Phase 2 (L34, L35). Both findings' line citations were accurate this time.
- L34 data race: process() (Rack's audio thread) called
imlShadow.get_example_features()/get_example_labels() directly on the WORKER
thread's private engine — which the file's own THREADING INVARIANT comment
says only the worker may touch — while workerLoop() concurrently
clear/refills those same std::vector<std::vector<float>> members via
load_examples(), train_(), randomise_weights and clear_dataset. Unsynchronised
reader/writer on a non-atomic vector: undefined behaviour.
Fix extends the staged handoff the file ALREADY uses for pendingWeights
rather than adding a second mutex: the worker deep-copies features/labels
into pendingFeatures/pendingLabels at the same instant it copies
pendingWeights, immediately before weightsPending.store(true), and the flag is
now released only after the whole batch is consumed — closing an early-release
window the old code had. process() no longer references imlShadow at all
(verified: the only surviving mention is a comment).
- L35: process() ran full jansson serialize on every weight-swap OSC push and
full json_loads + dataFromJson on incoming OSC state — heap-heavy tree work
on the audio thread. The plan said to move it to the worker. It is DELETED
instead: reading the actual transport shows both directions talk to nobody —
osc-client.ts only ever sends {params|input|feedback}, and bridge.ts has no
state/weights case and explicitly drops other addresses. Relocating
heap-heavy work to serve a confirmed-zero consumer is complexity without a
requirement; removing the cause is the smaller coherent design.
dataToJson/dataFromJson are UNTOUCHED — they remain the live consumers for
Rack patch save/load and the .nisps preset menu, both off the audio thread.
Neither is empirically reproduced: a real race needs a live Rack engine under
TSan, which is not available here. Justified by reading, and verified by
`cd vcv && make -j4` (clean) plus the host suite including
test_vcv_iml_parity.cpp, which pins iml.hpp bit-exactly against the core MLP —
iml.hpp was not modified, and parity holds.
Known remaining, pre-existing and out of scope: process() still takes a brief
lock_guard on feedbackMutex to copy a small staged struct, and several config
fields (slewMs, oscPort, output/input range flags) are written by the UI thread
without atomics.
Phase 1 group 7 (L33). vcv/test/smoke_test.cpp included a header that no longer
exists (a retired nisps-core path), asserted the pre-P6 2x12 module shape, and
ran in no gate; a compiled smoke_test binary was tracked alongside it. Removed
the directory plus Makefile.dist, updated BUILDING.md's two references, and
deleted the unreachable reply-to-sender branch in osc_server.hpp.
Gates: run-all-tests.sh ALL GREEN.
Replace the vendored runtime MLP in vcv/src/iml.hpp (DetRng + 3D-weight-store
MLP + Dataset + IML) with a THIN, Rack-free adapter over the shared core:
nisps::ml::MLPCore<nisps::ml::DynamicStorage> (8->[16,24,16]->16, the P2 dynamic
case), nisps::Rng, and the core MLP's own FIFO dataset. Behaviour changes from
the vendored approximation to core-exact firmware/WASM semantics.
- MEMLNaut.cpp: staged/pending weight buffers and patch JSON now use the core's
flat [weights..][biases..] vector (nisps::IML<float>::Weights); patch version
bumped to 3. Double-buffer / single-writer threading discipline unchanged.
- New ctest tests/cpp/test_vcv_iml_parity.cpp: seeded train/infer/move_weights
session through the adapter is memcmp-equal to a bare MLPCore<DynamicStorage>.
- Docs: vcv-module.md delta #5 marked CLOSED (2026-07-18); MAP.md vcv/ updated.
Closes vcv-module.md delta #5.
- vcv/build-mac.sh: bounded Docker osxcross build (system clang, NOT LLVM-from-
source; cached toolchain image for fast reruns) → ad-hoc-signed mac-x64 +
mac-arm64 .vcvplugin. MacOSX12.3 SDK from joseluisq/macosx-sdks.
- DISTRIBUTION.md: macOS is now a host bounded cross-build, not CI-only.
All four platforms (lin/win/mac-x64/mac-arm64) now published at /next/vcv.
