Tree search drops its flex-grow and pins to ~140px width. Check and
collapse-all icons shrink from 16px to 14px (2pt) for a tighter
register that matches the row icon size. Toolbar padding tightens
slightly to compensate.
cc
The select-mode toggle drops its backgrounded style for a flat icon
that just changes color when active. New collapse-all button next to
it clears every expanded folder. Tree mode gains its own dedicated
search input alongside the toolbar buttons, decoupled from the
global album search. Folder rows lose the inline (N) track count for
a cleaner Plex/foobar register; the underlying count fetch stays in
place.
cc
The compact album view's Play All / Shuffle buttons drop the pill
treatment and adopt the standard app button style. A new track search
input filters the displayed list by title or artist; empty results
show a short message. New i18n keys added to all five locales.
cc
Flat/Tree switch moves inline with the tabs row (right-aligned, low
weight, only on Folders). Page-level search icon hides on the Folders
tab — flat mode has its own inline search; tree mode gains a dedicated
tree-search in a follow-up. Two-column divider extends up to almost
touch the jumpnav by relocating the toolbar inside the tree column.
cc
The horizontal scrollbar never appeared on .folder-tree-scroll despite
rows declaring width: max-content. Root cause was contain: strict on
the scroll wrapper, which includes size containment; WebKit (Tauri)
under-reported scrollWidth so the bottom scrollbar was suppressed even
when long folder names extended past the rail. Drop size containment
(keep layout + paint for perf) and add flex-shrink: 0 to .row-name so
the flex row doesn't collapse the name and force the row's max-content
width to actually take effect.
The resize handle gains a centered vertical pill (4px x 36px, 40%
opacity by default, accent on hover/drag) so the divider has a visible
affordance independent of cursor rendering quirks under Wayland or
XWayland. Hit area widened to 8px for easier grabbing. Drag/keyboard
behavior and a11y attributes are unchanged.
cc
Default width drops from 432px to 302px (~30% smaller). Long folder
names no longer ellipsize - the sidebar scrolls horizontally inside
itself when content exceeds the column width. Drag-resize and the
persisted width preference are unchanged: the user can still widen
to 40% of the content area, and that choice survives restart.
The resize handle stays anchored to the column's right edge because
it sits as a sibling of the scroll wrapper (.folder-tree-scroll),
not inside it - horizontal scroll inside the wrapper does not drag
the handle along.
cc
Tighter row padding (3px), smaller typography (~12.5px), 12px depth
indent, 14px icons. Removes the artwork thumbnail in favor of always
showing the folder icon. Track count moves inline as a muted suffix
(N) instead of a subtitle row, hidden when zero. Visual register
matches Plex/foobar2000 file-tree conventions.
cc
New LocalLibraryFolderAlbumView renders a stripped album detail
(artwork + metadata + play + track list) sized for the tree-mode
right pane. The full-page album-detail view is unchanged; tree mode
no longer triggers the page-level takeover when an album folder is
selected, so the tree rail stays visible while browsing.
Tracks load into a separate `treeAlbumTracks` state via $effect on
`selectedAlbumForTree` (raceless: commits only when the selection
still matches at resolve-time). Play / shuffle / track-click set the
playback context against the tree-selected album so the now-playing
source label remains accurate.
cc
Sidebar default widens from 240px to 432px. A 6px right-edge handle
resizes between 200px and 40% of the content area. Width persists
across restarts via v2_set_library_folders_tree_sidebar_width on
drag-end. ResizeObserver tracks content-area changes (window resize +
app sidebar toggle) and re-clamps the user width when the cap shrinks.
