When hardware doesn't natively support a sample rate (e.g. 176.4kHz),
return None from try_create_direct_stream instead of a hard error.
This lets the player fall through to CPAL/rodio which can resample.
Applied to both hw and plughw paths — InvalidParams on rate now
triggers graceful fallback instead of killing playback.
When switching from PipeWire to ALSA Direct, PipeWire holds the ALSA
device open causing EBUSY on the first snd_pcm_open. Now detects
DeviceBusy, suspends the PipeWire sink via pactl, waits 200ms for
release, then retries. Applied to both hw and plughw paths.
cpal 0.15 returned "ALSA" from HostId::name(), cpal 0.17 returns "Alsa"
(now uses stringify! on the enum variant). Use eq_ignore_ascii_case()
for the comparison so it works regardless of cpal version.
Query the DAC's supported sample rates from /proc/asound/cardN/stream0
(maps PipeWire sink to ALSA card via pactl). If the track's native rate
isn't supported (e.g., 176.4kHz), fall back to the nearest rate in the
same family (e.g., 88.2kHz). Rodio resamples from track rate to stream
rate automatically, so audio plays instead of producing silence.
rodio 0.22's with_supported_config() resets buffer_size to Default via
..Default::default(), so our BufferSize settings in StreamConfig were
never reaching CPAL. MixerDeviceSink has zero internal buffering, making
CPAL's buffer the ONLY buffer between the mixer and audio hardware.
Chain .with_buffer_size(Fixed(sample_rate/10)) AFTER with_supported_config()
to actually pass ~100ms buffer to CPAL, matching old vendored cpal behavior.
PipeWire manages its own buffer scheduling via quantum. Forcing a fixed
buffer size (sample_rate / 10 = 19200 frames at 192kHz) conflicts with
PipeWire's internal scheduling and causes buffer underruns that produce
audible noise.
BufferSize::Fixed is kept only for:
- Exclusive mode (512 frames, low latency)
- ALSA backend (direct ALSA, not through PipeWire)
rodio 0.22's MixerDeviceSink has an internal mixer thread that cannot
synchronize with PipeWire's forced quantum scheduling. This caused
massive buffer underruns (600+) at sample rates >= 88.2kHz when Force
Bit Perfect was enabled.
clock.force-rate alone is sufficient for bit-perfect sample rate
switching. Logs confirm zero underruns at 88.2kHz and minimal underruns
at 192kHz (only during heavy I/O) when quantum is not forced.
Replace BufferSize::Default with BufferSize::Fixed(sample_rate / 10)
for non-exclusive mode. This preserves the 100ms period tuning that was
previously applied in the vendored cpal ALSA backend, reducing CPU
wakeups during playback.
- OutputStream/OutputStreamHandle -> DeviceSinkBuilder/MixerDeviceSink
- Sink -> Player (as RodioPlayer)
- current_frame_len -> current_span_len in all Source impls
- Source::channels() returns NonZero<u16>, sample_rate() returns NonZero<u32>
- SamplesBuffer no longer generic, constructor takes NonZero args
- SampleRate is now u32 type alias (no tuple struct)
- Decoder::new_mp4 no longer takes Mp4Type hint
- Applied across both crates/ workspace and src-tauri/ legacy code
Remove vendored rodio and cpal from src-tauri/vendor/ since rodio 0.22
now includes upstream f32 decoder support (our vendored modification).
Update all Cargo.toml files to use rodio 0.22 from crates.io and remove
the [patch.crates-io] section.
- Allow dead_code on TokenValidationResponse.code (deserialized but not read)
- Remove unused MusicProvider import in commands_v2.rs
- Restore #[allow(non_snake_case)] on v2_plex_cache_get_tracks
- Replace drop(tokens) with let _ = tokens (drop on reference is no-op)
Migrate downloaded_purchases to compound PK (track_id, format_id) so
the same track downloaded in MP3 and FLAC are tracked separately.
Album detail endpoint now returns downloaded_format_ids per track.
Qobuz redirects to play.qobuz.com/discover?code_autorisation=XXX
not ?code=XXX. Both the navigation intercept and the /oauth/callback
API call were using the wrong parameter name, so the code was never
captured and the exchange request was malformed.
Add get_all_custom_artist_images() database method and
v2_library_get_all_custom_artist_images command to support
loading all custom artist image overrides at once for the
global resolution store.
The server echoes every state report as a SET_STATE with only next_track
(playing_state=null, current_track=null). These echoes were triggering
align_corebridge_queue_cursor which uses the accumulated renderer state
current_track (from a different context, e.g. iOS). When the track
wasn't found in the local queue, it replaced the entire queue with a
single track, destroying the user's loaded queue.
Now echo SET_STATE commands are detected early and only update the
internal renderer state (for next_track tracking). CoreBridge actions
(align cursor, load track, play/pause) and state reports are skipped
entirely for these echoes.
The server echoes every RNDR_SRVR_STATE_UPDATED as a SET_STATE command
with only next_track updated (playing_state=null, current_track=null).
send_renderer_reports was replying to these echo SET_STATE commands
with another state report, creating an infinite loop that generated
47k log lines (8.2MB) in seconds and completely blocked playback.
Fix: only send state reports for substantive SET_STATE commands that
carry playing_state or current_track changes. Echo commands (next_track
only) are silently consumed.