- vcv/Makefile: include arch.mk early, link -lws2_32 when ARCH_WIN (the OSC
server uses Winsock; mingw ignores the MSVC #pragma comment lib).
- vcv/build-win.sh: bounded Docker mingw cross-build (hard CPU/mem caps, no
extra swap, prebuilt mingw — no toolchain compile). Produces win-x64 .vcvplugin.
- DISTRIBUTION.md: reflect the no-dedicated-box reality — Linux + Windows built
locally (supervised + bounded), macOS needs the Apple SDK (CI or a Mac).
- add .github/workflows/vcv-plugin.yml: official rack-plugin-toolchain
(pinned to commit 4fd1318, Rack SDK 2.6.6, image v19) cross-builds the
MEMLNaut plugin for lin-x64/win-x64/mac-x64/mac-arm64 on CI, uploads
.vcvplugin artifacts, and attaches them to GitHub Releases on v* tags.
- add vcv/DISTRIBUTION.md: build + publish flow, how to add CI release
artifacts to meml.lnfinitemonkeys.org/next/vcv.
- add vcv/.gitignore for build products (build/, dist/, *.vcvplugin, binaries).
- BUILDING.md: point at DISTRIBUTION.md; note make dist supersedes Makefile.dist.
Linux x64 .vcvplugin built + verified locally via make dist and published to
/next/vcv (additive subdir). Windows/macOS deferred to CI per prod-host policy.
8 inputs x 16 outputs; per-output LED ring widget (drawLayer+nvgArc); palette
from frontend tokens; OSC bridge verbs for bidirectional browser training;
vendored self-contained iml.hpp (retired nisps-core); compiles against Rack
SDK 2.6.4. See SPEC.md BUILD DELTAS.
Fixes from Opus 4.6 review (C1-C6, I1, I3, I4, I8):
C1: static lastOutputs → per-instance lastOutputsForDelta member
C2: add_example() now on audio thread only (owns iml); worker reads
from mutex-protected staging area (stagedFeatures/stagedLabels)
C3: Worker reads stagedWeightsForWorker (not iml.get_weights()),
eliminating concurrent read/write on iml's MLP
C4: Worker spins on weightsPending before writing pendingWeights,
preventing double-write race
C5: RAND and CLEAR now enqueue Randomize/Clear jobs through worker
instead of directly mutating iml on the audio thread
C6: OSC callbacks stage JSON into oscStagedJson + atomic flag;
audio thread applies in process() (no recv-thread mutation)
Also fixed:
- I4: Separate pendingJob field (enqueueJob no longer overwrites currentJob)
- I8: Removed redundant swapReady atomic
- noiseLevel, cachedNovelty, cachedConfidence now std::atomic<float>
- Worker syncs examples back to iml after training via load_examples()
Replace direct-mutation threading with shadow IML:
- Background thread clones weights from main → shadow IML
- Training and perturbation operate only on shadow instance
- New weights staged in pendingWeights, swapped atomically by audio thread
- Thumbs-down now enqueues Perturb job instead of calling move_weights directly
- Audio thread applies new weights via iml.set_weights() at safe point
- Examples copied to shadow for training, results copied back as weights only
Threading invariant now fully enforced: background thread never writes to
the inference IML. Audio thread applies staged weights between inference calls.
Standalone test exercising IML inference pipeline without VCV runtime:
- Basic inference: all 12 outputs valid in [0,1]
- Input responsiveness: 12/12 outputs change between corners
- Randomize: weights change produces different output mapping
- Spread parameter: measurably different distributions at 0 vs 1
- Expressiveness: 6/6 corner pairs distinct, full range utilized
- Smoothness: 100% of sweep steps are smooth (no binary jumps)
Finding: spread sigmoid saturation effect is architecture-dependent
with small [16,24,16] network (filed as meml-l5a for investigation).
Full spec for MEMLNaut VCV Rack module based on interview:
- 2-8 configurable CV inputs, 12 raw + 5 derived CV outputs
- RL feedback via panel buttons + CV triggers with learn gate guard
- nisps-core C++ engine with background thread training
- User-configurable inference rate (block to audio rate)
- Bidirectional state transfer with companion webapp (file + OSC)
- 10-phase development plan from skeleton to distribution