Keyboard parity: arrow keys nudge in 16px steps (64px with Shift),
Home/End jump to bounds. Aria semantics use role="separator" with
valuenow/min/max. Five-locale i18n for the handle label.
cc
New i64 field on LibraryPreferences for the tree-mode sidebar width
the user chooses via drag. Migration adds the column non-destructively;
NULL means "use default".
cc
The view used to hardcode `activeTab = 'tracks'` on mount, which on
large libraries (16K+ tracks) causes a startup freeze regardless of
the user's preferred ordering. Now `activeTab` initializes to the
first visible tab in the user's configured order. Existing reactivity
that follows tab visibility changes (e.g. when the user hides their
current tab) is preserved.
cc
Top-level scan-root rows now fetch their recursive descendant count
via v2_library_count_folder_tracks instead of showing a hardcoded 0
(closing the Task 7 trade-off). Cache invalidates when the exclude-
network filter flips. The tree itself loses its per-row play button:
navigation-only, with recursive playback available from the right-
pane detail view.
cc
Lightweight COUNT(*) helper for the tree-mode scan-root rows that
needed a recursive descendant count. Same network-folder filter as
the listing primitives.
cc
All tree-mode invokes route through shouldExcludeNetworkFolders() so
the tree rail mirrors the flat list and recursive-play boundary.
Closes the visible/play semantic mismatch flagged by Task 10.
cc
v2_library_list_folder_children, v2_library_list_folder_tracks,
v2_library_list_folder_tracks_recursive accept and forward the
exclude-network-folders flag.
cc
list_folder_children, list_folder_tracks, list_folder_tracks_recursive
now apply the same network-folder exclusion that flat-mode search
honors, so the tree rail and recursive playback see the same source
of truth. Tests cover the on/off cases.
cc
handleBulkPlayNext, handleBulkPlayLater, handleBulkAddToPlaylist now
call resetMultiSelect() so treeSelectMode and treeSelectedTracksById
are cleared alongside the existing flat-mode reset. Without this, the
BulkActionBar lingered with count 0 after firing from tree mode. The
tree-mode BulkActionBar's onClearSelection closure now delegates to
the same helper for consistency. resetMultiSelect was already
correct (it leaves trackPathById intact, which is what we want).
Also: warn when toggleTreeTrackSelection's cache-miss fallback hits a
track path with no parent folder, instead of silently returning.
cc
Verify the recursive listing returns an empty Vec (not an error) when
the requested folder path has no descendants — frontend treats empty
as "nothing to play/queue" and skips the toast.
cc
Folder-row checkbox in tree mode selects every descendant track via
v2_library_list_folder_tracks_recursive, deselects them on a second
click, and shows the standard indeterminate state when only some
descendants are selected. Track-kind rows in the tree honour the
same flow so users can deselect individual tracks from a partially-
selected folder. Both checkboxes share the existing selectedTrackIds
set, so the bulk-action bar (play next, play later, add to playlist)
operates on the union without any special-case handling.
Indeterminate state is computed cheaply: a SvelteMap<number, string>
caches track-id -> file_path (populated from the in-memory tracks
list and from each recursive fetch on toggle), and `countSelectedUnder`
walks the small selectedTrackIds set checking the prefix. Folder-row
state derives via a parent-supplied callback; the child reads the
prop reference of selectedTrackIds so reactivity propagates when the
parent reassigns the set.
Adds a dedicated treeSelectMode flag so the flat-mode tracks-tab
selection toggle stays independent. Renders the BulkActionBar at the
end of the tree-mode container when treeSelectMode is on. Adds a new
toast.failedSelectFolderTracks key across all five locales.
cc
Tauri wrapper for the qbz-library primitive added in the previous
commit. V2 cleanroom: direct call into qbz-library, no legacy
delegation. Registered in lib.rs alongside the existing folder-tree
commands.
cc
Returns full LocalTrack rows for every descendant under a folder path,
filtered by COALESCE(source, 'user') = 'user' so qobuz_download rows
never leak. Ordering is file_path ASC for stable on-disk reading
order across heterogeneous subtrees. Reuses escape_like_pattern so
literal '_' / '%' in path segments don't behave as LIKE wildcards.
Used by the tree-mode multi-select to populate selectedTrackIds when
the user ticks a folder-row checkbox; returns the full record (not
just IDs) so the frontend can build queue items without a second
round-trip.
