* Build: line tables only for workspace crates, no debug info for dependencies
Full debug info made most of the dev and test target size and link time. Line tables keep file and line numbers in backtraces; dependency frames lose them.
* Tests: build each crate's integration tests as one binary
Each tests/*.rs file was its own test binary: 98 binaries, each linking the crate and its dependencies. The files are now modules of tests/it/main.rs, one binary per crate, which cuts link time and target size. Filter by module: `cargo test -p <crate> --test it <module>::`.
Binaries with their own main (harness = false) in clap-host and vst3-host stay separate, as do the ui-egui tests that switch the process-wide Light/Dark palette and recovery.rs, whose autosave locks leak into child processes other tests start.
* Xtask: run tests with nextest, doctests apart, and no examples in the test build
cargo xtask ci runs the tests with cargo nextest when it is installed, else cargo test with a notice, and doctests with cargo test --doc. The test build leaves out examples (--lib --bins --tests); clippy --all-targets still compiles them. Test binaries with harness = false run under cargo test: their own main cannot answer nextest's --list. clippy and the wasm check keep going past a failing crate. cargo xtask test runs the test step alone.
Found while testing crates/engine/src/cmd/mix.rs:237 (pan); same pattern at mix.rs:76 (mix.trim)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Found while testing crates/engine/src/lib.rs:368-381 (Engine::execute_merged, with push_undo at 280-291)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Validate the trimmed 64-scalar stored track name before mutating the
document, return that exact name, and preserve no-op document state.
Retain routing-folder bus naming and repair with unique bus names,
stable routes and undo/redo coverage.
Retain the source ChannelFormat for committed audio tracks and reject
unsupported widths, overflowing ranges and incomplete renders before
replacing selected tracks. Preserve current routing and rollback behavior
and document the existing renderer limits with multichannel coverage.
It cut the new clip at MAX_RENDER_SAMPLES / channels without a word, after
allocating up to 2 GiB for it. Refuse such a selection up front, as #326
does for other renders, before anything is generated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Bounce Mix rendered the whole file inside one UI frame: the window stopped
drawing and nothing could cancel the bounce until the file was written,
which takes minutes for long or plugin-heavy sessions.
- engine: io::Bounce renders a slice at a time (step) and reports its
progress; dropping an unfinished one cancels it, removing the partial
file and keeping any older one. cmd::bounce_job builds it from
file.bounce_mix's params; the command and stems still run it to
completion.
- ui: the Bounce dialog starts the job and logic() renders it for 30 ms
per call (32 blocks in the browser, where Instant panics), requesting
another call each time, so it carries on while the window is hidden
(eframe then runs logic alone). A modal shows the file, a progress bar
and Cancel (or Esc); the status line reports the result, and the Task
Manager lists the running bounce.
- It stays on the UI thread because hosted plugins must be created there.
A bounce rendered the whole mix (or bus) into memory and then encoded a
second copy, and the render silently stopped at 2^29 samples: about 93
minutes of stereo at 48 kHz, 31 of 5.1, 15 of 7.1.4. Longer bounces came
out cut short while file.bounce_mix still reported the requested length.
- audio-io: FileEncoder writes WAV/BWF (RF64 above 4 GiB), AIFF and FLAC
block by block into any Write + Seek, byte for byte what encode writes
for the whole buffer. check_room fails up front when an AIFF (4 GiB) or
a FLAC (2^36 frames) can't hold the bounce. AIFF and FLAC encode run on
the same writers; their output is unchanged.
- mix: RangeRender renders the main mix or one bus over a range of any
length, a block at a time; render_range and render_bus (in memory,
still capped) are built on it.
- engine: io::Bounce streams the main mix or a bus, or its stereo
fold-down, into "<name>.part" next to the target, syncs it and renames
it into place, so a failed bounce leaves no partial file and keeps the
old one. In the browser it encodes into memory (failing instead of
aborting when memory runs out) and delivers the file as before.
Normalize renders twice (peak, then the scaled write); levels are
measured as blocks are written. Stems stream the same way and share the
4-hour limit.
A 100-minute bounce of the demo session now writes all 288,000,000
frames with 83 MB peak memory.
Opening, creating, closing or reverting a session in the middle of a
take dropped the recorder: the take was written to disk but never added
to any session. Quitting mid-take did the same, and on_exit didn't
finish the take either.
