_NETWORK
triggers immediate retry. Bounded by maxRetries (3 default) to
prevent tight loops on persistent failures.
- audioLoader seam — `lib/player/audioLoader.ts` exposes a
resolve(track) → URL + optional prefetch(track) hint. Default
returns track.stream_url; the planned Tauri shell can swap via
setAudioLoader() to return blob URLs backed by a Rust cache,
mirroring the Flutter drift+LockCachingAudioSource pattern.
- Hidden prefetch <audio> element. When current track passes 50%,
start loading queue[index+1]. On track-end the swap is gap-free
via browser HTTP cache + audio buffer warm-up.
Phase 3 (service worker chunk cache) deliberately deferred — that
work belongs in the Tauri shell as native Rust caching, not as a
web-only investment that would compete with it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The mobile-responsive design called for all 4 admin tables to use
RowActionsMenu. Requests + Quarantine + Users users-section landed
in earlier commits; the invites section in the same users page was
still rendering inline Copy/Revoke buttons. Now uses RowActionsMenu
with primary Copy + secondary Revoke (danger).
Extend expiry from the design's secondary list deferred — no
server endpoint exists yet (lib/api/admin.ts has no
extendInviteExpiry).
Closes the in-code portion of #358. Real-device verification
walkthrough at 375/414/768 + grid/safe-area checks remain
operator-side.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The MediaCard consolidation in the discovery doc would have needed
~14 props to cover the differentiating bits (wrapping element a vs
button, art-shape, art-fallback strategy, click semantics, title
align/size, subtitle count). That's a god-prop blob that's harder to
reason about than three explicit cards.
Narrowed to the actually-drifted bit: the action cluster
(LikeButton + Plus + bottom-right menu slot, including the
stopPropagation defensive wrappers). One small component covers
AlbumCard / ArtistCard / CompactTrackCard so the focus-ring and
button-styling drift can't recur.
Each card keeps its own data-fetching, art container, and wrapping
element since those have load-bearing differences. The consolidation
addresses drift, not LOC for its own sake.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces inline TrackRef literal redefinitions with calls to the
test-utils helper. Tests that asserted on default field values
(e.g. track titles, artist names rendered in the DOM) keep explicit
overrides; tests that only need a stub for shape now use makeTrack()
with no overrides.
PlaylistCard.test.ts, PlaylistTrackRow.test.ts, and playlist.test.ts
SKIPPED — they use PlaylistTrack (with track_id/added_at), not TrackRef.
ArtistCard.svelte and +page.svelte route files (matched by initial grep)
SKIPPED — live code, not test files.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After e8eff1b migrated playlists.ts from raw apiFetch to the api.*
wrapper, three test files mocked the wrong surface:
- playlists.test.ts: GET case asserted init.method === 'GET' but
api.get omits init.method entirely (fetch defaults to GET). Fall
back to 'GET' when init.method is undefined.
- playlists.refresh-discover.test.ts: re-mock ./client to expose
`api: { post: vi.fn() }` instead of `apiFetch`; assertions check
api.post call args.
- playlists.refresh-foryou.test.ts: same.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the hand-rolled { subscribe, data } quarantine stub with the
emptyQuarantineMock() helper. readable() satisfies the same store
contract the page actually uses (subscription via $store), and the
playlist tests don't assert on quarantine state — the mock is purely
a transitive module-load satisfier (PlaylistTrackRow → TrackMenu).
Companion to commit 7e8d196 (likes sweep), which initially skipped
this file because it shared the hand-rolled stub shape with the likes
mock; the quarantine half can safely move to the helper.
Five other files in the original 8-file batch (PlaylistTrackRow,
TrackMenu, TrackRow, PlayerBar, CompactTrackCard) were absorbed into
that likes-sweep commit; search/tracks was absorbed into commit
dd67f28 (pageUrl sweep). hidden.test.ts SKIPPED — it imports
createMyQuarantineQuery as a vi.fn() to mockReturnValue per-test and
asserts on unflagTrack call args; the helper would defeat both.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces duplicated inline vi.mock('$lib/api/likes', ...) blocks with
the emptyLikesMock() helper, and (where previously left in working
tree by a parallel quarantine sweep) rolls in the matching
emptyQuarantineMock() switchovers in the same files.