Adds three unit tests covering the recursive walk, file_path
ordering, and special-character escape handling.
cc
When the user types in the search input while in tree mode, matching
folders/tracks stay visible and every ancestor path expands to keep
context. Empty branches collapse out. Clearing the query restores the
user's pre-search expand state. Implementation: in-memory contains over
the loaded `albums` array, debounced 200ms. The match expands the union
of matched paths and their ancestors in `searchExpandedPaths`, which
replaces `treeExpandedPaths` while search is active. Pre-search snapshot
is restored on clear. Tracks remain rendered under any visible folder.
cc
- Surface a toast on playback failure so the user gets UI feedback
rather than only a console error.
- Honor shouldExcludeNetworkFolders() for consistency with loadTracks
and the rest of the view's search invocations.
cc
Wires LocalLibraryFolderTree (left rail, 240px fixed) and the
LocalLibraryFolderDetail / album-detail flow (right pane) into a
two-column shell that mirrors the Artists tab CSS byte-for-byte.
View-mode segmented control on the Folders toolbar toggles between
Flat (the existing folder-grouped list) and Tree (the new hierarchical
view). The toggle is hidden when browsing Hidden Albums so that
sub-view keeps its existing flat rendering.
Selection routes the right pane: folder paths matching a known
album_group_key drive the existing handleAlbumClick flow (which sets
selectedAlbum and renders the top-level album-detail view); non-album
folders open LocalLibraryFolderDetail with subfolder cards plus
direct-children tracks. Empty-state when nothing is selected uses the
library.foldersTree.selectAFolder string.
Recursive play: handlePlayRecursive fetches all tracks via
v2_library_search (limit:0), client-side prefix-filters on file_path,
sorts by path and queues the result. Backend does not yet expose a
recursive-tracks-under-path command — flagged as a Task 7 trade-off
to revisit if telemetry shows tree-mode play stalling on large
libraries.
Mode persists in library_preferences.folders_view_mode via the
existing v2_set_library_folders_view_mode command, hydrated on mount
through v2_get_library_preferences. Flat mode rendering is
byte-for-byte unchanged: zero deletions in the existing markup, the
new tree shell wraps it via an outer if/else.
LibraryPreferences type extended with the optional folders_view_mode
field so TS strict-mode passes on the new prefs.folders_view_mode
read.
cc
The folderPath snapshot existed but wasn't used to guard the async
writes. Rapid folder switches could land an older fetch's results on
top of the newer selection. Add three short stale-guards on .then,
.catch, and .finally so only the latest in-flight result wins.
cc
Renders the contents of a folder that is NOT a known album_group_key —
shows immediate subfolders + direct-children tracks side by side, with
a "Play all recursively" button for the whole folder. Used in the right
column of the upcoming tree-mode two-column layout (Task 7). Extracts
FolderTreeEntry to a shared $lib/types/folderTree.ts module so both
LocalLibraryFolderTree and FolderDetail import the same type.
cc
- Keep self-import for recursion and replace the misleading "deprecated
in Svelte 5" comment with a corrected rationale. svelte-check confirms
`<svelte:self>` is in fact deprecated in Svelte 5 (svelte_self_deprecated)
and self-import is the recommended replacement; the original code was
right, only the comment was wrong.
- Switch the row role from `button` to `treeitem` so that `aria-selected`
is valid (svelte-check rejects aria-selected on role=button) and add
`aria-expanded` for folders. treeitem is also the semantically correct
role for a tree row, so screen readers now announce both selection and
expansion state.
- Add :focus-within parity for the play button so keyboard users see it
when the row receives focus.
cc
Lazy-loaded recursive tree-row component for the upcoming tree mode in
the Folders tab. Renders one node per call; children loaded on first
expand via v2_library_list_folder_children. ADR-001 compliant (no $t
inside $derived). Indentation by depth * 16px matches Poweramp-style
hierarchy. Recursion uses self-import pattern (svelte:self deprecated
in Svelte 5). Not yet wired into LocalLibraryView - that lands with
the two-column layout integration (Task 7).
cc
Task 4's structural change of `library.folders` from string to object
broke LocalLibraryView.svelte:4337's dynamic tab label
($t(`library.${tab}`)) which expects a string. Move the new tree keys
to a sibling namespace `library.foldersTree.*` instead, and restore
`library.folders` to its original flat string value.
cc
Five locales (en, es, de, fr, pt). en and es are human-written in the
project's canonical Mexican neutral register; de/fr/pt are machine-
translated drafts flagged in qbz-nix-docs/i18n/pending-translations.md
for community review. Seven keys: modeLabel, flatLabel, selectAFolder,
treeFolderTracks (ICU plural), treeLoading, treeEmpty, playAllRecursive.