New, Open, Close, Revert and Quit now wait for the take: the engine
refuses them with "stop recording first", and their menu items are
greyed out while recording, as is Recover in the recovery window. Any
exit that still happens mid-take (closing the window, the system
quitting the app, a remote quit) finishes the take first and keeps the
session, take included, in the autosave folder for the next launch.
The edit modes only partly worked. Shuffle changed Cut, Clear and Paste
but not moves or trims. Spot froze drags and had no dialog. Grid snapped
relative to whichever clip was under the pointer, and Relative Grid
acted like Absolute. Behaviour now follows the Edit Modes chapter of the
Pro Tools Reference Guide.
Shuffle
- Grabber drags lift the clips out (later clips close the gap) and land
them at the nearest clip boundary, pushing later clips right; an empty
track takes them at the session start. The drag outline shows where
each track's clips will land.
- Trims move the later clips (a trimmed start stays put and the material
slides under it). Duplicate, Duplicate and Extend, Repeat and Clip List
drops make room, and Paste over a selection replaces it.
- Moves lift each clip by id and close the hole it leaves, so crossfade
partners and unselected clips keep their audio, and a crossfaded pair
moved together stays a crossfade. Ripples never push clips past the
session start, never split a crossfade (refused with a reason), and
drops never land inside a clip or crossfade.
- A time- or edit-locked clip and everything after it stay put when
Shuffle closes or opens gaps. Edits that would push into one are
refused ("not enough room before a locked clip"), and trims stop short
of it. Insert Time and Cut Time are timeline operations and still move
everything.
- With Automation Follows Edit, automation moves with every ripple, up to
a locked clip, and travels with moved clips; with it off, Shuffle edits
leave automation alone. Shuffle Clear used to move the clips but leave
their automation behind (#232).
- Shuffle Lock: Cmd-click (Ctrl-click on Windows and Linux) the Shuffle
button while in another mode. While it's on, the button shows a padlock
and every way into Shuffle (keys, button, commands) is refused.
- A drag works out every lift and landing point once per target track
(`lift_plan`, `shuffle_slots`, and a `DragPlan` kept for the drag). In a
release build, dragging 100 clips on a track of 20,000 took 29 ms a
frame and a 135 ms drop; now it's under a microsecond a frame and a 5 ms
drop.
Slip, Spot and Grid
- Slip: trims stop at the neighbouring clips instead of overlapping them.
- Spot: a Grabber click or drag, a Trim click or drag and a Clip List drop
open a Spot dialog: Start, Sync Point and End in any time scale, the
Duration, and the clip's Original Time Stamp. Trim clicks pick the
nearer edge.
- Grid: a drag snaps the dragged clip's start (its sync point, when it
has one inside the clip) to the grid. Relative Grid moves by whole grid
steps and keeps the clip's offset from the grid; a cut or copy pasted
in Relative Grid lands as far after a grid line as the copied range
started after one. Holding Cmd (Ctrl on Windows and Linux) while
dragging or dropping from the Clip List swaps Grid and Slip. Grid lines
no longer fall a step short at tempos where a line isn't a whole number
of samples.
- Choosing Grid from another mode picks whichever of Absolute and
Relative Grid was used last; F4 in Grid toggles them, like clicking the
Grid button again. Options > Edit/Tool Mode Keyboard Lock, which did
nothing, now stops F4 and Alt+4 from switching them (toolbar clicks
still do).
- Snap to Grid in Shuffle, Slip and Spot: Shift+F4, Shift-click on the
Grid button or on another mode's button, or hold F4 with F1-F3. The
Edit cursor and selections snap to the grid while clip moves and trims
follow the main mode; the Grid button is outlined while it's on.
- Option+1-4 (Alt+1-4 on Windows and Linux) work like F1-F4, and the `
key cycles Shuffle, Slip, Spot and Grid, skipping a locked Shuffle.
- Insert Silence pushes later material only in Shuffle; in the other
modes it clears the selection, as in Pro Tools.
Also
- Dragging clips from several tracks keeps each track's distance from the
grabbed clip's track. Drops past the last track, or onto a track that
can't play the clips (audio onto MIDI, an aux or the master), are
refused; clips moved onto an aux used to vanish into its unused
playlist.
- Undo and Redo keep the Edit mode, Shuffle Lock, Snap to Grid and the
Grid kind. Start trims keep a clip's sync point on its audio, or drop
it. Imported BWF time stamps are converted from the file's rate (read
from its header) to the session's, and the Original Time Stamp no
longer adds the session start twice. Typed positions that overflow are
ignored instead of panicking.
