A paused or stopped QMediaPlayer on Qt's FFmpeg backend keeps its PipeWire
stream open and never corks or drains it, so the few seconds the backend
decoded ahead sit there live. A later audio-graph change — a USB DAC waking
from idle suspend, a device appearing — flushes that stale buffer to the
speakers, hours after the app was last touched.
That is the "LinTunes plays by itself" haunting: the v0.1.4 provenance log
recorded zero control events between the 17:48 pause and the 00:30 incident,
while MPRIS still reported Paused at exactly 6974000us — the position it was
paused at seven hours earlier — and PipeWire showed the stream state=running.
It explains the whole shape of it: always a short burst (only what was
buffered), always mid-song, never a recorded play count.
After 30s idle, release the pipeline: stop() then setSource(QUrl()), which
removes the PipeWire node outright, so no buffer survives to leak. Gain drops
to zero first as an independent second layer. stop() arms the brake too — Qt
leaves the source loaded there, and _advance() takes that path when a playlist
runs out. Resume rebuilds the source and seeks back via the existing
custom-start-time machinery.
Verified end-to-end against the real LocalSink with a silent WAV, watched
through pw-dump: present while playing, still present right after pause, gone
once parked, rebuilt on resume with position and duration intact.
This is a workaround for an upstream Qt Multimedia bug; TASKS.md now carries
the assessment for moving the local sink to GStreamer, which gets correct
pause behavior for free and would unlock the parked equalizer and gapless.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Right-click a track (or a multi-selection) for "Remove from Library" or
"Remove from Library and Delete File". The second moves the file to the
desktop trash rather than unlinking it, so it stays recoverable by Ctrl+Z
in-app and by "Restore" from the file manager afterwards.
The trash is per-filesystem: music lives on a mounted volume, so the file
belongs in <topdir>/.Trash-<uid> with a topdir-relative, percent-encoded
Path. Using ~/.local/share/Trash would be a cross-device copy recording an
original path "Restore" can't reach. lintunes/trash.py implements the
freedesktop spec directly rather than adding a dependency that would need
a manual reinstall on the other machine.
Playlist cleanup is synchronous with the removal: playlist edits address
tracks by row index into track_ids while the view skips ids missing from
the library, so a dangling id would desync the two and make a later
"Remove from Playlist" hit the wrong track.
A file that can't be trashed keeps its track — better a song to delete
again than a library entry orphaned from a file still on disk. Player
gains drop_tracks() so deleting the playing track stops cleanly instead
of erroring out when the queue walk later reaches a dead id.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A "✕" button sits left of the sync progress group and expands to
"cancel transfer" on hover; clicking asks Cancel Transfer / Keep
Copying (the copy keeps running under the dialog). Cancel stops the
worker at the next chunk boundary, removes the in-flight partial file,
and rewrites the m3u to list only tracks actually on the device — so a
cancelled sync always leaves a coherent partial playlist. That last bit
also closes a pre-existing gap: stale files are deleted before copying,
so an interruption could previously leave the old m3u pointing at
deleted files.
Recoverability proven both ways: a test asserts cancel-then-resync ends
byte-identical to an uninterrupted sync, and a live run against the
real Rabbit (cancel mid-file-2 of 3) recovered with kept=1/copies=2.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Quitting mid-transfer now asks (Keep Syncing / Quit Anyway) from
closeEvent, which every quit path hits — including the self-update
restart, which previously bypassed closeEvent via a bare quit() and now
routes through close(). A running sync holds its own GNOME inhibitor
(suspend + logout, "Syncing a playlist to a device") so the machine
won't sleep or shutdown/restart under a copy.
Along the way: the Inhibit D-Bus call marshaled Python ints as signed
against GNOME's (susu) signature, so every call was rejected and the
playback sleep inhibitor had silently never worked — _uint fixes both
holders (verified live: cookies taken, IsInhibited flips, releases
clean).
Also investigated trav's mid-copy Syncthing question: the transfer works
from a click-time snapshot and never reads live library state, so remote
changes can't corrupt it — no state freeze needed. Hardened the one real
gap: a source file relocated under the queue (remote metadata edit
moving files) is now skipped, dropped from the m3u, and reported,
instead of aborting the whole sync.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The QProgressBar overlay text didn't fit at 160-220px and clipped to two
broken lines. Split it: a plain label "Copying to Rabbit R1 · 1.2 GB /
3.4 GB" next to a textless bar, both permanent right-side widgets hidden
when idle. And the Device menu now sits to the right of Track (trav's
intended spot).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
New Device menu (left of Track) with "Sync Playlist to Rabbit", enabled
only when a playlist is showing and the Rabbit is plugged in. The Rabbit
mounts over MTP/gvfs (not mass storage), so device_sync.py drives it
with plain file I/O honoring the MTP caveats: no copystat, diff by
name+size. One-way mirror into Music/<Playlist>/ on the device — stale
files deleted (that folder only), Auxio-importable .m3u carries the
order, free space checked up front with a needed-vs-available alert, and
a right-justified status-bar progress bar tracks the chunked copies.