The flat library.folders string ("Folders" / "Carpetas" / etc.) was
migrated to an object; the original label is preserved at
library.folders.label. No source code currently references the key,
so the restructuring is non-breaking. "Track" is translated to match
each locale's existing convention (pista, Titel, titre, faixa).
cc
New persisted field on LibraryPreferences gating the upcoming Tree
sub-mode in the LocalLibrary Folders tab. Default flat preserves
existing behaviour for current installs. Migration adds the column
non-destructively. Setter v2_set_library_folders_view_mode exposes
the toggle to the frontend; the existing v2_get_library_preferences
returns the new field automatically via serde.
cc
- Replace the false "SQLite ORDER BY fragility" rationale on the
Rust-side sort in list_folder_children with the actual reason
(rows already in memory; Rust case-insensitive compare clearer
than COLLATE NOCASE on a CASE-derived column).
- Drop the redundant ESCAPE '\\' clause on the literal '%/%' pattern
in list_folder_tracks. The pattern contains no escape character;
the clause was a copy-paste artifact from the user-input LIKE
patterns where ESCAPE is required.
No behaviour change.
cc
Two new SQL primitives that walk local_tracks.file_path to reconstruct
filesystem hierarchy without schema changes. Both filter
COALESCE(source, 'user') = 'user' so Plex and Qobuz downloads are
excluded from the eventual tree-mode view in the Folders tab. Both hit
idx_tracks_file_path for the LIKE prefix% portion; per-row level
extraction is in-memory over the prefix-narrowed candidate set. New
FolderTreeEntry enum carries the child rows. Unit tests cover folder/
track interleaving, source filter, special-character paths, and
direct-children vs recursive scoping.
cc
Issue #375: pause-then-resume sometimes leaves audio silent. Diagnosis
in user-reported logs:
Synced audio settings to CoreBridge player: backend=None, exclusive=false
...
Resume: reinitializing stream at 44100Hz/2ch
Using default audio device <- legacy CPAL path
Using audio device: Default Audio Device
Audio output initialized successfully <- silent stream
backend_type=None falls through to init_device's legacy CPAL path which
races against rodio's DeviceSink drop on resume — CPAL hands us a new
default-device handle, but PipeWire is still tearing down the previous
sink, and the new handle ends up routed to a node that produces no
sound. CPAL doesn't fail the open, just gives a silent stream.
The "Sync audio settings on startup" toggle Rico29 was asked to enable
fires correctly (`sync_audio_on_startup=true` in his log) but only
re-syncs whatever is persisted. His persisted backend_type was still
NULL because he never explicitly picked a backend, so the sync was a
no-op.
Fix:
- Default::default() now sets backend_type to Some(AudioBackendType::default()),
which picks PipeWire on Linux and SystemDefault elsewhere. New installs
go directly onto the proper backend init path; the runtime still
degrades gracefully if the chosen backend isn't available on the host
(try_init_stream_with_backend returns Some(Err(_)) and the existing
fallback warning fires).
- Migration backfills existing rows where backend_type IS NULL with the
same per-OS default, serialized via serde_json so the row read path
round-trips cleanly. macOS users were already getting effective
SystemDefault dynamically at the call site (audio_settings.backend_type
.or_else(...)) so behaviour there is unchanged; the migration just
persists what the runtime was already substituting.
reset_all picks up the new default automatically because it reads
AudioSettings::default().backend_type.
cc
Three issues observed when fast-scrolling collection detail views
(reported on the Macross collection, ~67 items, list mode):
1. Horizontal scroll artifact at the bottom-right corner — same root
cause as AlbumDetailView's: an absolutely-positioned artwork backdrop
widens past 100% of the .detail-view. Add overflow-x: hidden alongside
the existing overflow-y: auto.