Together with commit dd67f28 — which already swept the four search/*
test files for likes — this completes the 18 likes-mock conversions
called out in #375.
LikeButton.test.ts kept inline — it uses a state-driven mock to
assert the toggle behavior.
SKIPPED:
- liked.test.ts: re-exports createLikedTracksInfiniteQuery and the
album/artist variants that emptyLikesMock() doesn't provide.
- playlist.test.ts: hand-rolled { subscribe, data } stub instead of
readable(...) — different store contract.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Centralizes the inline { page: { get url() { return state.pageUrl } } }
mock shape via test-utils/mocks/appState.ts. The vi.hoisted state
declaration remains per-file (vitest mock-hoisting requires module-
level declaration). Adds a $test-utils alias to svelte.config.js so
the helper can be imported without ../../../ chains.
SKIPPED (mock exposes more than url):
- Shell.test.ts, MobileNavDrawer.test.ts, admin.test.ts: static
page: { url: ... } shape, no hoisted state to centralize.
- album.test.ts, artist.test.ts: page exposes both params and url.
- playlist.test.ts, reset-password.test.ts: page exposes params only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wraps the dominant pattern (try { await fn } catch { pushToast(errMessage,
'error') }) for the 5 admin/+page.svelte handlers that surface errMessage
directly: onApprove, onReject, onResolve, onDeleteFile, onDeleteLidarr.
Tighter scope than originally planned — the 3 handlers in
admin/users/+page.svelte (onToggleAutoApprove, onGenerateInvite,
onRevokeInvite) use errCode + verb prefix ("Generate failed: ${code}")
which is structurally different and would change UX wire output if
forced through the helper. Skipping them keeps behavior identical.
The helper SWALLOWS errors so callers no longer need try/finally for
busy-state — `saving = true; await runMutation(...); saving = false;`
is correct because runMutation can't throw.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
reconstruction (#375)
PlaylistTrackRow.svelte and PlaylistCard.toTrackRefs both rebuilt
TrackRef from PlaylistTrack with the same field-by-field literal.
Server adding a new TrackRef field would have needed both sites
updated; now one helper owns the mapping.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three nearly-identical try/catch wrappers across theme + player stores
collapse to read()/write()/remove() in lib/util/safeLocalStorage.ts.
Sets the pattern for future stores. Caller still does parse/serialize
since the existing call sites store strings (theme preference, volume
number) — no JSON wrapper needed yet.
persisted.ts left alone — its JSON-payload + per-key-suffix shape is
distinct enough to keep self-contained.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Drop 9 hand-rolled apiFetch calls with redundant Content-Type
headers + manual JSON.stringify. The api.* wrapper in client.ts
already handles both. Identical wire shape; existing tests cover
all endpoints.
Also makes api.del generic so DELETE endpoints with typed returns
(playlist track removal) can use the wrapper instead of raw
apiFetch. Default `T = null` preserves existing callers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes the last theme-color hardcode. Runtime side already imports
tokens.colors.{dark,light}.obsidian via applyMetaThemeColor; build
side now matches via JSON import-attributes. Drops the // TODO(#375)
tripwire.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
auth/store.svelte.ts no longer imports from $lib/player. Logout
broadcasts via auth/sessionEnd.svelte.ts (a tiny tick + outgoing
userId pair); player subscribes through $effect.root and owns the
player-specific teardown (clearPersistedQueue + playQueue([]) +
closeQueueDrawer()).
Cycle fully inverted — auth is now a pure leaf for player; future
session-end consumers (download cache, recent-search history) plug
in by adding their own effect on the same signal.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- cache_first.dart: backtick-fence List<T> doc comment to dodge
unintended_html_in_doc_comment.
- library_providers.dart:163: switch single-row insert to
insertAllOnConflictUpdate; Batch only exposes the *AllOnConflict*
variant.
- 5 test files: StreamProvider.overrideWith takes Create<Stream<T>>,
not Create<Future<T>>. Wrap data emissions in Stream.value(...).
- artist_detail_screen_test: add models/album.dart import for AlbumRef
type annotation.