Commands: `edit.move_clips` follows the mode, and its `track` moves the
first clip's track there and keeps the others' distances. `edit.trim_clip`,
`edit.snap_to_grid` (Shift+F4) and `edit.shuffle_lock` are new, and
`edit.mode` takes `snap_to_grid`. Mode buttons expose their names to
accessibility tools.
Tests: crates/{engine,ui-egui}/tests/edit_modes.rs and the op tests in
edit.rs; the random-edits property test also toggles clip locks.
Commands clamp every position they take to ±2^40 samples so the
arithmetic after them can't overflow, but session and clip group files
skipped that. A clip, MIDI note, marker, automation point or selection
near i64::MAX made Duplicate, Insert Silence, Insert Time, Heal, Trim to
File Boundaries, Quantize to Grid, Separate on Grid, Thin Automation,
Copy, Move Selection, Extract Chords and Track Commit overflow: a panic
in debug builds, wrapped positions in release.
MAX_POSITION moves to soundcraft-time so the model can use it. Loading
a session clamps clip starts, lengths, offsets, sync points, gain
envelopes and MIDI notes, automation points, markers, key changes, the
selection, playhead, pre/post-roll, scroll and timecode start. A clip's
repairs are now Clip::sanitize, which clip group import uses too, and
Session::sanitize visits each collection once. The Edit window adds a
clip's offset to its file's time stamp without overflowing.
One offline render holds MAX_RENDER_SAMPLES samples, about 93 minutes
of stereo at 48 kHz, and the offline time stretch, pitch shift and
resample stop at MAX_OFFLINE_SAMPLES frames. Longer audio was cut short
without a word: Render Clip Gain and Clip Effects Render left the end
of the clip silent, every AudioSuite process shortened the clip, the
track Commit made fell silent at the limit, and Export Clips wrote
shortened files. Above 74 kHz an hour no longer fits, so TCE, Elastic
Properties and Remove Pitch Shift padded the cut end with silence.
Each now checks the length before it renders and refuses with the
limit in seconds, as Consolidate does in #212. AudioSuite counts what
Time Stretch, Pitch Shift and Varispeed make, so a fourfold stretch is
refused at a quarter of the limit. Export Clips checks every clip
before it writes a file.
When a third-party plugin refused its stored state (a newer plugin
version that can't read it, a damaged blob), it ran with its defaults,
and the next save or two-minute autosave read its state back into the
session: the saved settings were gone, even after updating the plugin.
create_instance now says whether the plugin refused the state
(NewInstance::refused_state). Playback leaves such an instance out of
the states it reads for saving, so the session keeps the stored state,
until the plugin's own editor is opened: from then on its state is the
user's work. Parameter values are saved apart from the state, so edits
made with SoundCraft's own controls survive either way.
Saving, bouncing and the two-minute autosave read every third-party
plugin's state through the audio callback: a command made the callback
call each plugin's save_state between two blocks and reply, while the
UI waited for up to 2 s. So:
- CLAP's clap.state and VST3's getState ran on the audio thread,
although both formats want them on the main thread.
- The callback allocated and waited on the plugins; a plugin with a
large state (a sampler) could make the audio drop out on every save
and autosave.
- Without a running callback (a stalled or lost device), every save
waited 2 s and then kept stale plugin states.
Each instance now hands Side::prepare a state reader before it leaves
for the audio thread, the way it hands over its editor handle. CLAP's
GuiLink and VST3's EditorLink (which now holds the component as well)
read the state under their lock; Audio Units get a StateLink that does
the same. Each instance cuts its link off before it is destroyed, so a
reader never calls into a dead plugin. Player::capture_states reads
them on the calling (UI) thread, and the CaptureStates command is gone.
Each host's tests read the state while another thread processes the
instance, and get nothing once it is dropped. Playback reads states
with no audio thread at all.
event.midi_track_offsets stored midi.offset.<track> but the note
scheduler never read it, so the render was unchanged. schedule_notes
now looks that many samples back into the clips, which plays the track
exactly as if its MIDI had been moved by the offset. The scheduler is
shared by offline bounce and realtime playback.