2. Empty container when scroll position briefly exceeds the content
tail. The virtualWindow math could compute firstRow * cols >= total
during overscroll moments, slicing windowedItems to an empty array.
Clamp firstRow / firstIdx to total - 1 so at least one row renders
even when the scroll position has overshot.
3. The user-visible regression that started this thread: scrolling
UP from deep in the list could not reach scrollTop = 0 even though
the "Top" button worked. Two fixes layered together:
a. overflow-anchor: none on .detail-view. Browser scroll-anchoring
tries to preserve visual stability when content above the viewport
changes height — but the virtualizer manages its own spacer
heights, so the auto-anchoring nudges scrollTop in the opposite
direction of user input. Disabling it lets the wheel input win.
b. content-visibility: auto gated to expanded mode. Until now
.item-block applied it unconditionally. The grid view's .grid-card
already had this exact issue diagnosed and fixed (see the inline
comment "doubling up with content-visibility: auto caused cards to
re-eject and re-paint on scroll-back"); .item-block had been left
behind. Layering native content-visibility on top of manual
virtualization caused the browser to intermittently treat rendered
rows as off-screen during fast scroll, snapping them to the
contain-intrinsic-size hint for a frame, which the spacer math
doesn't predict — producing the empty-viewport-with-stuck-scroll
symptom. The .lazy class now scopes content-visibility to expanded
mode, where virtualization is OFF and the optimization is needed.
Smoothness during very fast wheel scroll is still imperfect but the
viewport no longer goes blank and scroll-up reaches the top reliably.
cc
The .has-art-bg ::before pseudo-element renders the album artwork
heavily blurred at transform: scale(1.15) for the Qobuz/Feishin-style
header. Without overflow-x: hidden on the .album-detail container, the
15% horizontal upscale of the absolutely-positioned backdrop bleeds
past the right edge and surfaces a horizontal scrollbar on the page.
Add overflow-x: hidden to the existing overflow-y: auto so the blur
stays clipped without affecting vertical scroll.
cc
Adds CoreAudio Hog Mode integration on macOS so Exclusive Mode actually
takes the device exclusively (previously the macOS path used the
Linux-shaped logic without any kernel-level lock). Pins the CoreAudio
device id through Hog Mode acquisition, sample-rate switching, and
stream creation so System Default doesn't drift mid-session. Linux
behaviour is unchanged.
Review asks (all four resolved by Vudgekek + afonsojramos):
- set_coreaudio_hardware_volume falls back to software volume on Err
rather than reporting success, so knob-only DACs keep a working
volume slider via engine.set_volume().
- CoreAudioExclusiveGuard snapshots the device's hardware volume on
acquire and restores it before releasing Hog Mode, so leaving
Exclusive Mode returns the DAC to the user's pre-Exclusive level.
Guard is constructed before the FFI call so partial-acquire failures
still trigger Drop and release.
- Linux unused-variable warnings on stream_opt and exclusive_guard
cleaned up via cfg_attr(allow(unused_variables)) on the macOS-only
guard plumbing. cargo check is clean.
- Silent shared-mode regression on macOS when backend init fails:
init_device now sets_current_device(None) and returns None on
exclusive_mode==true so the user sees an error instead of
unknowingly playing in shared mode.
Bonus: 5f10b4ba adds real CoreAudio device-rate detection for macOS
Shared Mode, +90 lines in backend.rs.
Conflict resolution: single textual collision in
crates/qbz-audio/src/lib.rs (both branches added a pub use line).
Resolved by keeping both, alphabetical order — coreaudio_direct
re-export from pr-391, device_reservation re-export from pre-release.