- track_actions_sheet_test: drop _StubLiked extends LikedIdsController
(notifier no longer exists post-migration); override with
Stream.value(LikedIds(...)) instead.
- like_button_test: rollback assertion now requires drift writes via
LikesController. Skip under _skipDrift until libsqlite3-dev lands on
the runner image (Fable #399). Replace stale .notifier reference
with likesControllerProvider for completeness.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
playlistsListProvider — StreamProvider over cached_playlists. Returns
all user-owned + others' public playlists. The 'kind' family arg only
affects the REST cold-cache fetch (server-side filter); drift can't
distinguish 'user' from 'system' kind because systemVariant isn't
persisted. v1 limitation: add-to-playlist sheet may briefly show
system playlists too. Follow-up: add systemVariant column + schema bump.
playlistDetailProvider — async* over the playlist watch stream + a
one-shot tracks query per emission (joins playlist_tracks → tracks →
artists → albums for snapshot fields). Cold-cache fallback inserts
playlist + playlist_tracks rows in one batch, then awaits re-emission.
Pull-to-refresh on these screens now triggers re-subscription rather
than forcing a REST fetch — drift is the source of truth and stays
fresh via SyncController's delta sync. Documented behaviour shift; if
operator wants force-refetch we can add it via a sync trigger later.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two consumer updates left out of 8a6c926: like_button.dart and
track_actions_sheet.dart still referenced the removed
likedIdsProvider.notifier API. Switched to the new
likesControllerProvider.toggle().
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Splits the previous LikedIdsController into two:
- likedIdsProvider — StreamProvider reading from cached_likes via
drift watch(). Reactive: SyncController writes propagate
automatically. Empty + online triggers REST cold-cache fallback
that populates cached_likes via insertOrIgnore (sync may have
already written some rows).
- likesControllerProvider (new) — exposes toggle(LikeKind, id).
Optimistic: writes drift first (UI updates instantly via the
watch stream), then REST. Rolls back drift on REST failure.
Two consumer updates: like_button.dart + track_actions_sheet.dart
switch from likedIdsProvider.notifier.toggle to
likesControllerProvider.toggle.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three providers all become StreamProviders driven by drift watch():
- artistAlbumsProvider: cacheFirst over the join of cached_albums +
cached_artists for artist_name on each row.
- artistTracksProvider: cacheFirst over the join of cached_tracks +
cached_artists + cached_albums for snapshot fields.
- albumProvider: composite ({album, tracks}) shape, so uses async*
with two queries (album + tracks) for cleaner reactive behavior.
Cold-cache fallback inlined; populates both album and tracks tables
in one batch.
.future call site in artist_detail_screen.dart left as-is —
StreamProvider.future returns the next emission with the same
signature, so no consumer change needed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Foundation for the provider migrations. cacheFirst<D, T> wraps the
drift.watch() + REST cold-cache fallback pattern: yields cached rows
when present, kicks off REST fetch + drift populate when empty +
online, yields empty when offline. REST failures swallow to empty so
callers can surface errors via toast.
adapters.dart adds CachedX → XRef extension methods + reverse
XRef.toDrift() companions for Artist/Album/Track/Playlist. Adapters
accept some loss of server-derived fields (coverUrl, streamUrl,
ownerUsername) — UI already handles empty values; cold-cache fallback
briefly shows the real values before drift takes over.
cache_first_test covers all 4 branches (non-empty, empty+online,
empty+offline, REST failure). adapters_test covers basic round-trips.
Both safe to run on CI runner — no drift open required.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pending-timer failure: CachedIndicator (now in TrackRow) reads
audioCacheManagerProvider, which constructs AppDb via drift_flutter's
driftDatabase(), which calls libsqlite3 — missing on the flutter-ci
runner. The async chain leaves a pending timer at test teardown.