Setup > Current Feet+Frames Position stored its origin in
feet_frames.start but nothing read it, so every Feet+Frames reading
still counted from sample 0. format_position/parse_position now take
the origin alongside the timecode start, and Session::format_position
and Session::parse_position pass both, so the counters, rulers,
session.format_time and typed positions share one reference.
Consolidate clamped its range to the track's last clip end plus one
second, so on an empty track a 0-5 s selection made a 1 s clip and a
2-7 s selection made nothing. It now renders, clears and places exactly
the selection, and refuses a selection too long to render instead of
placing a shorter clip. Prepare DPE keeps the old per-track bound.
Fixes#164
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Pinch and Cmd+scroll zoom the timeline around the pointer. egui 0.36 turns
both into zoom_delta, so the old Cmd+scroll branch never ran; Shift+wheel
pans sideways again for the same reason.
- Two-finger panning keeps the sub-sample remainder, so slow motion still
moves at the closest zoom; the scrollbar drag pans the same way (its range
now takes in the view, so the thumb stays in the bar), and the rulers pan
too. Panning stays open-ended to the right, as before.
- A swipe keeps to the axis it starts along, so panning doesn't drift between
tracks and scrolling doesn't drift in time. The larger share of its first
6 points decides until egui reports the scroll over: the fingers lift, or
150 ms pass without wheel events.
- The MIDI editor scrolls by the point instead of rounding each frame's delta
to whole rows (slow swipes did nothing), and its range fits the panel
(pitch 0 could not be reached; a taller panel showed blank rows).
- Zoom and scroll write through `Engine::zoom_mut` (`view` commands): no
playback resync (a session copy sent to the audio thread every frame), no
copy of the whole session per frame, no journal entry per frame, and no
end to an undo gesture (scrolling during an automation pass made each
write its own undo step). New `view.zoom_by {factor, x?}`;
`view.scroll` saturates; Fill Window and Zoom to Selection stay within the
zoom range (zooming out after them on a long session zoomed in).
- Control channel: `ui.scroll` (trackpad swipe or wheel notches) and
`ui.zoom` (pinch); `session.inspect` reports zoom.
Video (new standalone crate soundcraft-video):
- Pure-Rust H.264 (Baseline/Main/High), ProRes (all flavours) and Motion JPEG decoding with
ISO-BMFF demux, adapted from FilmCraft's crates; fuzzed with mutated/truncated movies
- TrackKind::Video with ClipContent::Video referencing Session.videos; file.import_video adds
the picture track and the movie's audio; relinking on open; video.offset/online/list
- Video and Video Universe windows, thumbnails on the timeline, background decoding
Third-party plugins:
- Plugin state (CLAP state ext, VST3 component + controller state) saved in sessions
- Hosted instances created, state-loaded and dropped on the UI thread, never the audio thread
- Plugin editors: floating CLAP GUIs and VST3 views on macOS (verified with Melodyne)
- New isolated FFI crate soundcraft-au-host: Audio Units on macOS (effects and instruments,
parameters, latency, MIDI, ClassInfo state), wired into the mixer and plugin browser
UI: Window > Renderer (speaker layout with live levels and object positions), Window > UI
Customization (interface scale, reduce motion), Window > Configurations > Auto-Update.
Fix: saturating sample-index math in clip rendering (hostile stretch values overflowed).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Surround:
- main and bus formats up to 9.1.6 and Ambisonics; pairwise constant-power
surround panner (divergence, centre, LFE, height) per track and send;
ITU fold-down; multichannel bounce with correct WAV channel masks;
multichannel device output; surround panner and multichannel meters in the
Mix window; setup.main_format, mix.surround_pan, mix.send_surround_pan
VST3:
- soundcraft-vst3-host: isolated unsafe crate (vst3 bindings, MIT), scans
standard folders, hosts components/processors/controllers, params,
latency, notes; VST3 menu; engine.vst3_plugins; e2e tests with a fixture
Hardening:
- positions from parameters clamped; offline renders capped and bounded to
track material; loop/transport limits; hostile-parameter property test
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- soundcraft-clap-host: scans standard CLAP folders, loads plugins via
clap-sys/libloading, implements the Plugin trait (audio, params, latency,
notes); the only crate allowed unsafe, every block documented; stubs on wasm
- mixer, engine (mix.insert, engine.clap_plugins) and Mix window CLAP menu
- end-to-end tests against a Rust-built fixture plugin
- file.bounce_stems: one soloed file per track
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
aiongg