Verified live against the real Rabbit end to end.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
LinTunes has spontaneously resumed playback for ~4s (then paused) four
times while trav was away — suspected phantom media-key events from the
Audioengine USB DAC's HID 'keyboard', but the pathway (MPRIS vs the
focused-window key filter) is unproven. Every control path now drops a
timestamped line in ~/.cache/lintunes/control-events.log: MPRIS Player
methods, the media-key eventFilter (with source input device where the
compositor exposes it), and Player.toggle_play/pause with position. New
lintunes/eventlog.py, 1MB rotation, never raises; conftest autouse
fixture keeps tests off the real log file.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
QMediaPlayer.setSource() no-ops on an unchanged URL, so replaying the
track that was already the loaded source never re-fired LoadedMedia and
the armed _pending_start_ms was silently dropped — a freshly edited
start time appeared to save but playback began at 0:00. _load_current
now detects the unchanged source and stop()+seeks directly (a forced
clear+reload was tried and races the FFmpeg backend, snapping the seek
back to 0). previous()-restart likewise rewinds to the custom start
time instead of 0:00.
Verified live via an offscreen GUI harness: replay-while-playing,
replay-after-EndOfMedia, and cleared-start-time replay.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Prev/play/next (and shuffle) used to hug their ~20px glyphs in the middle
of the 62px rounded box, so most of the bubble was dead space. _box() now
takes split=True: the buttons tile the bubble interior in equal shares,
expanding to the (tightened) margins with the glyph centered in each.
Minimum button widths keep the bubbles at their old footprint — no bigger,
no smaller. Round 21 tests cover the equal split, vertical fill, edge
spans, and the preserved footprint.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Diagnosing the Ctrl+I save hang on the Debian machine (GUI thread blocks
past mutter's 5s check-alive):
- lintunes/perf.py: timed() context manager logging wall-clock ms at INFO
on lintunes.perf — instruments tag/artwork saves, file moves,
edit_track(s)_fields, the browser rebuild, smart recompute, and the
debounced JSON flush, so the slow machine can tell us which stage
eats the time (journalctl --user or a terminal run).
- tagging.write_tags: ONE parse + ONE save per edit. grouping/
compilation/bpm ride the main save via registered Easy keys
(GRP1/TCMP/TBPM on EasyID3, cpil on EasyMP4) instead of
_write_extra_tags re-parsing and re-saving the audio file.
- LibraryView: browser rebuilds coalesce through a 0ms single-shot
timer — an N-track Get Info edit rebuilds the genre/artist/album
cascade once instead of N times over the whole library.
Moving the writes off the GUI thread is deliberately deferred until the
Debian timings say which stage dominates.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CLAUDE.md's never-move rule rewritten for the Round 18 rename-move
feature; TASKS.md New backlog folded into a checked-off Round 18
section (plus trav's by-eye verify list). The equalizer item moves to
Parked/deferred with trav's bass/mid/treble idea and the QMediaPlayer
no-effects-hooks blocker recorded.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Rediagnosed (was thought Bluetooth-specific): on the Debian 13 machine,
resuming after the machine sat idle plays nothing even though decoding
runs (visualizer moves) — the OS suspends the audio sink and it comes
back dead until a seek re-primes it. Resuming after a pause of 30s+ now
does a seek-in-place before play(), automating the manual "rewind
slightly" workaround without losing the position. The volume re-apply
from the earlier fix stays (harmless, covers device swaps).
Also fold the new backlog items into the TASKS.md board.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Features:
- Rating hover dots: hovering a rating cell shows 5 clickable slots
(RatingDelegate); click slot k sets k stars, clicking the current count
clears. Undoable, library-only (ratings never rewrite music files).
- Search bar moved into the Library header strip, right of the Library
button, so the tracklist top aligns with the playlist tree.
- Right-click "Download Album Art…": iTunes Search API (no key), off-thread
fetch, preview/confirm dialog with Next Result, embeds via write_artwork
+ size refresh, invalidates the MPRIS art cache.