All other auto-merges verified line-by-line: AlsaDirectStream's
DeviceReservation::acquire flow (alsa_direct.rs:60-97), the macOS
exclusive_mode guard (player/mod.rs:1077-1085 in qbz-player and the
mirror in src-tauri:1024), and CoreAudioExclusiveGuard's volume
snapshot/restore (coreaudio_direct.rs:351,369,373,388) all coexist
correctly because each branch touched different functions.
cargo check clean across all four crates. 9 device_reservation +
3 reserve_dac + 214 src-tauri unit tests pass.
cc
Mirrors the HiFi Wizard control (landing in Task 8) inline under
Exclusive Mode. Status row polls v2_get_dac_reservation_status every
5s while the row is rendered. Hidden when the selected device is not
a card-specific ALSA prefix (hw:, plughw:, front:, surround*, iec958:,
hdmi:). Six i18n keys consumed at settings.audio.reserveDac.*.
cc
Five locales (en, es, de, fr, pt). English and Spanish are human-written
in the project's canonical Mexican neutral register; de/fr/pt are
machine-translated drafts. Pending-translations note recorded in
qbz-nix-docs/i18n/pending-translations.md (outside git scope).
Keys: settings.audio.reserveDac.{label, description, statusActive, statusContested, statusUnavailable, wirePlumberNote}.
UI surfaces (Settings mirror + HiFi Wizard) consume these in Tasks 7-8.
cc
- Document the Contested-vs-Unavailable classification limitation on
the DacReservationStatus enum and its call site, deferring the
stash-the-last-error refactor until Tasks 7-8 confirm the UI needs
the distinction.
- Lift is_card_specific_device to pub(crate) in lib.rs and delete the
duplicate in commands_v2/audio.rs to prevent the allow-list from
diverging when new ALSA prefixes are added.
- Drop the misleading "or non-card device" clause from the
apply_dac_reservation degraded-path log; non-card devices are
filtered earlier and cannot reach this log line.
No behaviour change.
cc
Lifetime B from the design spec: a long-lived DeviceReservation guard
held in AppState, gated by the reserve_dac_while_running setting.
Acquired at startup if the persisted flag is true and the selected
device targets a single ALSA card; reapplied when the user picks a
different DAC; released on toggle-off, on a switch to a non-card
device, or AppState drop.
DB persistence: ALTER TABLE migration adds the column, SELECT and
reset_all are extended, set_reserve_dac_while_running setter parallels
the established pattern for boolean audio settings.
Tauri commands (V2 cleanroom):
- v2_set_reserve_dac_while_running(enabled) - persist + apply
- v2_get_dac_reservation_status() -> DacReservationStatus
Status enum mirrors the four cases in the spec: Inactive (toggle off,
or non-card-specific device), Active (we hold the bus name), Contested
(another app holds the device at higher priority - guard is None but
the toggle is on), Unavailable (D-Bus unreachable, degraded guard).
v2_set_audio_output_device gains an AppState dependency so DAC changes
trigger a reservation handover when the toggle is on.
UI surfaces (Settings mirror + HiFi Wizard) land in Tasks 7-8.
cc
Opt-in flag for the per-process DAC reservation lifetime (Lifetime B
in the spec). Defaults to false; serde default ensures legacy settings
JSON files round-trip cleanly without the field. Task 5 will read this
flag in AppState startup and the v2_set_reserve_dac_while_running
Tauri command.
Strictly additive: no existing field changed, no behaviour change for
users with the flag at its default (false).
cc
The CARD= parameter in ALSA device strings refers to the card's short
id (e.g. "C20", "HDMI", "Generic"), not its long descriptive name
("Cambridge Audio USB Audio 2.0"). The previous resolver called
Card::get_name() and silently swallowed errors, so every CARD=<id>
lookup failed and acquire() degraded gracefully -- meaning the
reservation has been non-functional for users with named-card device
strings since Task 3 landed.
Reported by the maintainer on Cambridge DacMagic Plus exposed as
front:CARD=C20,DEV=0, 2026-05-07. Smoke-test scenario 2 had been a
false positive: audio worked because we degraded silently, not because
the reservation succeeded.
In alsa-rs 0.10, Card::get_name() wraps snd_card_get_name and despite
the name returns the LONG descriptive string. The short id (what
aplay -l shows in brackets and what CARD= references) is exposed via
ctl::CardInfo::get_id() (snd_ctl_card_info_get_id). Open a Ctl per
card, read the id from its CardInfo, and surface ALSA enumeration
errors as ReservationError::AlsaError instead of unwrap_or_default()
into an empty string.
cc