The exact bit and file-hash pins were taken on one machine and failed
in CI, where the platform maths library rounds differently in the last
bit. Levels are now checked within 1e-3 dB of the pinned values, and
the maximizer and plate reverb tests check that the output does not
depend on the block size.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Profiling a demo-session bounce showed the time going to per-sample work that
can be done more cheaply while producing bit-identical results:
- mix: the fader ramp converted dB to gain (two powf) for every sample of every
strip and of the master; GainRamp converts once per block.
- mix: apply_matrix_span has slice fast paths for one and two taps per output,
summing in the same order as the general loop.
- mix/dsp: strip meters and the bounce peak use an 8-lane max (channel_peak);
max is order-independent, so the result is unchanged.
- dsp: LoudnessMeter true peak keeps its history in a doubled ring instead of
shifting it every sample and runs the four phases side by side, each phase
still summing its taps in the original order.
- dsp: Maximizer finds the window minimum with a monotonic queue instead of
scanning the whole lookahead window every sample.
- dsp: flush() tests both bounds without short-circuiting; audio-io quantize
rounds half away from zero without the libm call (exact on its clamped range).
Demo bounce (24-bit dithered WAV, 55.9 s, 3 bounces per run, interleaved,
15 runs each, Ryzen 16 threads, shared machine): median 3.677 s -> 2.955 s
(-19.6 %), min 3.116 s -> 2.394 s (-23.2 %). Output WAV byte-identical, reported
LUFS/true peak/peak identical. New tests pin meter, maximizer, plate reverb and
demo bounce output against values taken from the previous implementation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
AudioSuite > Pitch Shift > Vari-Fi resampled the clip at a constant
speed. It now ramps the speed between normal and a complete stop, so
the pitch falls or rises with it. Slow Down and Speed Up set the
direction. Fit To keeps the clip's length and plays two thirds of it;
Extend plays all of it and makes the clip 150 % long. Fades On fades
the audio out with the slowdown or in with the speedup.
The ramp is offline::vari_fi, run by audiosuite.process as vari_fi
with change, selection and fades params. The constant-speed varispeed
process stays available to commands.
* 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.