Fixes:
- MPRIS media keys: PropertiesChanged sent invalidated_properties as "av"
instead of "as", so gsd-media-keys dropped it and never MRU-bumped
lintunes (why the play/pause key kept waking stale players). Now an
explicit QDBusArgument string array; loopback-verified sa{sv}as.
- Exit segfault: ordered Player.shutdown() (stop, clear source, detach
buffer/audio outputs) from closeEvent/aboutToQuit; scripted run exits 0.
- BT zero-volume after pause/resume: volume re-applied on resume, device
swap, and BufferedMedia (needs verify on the affected machine).
Cruft sweep:
- Tag writes filtered to EDITABLE_FIELDS; failures logged + surfaced in
the status bar (was a swallowed print).
- O(n²) import fixed (location index + cached max track id); O(1)
refresh/reveal via TrackTableModel row index.
- lastfm: scrobble-queue thread lock, login prefs write marshalled to the
GUI thread, JSON/raise_for_status order fixed.
- Shared read_json/write_json; TableSettingsMixin dedupes view settings.
TASKS.md rewritten as a resumable board; tests in tests/test_round17.py
(251 total pass).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Both machines now reflect each other's changes within a couple seconds, and
genuine Syncthing conflicts auto-merge with a backup and an alert.
- sync_watcher.py: QFileSystemWatcher (debounced, re-arms after atomic renames)
emits a single `changed`; the manager decides if it was external.
- library_manager: `reload_from_disk()` re-reads and reconciles disk into memory
(max play/skip counts, newest edit wins, playlist membership unioned), keeping
local unsaved edits and object identity so open views stay valid, then refreshes
the UI without touching the player. `flush()` records each file's (mtime, size)
so our own writes are never mistaken for an external change. New signals
library_reloaded / conflict_resolved.
- conflict_resolver: back up BOTH sides into .resolved/<ts>/{original,incoming}/
before merging, return list[ConflictSummary], add restore_backup(); runs at
startup and live.
- gui/conflict_dialog.py: modeless summary with Open/Restore backup; main_window
shows it from a status-bar notice and preserves scroll+selection on reload.
Tests: tests/test_round16.py (9) + full suite green (225).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The library is going live as the canonical store, synced to a second machine
via Syncthing. Track locations were stored absolute and only remapped at import,
so on a second machine (where Syncthing mounts the folder at a different path)
every location would break.
- lintunes/paths.py: to_relative/to_absolute. Locations are stored relative to
the data dir and resolved back on load, applied only at the json_storage
boundary (save_tracks/load_library). Track.location stays absolute in memory,
so the player, tagging, and art code are unchanged. The data dir and the music
move together inside one synced tree, so paths resolve wherever it's mounted —
no per-machine music_root config. Absolute paths in older library.json files
still load (back-compat).
- tests/test_round15.py: helper round-trips, back-compat, and a machine-2
scenario (save under root A, load the copied tree under root B).
- TASKS.md/tasks-done.md: mark the data-dir move + real import done; log the
benign exit-time Qt/FFmpeg teardown segfault.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
While a track drag hovers within 28px of the track list's top/bottom
edge, the list scrolls that way (40ms timer) so the user can drop onto
off-screen rows. Qt's built-in autoscroll never fires because our
dragMoveEvent accepts the event itself, so this is driven manually; the
drop line refreshes as content scrolls under a stationary cursor.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Cut (Ctrl-X) marks the selection on the clipboard with a `cut` flag; a
paste of a cut from a real playlist is a *move* (single-undo across
playlists via move_tracks_between_playlists; same-playlist reorders),
while a copy — or a cut from the library, which has nothing to remove —
just adds. Paste now inserts *above the selected track* rather than
appending. Cut/Paste added to the track context menu; parse_tracks_mime
now tolerates a None/empty clipboard.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Five grayscale (black->white) sliders in Preferences that update live:
outer-chrome background (QPalette.Window), the rounded transport button
boxes, the now-playing panel, the now-playing text, and the row stripes
(QPalette.AlternateBase). Defaults are None ("inherit current") so nothing
changes until the user drags. Decouples the button boxes from the stripe
color (they used to share palette alternate-base). Live preview via
Preferences.set_live (no disk write); persisted ~0.4s after the drag.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Snapshot of the existing codebase before working through the TASKS.md
backlog. Real library data (data/) and the iTunes import fixture
(itunes-test-library/) are gitignored.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
travandClaude Opus 5