Skip cohort matches sync_controller / audio_cache_manager / storage_section
tests; all re-enable once Fable #399 lands the runner image fix.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CI analyze flagged storage_section_test.dart's `skip: _skipDrift` calls.
testWidgets has signature `skip: bool?`; only test() takes `String?` as
the skip-reason. Changed _skipDrift in this file to `const bool = true`
and moved the rationale into a comment. The other two drift-cohort
files (sync_controller_test, audio_cache_manager_test) use test() so
their String const is fine.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CI's flutter-ci runner doesn't ship libsqlite3.so. drift's NativeDatabase
fails at first use ("Failed to load dynamic library 'libsqlite3.so'").
Affected files:
- test/cache/sync_controller_test.dart (4 tests)
- test/cache/audio_cache_manager_test.dart (5 tests)
- test/settings/storage_section_test.dart (2 tests, was silently
succeeding because the drift call was best-effort but emitted
multiple-AppDb warnings)
All 11 marked with skip: '...' + a top-level @Tags(['drift']) library
declaration so they can be re-enabled by tag once the runner image has
libsqlite3-dev installed (or once we move VM tests to sqlite3/wasm).
On-device verification covers the actual cache + sync logic.
Plus two collateral fixes:
- test/cache/connectivity_provider_test.dart: connectivity_plus needs
a platform channel that doesn't exist in unit tests; reduced to a
non-null import smoke check.
- test/library/widgets_smoke_test.dart: TrackRow now contains
CachedIndicator (ConsumerWidget); wrapped TrackRow test in
ProviderScope.
Filing follow-up for the runner image fix in next commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Cleared 5 errors + 1 warning + 1 info from CI flutter analyze on 5114a81:
- prefetcher.dart: Riverpod listener callbacks used (_, _) for two
placeholder params — Dart 3 enforces unique names. Changed to (_, __).
- prefetcher.dart: Riverpod 3's AsyncValue exposes `.value` (nullable)
not `.valueOrNull`. Three call sites updated.
- sync_controller.dart: doc comment had `?since=<cursor>` → analyzer
warned about unintended HTML. Wrapped in backticks with `{cursor}`.
- sync_controller.dart: delete loop over heterogeneous Table list
inferred as List<Table>; drift's delete() expects TableInfo. Unrolled
to explicit per-table deletes.
- audio_cache_manager_test.dart: db.into(...).insertAll([...]) doesn't
exist on InsertStatement — only insert/insertOnConflictUpdate. Used
db.batch((b) => b.insertAll(table, [...])) instead, in two test cases.
- audio_handler.dart: LockCachingAudioSource is marked experimental in
just_audio. Added // ignore: experimental_member_use — operator
acknowledged the experimental status during brainstorming and we're
the project; CI's --fatal-infos would otherwise gate.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CachedIndicator (lib/library/widgets/cached_indicator.dart) — small
download glyph rendered next to a track row when AudioCacheManager
reports the track as cached. FutureBuilder one-shot.
Wiring:
- track_row.dart: render CachedIndicator before the duration label
- playlist_detail: 'Download' OutlinedButton next to Play; pins all
playable tracks with source: autoPlaylist + SnackBar feedback
- album_detail: 'Download' IconButton in the header; same pin pattern
- app.dart: now ConsumerStatefulWidget — initState fires the initial
sync + activates the prefetcher provider
Together these complete the operator-facing surfaces of the offline
slice: visible cache state, explicit download trigger, automatic
sync + queue-ahead prefetch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
_buildAudioSource is now async and cache-aware:
1. Cache hit → AudioSource.uri(file://path)
2. Cache miss with manager → LockCachingAudioSource (cache-as-you-play)
3. No manager configured → plain AudioSource.uri (legacy fallback)
playerActionsProvider.playTracks now passes audioCacheManager into
configure() alongside coverCache. setQueueFromTracks awaits the source
build (Future.wait over the track list).
Out of scope: registering an index row when LockCachingAudioSource
finishes downloading (no clean hook from just_audio). Prefetcher /
Download buttons cover the index path.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Listens to mediaItemProvider + cacheSettingsProvider. On track change
(or settings change), computes the [currentIdx, currentIdx + N] window
in the queue and pins each uncached track via AudioCacheManager with
source: autoPrefetch.
Window N comes from cacheSettingsProvider.prefetchWindow (default 5,
configurable in Settings → Storage card).
Smoke-test only at the unit level — real coverage via on-device pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
bvandeusen