Also demote diagnostic logging from info to debug to reduce terminal
noise during normal operation.
The server rejects renderer state reports with "Current track not found
in queue nor autoplay" after QBZ-initiated queue loads, because it assigns
non-standard queue_item_ids (first track = track_id, rest = 1..N).
Omit current_queue_item_id and next_queue_item_id from all outbound
renderer reports. The server already knows the current track from
SET_STATE commands and adds the correct queue_item_id when forwarding
to controllers.
Also add info-level diagnostic logging to both report paths (periodic
frontend reports and command-response reports) and the protobuf encoder
for verifying wire content.
Root cause: server pause/resume commands send None for position and
tracks, which the renderer state blindly overwrote. Then renderer
reports sent null position to the server, making controllers show the
track as finished.
Three-part fix:
1. renderer.rs: only overwrite position/tracks when server provides
a non-None value (retain last known values through pause/resume)
2. app.rs + qconnect_service.rs: update renderer state position from
frontend periodic reports so the internal state always has the
real playback position
3. +page.svelte: increase position report frequency from 5s to 2s
for smoother seekbar updates on controllers
ErrorMessage.code was defined as int32 but server sends string codes
(e.g. 'ERROR_QUEUE_INSERT_TRACKS'). Changed to string type and added
logging for queue error responses.
Added context_uuid (random UUID v4) to queue insert and add commands -
the server likely requires this to associate tracks with a context.
track_ids in QueueLoadTracks, QueueInsertTracks, QueueAddTracks, and
AutoplayLoadTracks was int32 packed=false but the server expects
fixed32 (packed by default). This silently broke all outbound queue
manipulation commands.
Server sends track_id as fixed32 (wire type 5) and shuffle_seed as
fixed32, but our structs declared them as int32 (wire type 0/varint),
causing decode failures on QueueState, TracksLoaded, and
RendererSetState messages.
Root cause: inner_payload from CloudPayload.payload IS a QConnectBatch
(QConnectMessages) with repeated QConnectMessage entries in field 3.
The QCloudInnerEnvelope struct incorrectly treated field 3 as opaque
bytes, keeping only the last repeated entry and losing all messages.
Additionally, QConnectMessages.messages_time was declared as uint64
(wire type 0/varint) but the server sends fixed64 (wire type 1),
causing a wire type mismatch that would fail decoding.
Changes:
- Remove QCloudInnerEnvelope struct entirely
- Pass inner_payload directly to decode_queue_server_events/decode_renderer_server_commands
- Fix messages_time from uint64 to fixed64 to match server wire format
Verified against qonductor proto definitions (QConnectBatch) used in
working POC v1 and v2 implementations.
Root cause: server wraps QConnectMessage inside an inner envelope with
fields 1 (correlation_id), 2 (sequence), 3 (actual message bytes).
We were passing the envelope directly to QConnectMessage::decode(),
which silently skipped the unknown wrapper fields and found nothing.
Also restructure bootstrap: controller join first (works without
session_uuid), renderer join deferred until SESSION_STATE (type 81)
arrives with the session_uuid the server requires.
Add missing SRVR_CTRL session management message types (81-87, 97-101)
that were completely absent from the protobuf definitions. Without these,
prost silently discarded all session topology messages from the server.
- Add 12 new proto structs for session/renderer/volume/quality events
- Add decoder branches and map functions for all new message types
- Add SessionManagementEvent variant to bypass queue reducer
- Add v2_qconnect_report_playback_state and v2_qconnect_report_volume
- Add comprehensive logging for inbound message diagnostics
- Add DeviceCapabilitiesMessage to protobuf (tag 7 on DeviceInfo) with
volume_remote_control=ALLOWED, min/max audio quality
- Extend JoinSessionMessage with reason, initial_state, is_active fields
matching Android renderer proto (tags 3-5)
- Bootstrap sends renderer JoinSession with is_active=true and initial
RendererState (STOPPED) so other devices see QBZ immediately
- Send initial StateUpdated, VolumeChanged, MaxAudioQualityChanged
reports after join for full renderer presence
- Default subscribe channels to 0x01,0x02,0x03 instead of empty
- Request queue state from server after bootstrap
cc
The save was broken in two ways:
1. Only current_track + upcoming (capped at 20) were saved, dropping
all history tracks. Queue was reconstructed incomplete on restore.
2. Track metadata (hires, bit_depth, sample_rate, source, etc.) was
not being persisted, only the basic 6 fields.
Add v2_get_all_queue_tracks command that returns the full uncapped
track list with correct current_index. Use it in saveSessionBeforeClose
instead of the capped v2_get_queue_state. Map all metadata fields.
Replace the flat album grid LabelView with a rich detail page featuring:
- Header with label image, name, description, and play button
- Popular Tracks section (5/10/all expand, same UX as ArtistDetailView)
- Releases horizontal scroll row with "See all" → LabelReleasesView
- Critics' Picks section from second releases container
- Playlists, Artists, and More Labels sections
Backend: add /label/page and /label/explore API endpoints with V2
commands (v2_get_label_page, v2_get_label_explore). Models use
serde_json::Value for flexible section parsing.
Old LabelView renamed to LabelReleasesView for the "See all" flow.
i18n keys added to all 4 locales (en, es, de, fr).
vicrodh