The track's "ShaderEffect / ColorOverlay / MultiEffect render NOTHING on the
software path this port runs on" doctrine was measured FALSE on 2026-07-29:
qbz-qt runs on the GPU (OpenGL RHI, threaded render loop, vsync 5.56ms). The
claim came from an offscreen session, and QT_QPA_PLATFORM=offscreen forces
software by definition. Every design it justified needed re-deciding.
- RoundedImage was a Canvas (renderTarget: Canvas.Image) that rasterised with
QPainter on the CPU AND uploaded a texture per repaint, for every cover in
the app, purely to round a corner identical in every frame of that image's
life. It is now layer.enabled + MultiEffect { maskSource } — an FBO rendered
once and cached for static content. Rectangle { radius; clip: true } cannot
do this: clip in Qt Quick is a RECTANGULAR scissor and ignores radius.
maskThresholdMin/maskSpreadAtMin are 0.5/1.0, NOT the 0.0/0.0 that looks
natural: 0.0 collapses Qt's mask smoothstep to step(0, alpha) and every
pixel with any alpha snaps fully opaque — measured 0 antialiased pixels at
radius 40 against the Canvas's 292. At 0.5/1.0 it is 308, ramp [0,51,219,255].
A software machine keeps the Canvas arm (GraphicsInfo.api), and
QBZ_QT_ROUND_MODE=canvas pins it on a GPU machine where the mask misbehaves.
- VizSettle drove the spectrum from FrameAnimation, i.e. every rendered frame:
180fps on this box for data the producer publishes at TARGET_FPS = 30. Now a
16ms Timer; the wall-clock maths is unchanged.
- SpectrumBand's per-bar y was a SECOND JS binding derived from height. An
anchor does the same arithmetic in C++ with no binding at all, and each FFT
frame is materialised into a plain array once per publish instead of being
read element by element across the JS/C++ boundary. Bindings per tick: bars
28 (was 56), waveform 48 (was 96), energy 5 (was 10).
QQuickPaintedItem was investigated and rejected with evidence: cxx-qt 0.7.3
emits X(QObject* parent) : base(parent), which does not compile against
QQuickPaintedItem(QQuickItem*), paint() is pure virtual, and there is zero
precedent for a custom base class in this workspace. Qt 6 ShaderEffect takes
no GLSL, only precompiled .qsb, which would make the build depend on a Qt
tool whose path differs on macOS.
- PlaylistCollage point-sampled its mosaic in its own Canvas
(Context2D.drawImage does not filter when it scales — measured RMSE 0.000
against nearest-neighbour). It now draws through RoundedImage.
- The scaled-derivative cache had no eviction at all and the .jpg -> .png key
change had orphaned 1251 files. Boot housekeeping sweeps the orphans once and
bounds the directory at 64 MB, LRU by mtime. Verified by effect: first run
deleted 1259 files / 10.05 MB, every run since reports 0, and the directory
is 14 MB. The eviction is only survivable because RoundedImage clears
_scaled on Image.Error and re-requests.
- For You artwork patched the rails in place instead of republishing the
document, which was resetting contentY. HomeView is destroyed on every
navigation, so the patch map is re-handed on mount through
QbzHome.refreshForyouArt() — otherwise the Radio rail and the Spotlight hero
went blank for the rest of the session after one navigation, and the skeleton
pulse Timer stayed pinned ON, defeating the fix it belonged to.
- Spotlight's placeholder gated on the local path instead of the remote url, so
it flashed a glyph on a cold cache where the Slint does not.
The retired doctrine is corrected wherever it was load-bearing for a decision.
The pre-baked icon tints are deliberately NOT reopened and no drop-shadows are
added: both are visual changes, not perf fixes.
cc
The owner's clue was the whole diagnosis: the same image looked fine at
220px in the hover preview and bad at 150px in a card. So not the source,
and not the URL variant either -- every Qt producer picks exactly what
Slint picks, `large` (600px) into a 200px cell, 3x headroom.
`artwork_qt` was producing the right derivative and `RoundedImage` was
drawing the untouched 600px original next to it. `Context2D.drawImage`
does not filter, so every card was a nearest-neighbour downscale: measured
on a card cropped out of the running binary, RMSE 0.000 against a
point-sampled reference. The preview looks right because it is a plain
Image with smooth+mipmap and never touches this path.
Three defects were latent behind that dead derivative: `thumbnail(w,h)`
does not preserve aspect (46 square covers on disk written as 32x28),
`to_rgb8` matted transparent logos onto black, and the file was JPEG q75 --
more damage than the resample it was meant to cure.
Qobuz playlist art is 800x380, and both the header and the card were
stretching it to fill a square. It now keeps its ratio, and the uncovered
area takes a gradient derived from the image itself, reusing the ambient
colour pipeline rather than starting a second one.
cc
The owner hit a segfault after clicking around. Cooperative and Threaded
rasterize on the scene-graph RENDER thread, and RoundedImage is instantiated in
25 files — every card, row and queue row — so list recycling creates and
destroys hundreds of these while async artwork callbacks schedule repaints on
them. The render thread can be mid-draw on a pixmap whose Canvas is already
gone. The ambient field makes it worse by repainting 30 times a second for the
whole window lifetime on that same thread.
Immediate keeps paint and destruction on one thread. renderTarget: Image
already meant CPU raster, so nothing looks different.
Not yet confirmed on hardware: the decisive check is running the previous
binary with QSG_RENDER_LOOP=basic, which gives Cooperative no separate thread
to race with. If that survives the repro and the default loop does not, this is
the cause.
cc
library_bridge.rs has sat in the tree unreferenced since the phase-23 work
stopped: no mod declaration, not in build.rs, zero call sites, and duplicating
members that were still live in bridge.rs. It could not have been wired as it
stood — its module was named qbz_library, which is ambiguous with the
qbz-library crate this binary depends on, so it never compiled. The module is
now qbz_library_bridge; the QML type name comes from the QObject, so QML is
unaffected.
Registering it and deleting the duplicates had to happen together. Wiring it
alone would have compiled cleanly and produced two independent Library states —
QML reading one object while Rust published to the other — with the view simply
frozen and nothing in the log.
Moved with it: four properties, five invokables, three signals, the seven
library-only ui() hops in main.rs, and thirty-four QML references. The
Connections blocks in LibraryView, Sidebar, QueuePanel, AlbumView and
PlaylistCollage retarget wholesale because every handler in them is a library
signal. ArtistView's does not: it mixes onLibraryArtworkReady with
onReleaseSectionReady, so it becomes two blocks — retargeting it whole would
have orphaned the releases pager, and QML resolves handlers lazily enough that
the discography would just have stopped loading with nothing in the log.
bridge.rs is down to 389 lines and five domains.
cc
Two bugs with one root: the card bound the playlist's `images*` lists, which
are MEMBER-ALBUM covers, not the playlist's own artwork. So a Qobuz editorial
playlist showed one of its albums' sleeves, and a user playlist — which has no
`images*` at all — showed nothing.
Qobuz playlists carry their graphic in `image_rectangle`, which the model did
not deserialize; the field is added (additive, `serde(default)`, and Playlist
is only ever built through serde) and the card prefers it, contain-fitted so
the artwork is not cropped. Playlists without one fall back to the member-cover
mosaic, reusing the collage the sidebar already draws rather than a second
implementation. Both the All-tab grid and its list mode are covered — the 44px
list thumb binds the same map and would otherwise have gone blank.
cc