geo_push.hpp and replay.hpp cited memllib @ 0a541cc. upstream/main pins
e291192, where the same code had been deliberately redesigned — and because
InterfaceRL was not in the tree (fixed one commit ago), we carried the
superseded version for months. Three changes, all upstream's:
kGeometricPushScale 0.5 -> 1.0 (InterfaceRL.hpp:409)
kNegLRBase 0.5 -> 1.5 (InterfaceRL.hpp:410)
/(1+len) taper deleted (InterfaceRL.tpp:724)
Upstream's own comment on the taper: "a 'no' should clearly move the mapping
away even from a sound already far from the liked region (the taper used to
kill exactly that case)". The direction is already a unit vector, so the
taper only ever shrank the push for exactly the sounds a user is most likely
to be rejecting.
Cold start is folded into the same path. Upstream's useRandom is
`!havePositives || len <= 1e-4`: with nothing liked yet there is no centroid
to push away from, so every dim goes in a random direction. Ours instead
kept the older 0a541cc fallback — train AWAY from the heard action at a
NEGATIVE lr — which was inert whenever the heard action equalled the net's
own output, i.e. in the common case. One path now, and a "no" moves the
mapping before any likes exist (ml_bench E1: 0 -> 2.3e-3). The
GeometricColdStart action is still reported so callers keep their "like a
few sounds first" prompt; only the training changed.
Measured (ml_bench, one dislike at a point):
A4 0.0157 -> 0.0533 at-point displacement (3.4x), so end to end across
this and the RMSProp fix: 5.3e-5 -> 5.3e-2, ~1000x. The gap to the
legacy Diffuse design closes from ~4100x to ~4.2x.
D1 effective_lr 4.7e-4 -> 1.5e-3; 10 presses now reach 0.34, 100 reach
the full intended push.
A5 compounding 0.87 -> 0.96 (a second press at the same spot is no
longer noticeably weaker than the first).
A7 damage_ratio essentially unchanged (0.87-1.57) — the collateral
damage to protected positives scales with the push and is NOT
addressed here; it is the negative-feedback design question.
NOT adopted, deliberately: upstream's per-tick batch retraining over all
live negatives, and its fixed kDislikeLifetimeMs=2500 in place of our
proportional decay. Both need something the core does not have — a per-tick
call site and a millisecond clock inside nisps/ml — so they change
FeedbackControllerCore's interface rather than its constants. Recorded in
ALIGNMENT's deferred-debt entry alongside the existing one-press-one-step
divergence, and filed as its own task.
test_mlp_geo_dislike.cpp: the taper test now pins its ABSENCE (equal
displacement near and far), the cold-start test pins movement where it used
to pin inertness, and a new test covers the random-direction branch.
ALIGNMENT defect 6b resolved. Gates: build-cpp-tests 139 tests / ctest 4/4,
parity-check PASS (WASM rebuilt), lint-cpp clean, firmware slpworkshop
SUCCESS.
Upstream memlp (github.com/MusicallyEmbodiedML/memlp @ ea777502, the commit
upstream/main pins) applies gradients with RMSProp everywhere: Layer.h:239
ApplyAccumulatedGradients, the m_sq_grad_avg running average at Layer.h:601,
StaticMLP.h:268. nisps/ml/training.hpp shipped SGD only and filed the
difference as an optimiser-choice research question. It was not one.
RMSProp divides each step by the running gradient magnitude, so an upstream
lr is a NORMALISED step; under SGD the same number multiplies the raw
gradient. Every learning rate ported from upstream therefore landed in an
optimiser that reads it differently — most visibly feedback.hpp's
`geo_lr_ = 0.001f // upstream InterfaceRL.hpp:312`, an RMSProp LR pasted
into a single SGD step.
rmsprop_step() ports Layer.h:239 exactly: clip at +/-10, sq = min(0.9*sq +
0.1*g^2, 1e6), adj = min(lr/(sqrt(sq)+1e-6), 1.0), w -= adj*g. The
adjusted-LR clamp stays one-sided as upstream's std::min is, so the negative
lr used by train_targets' "train away from this target" path behaves as it
does upstream. The per-weight squared-gradient average is new persistent
state and lives in the storage policies (FixedStorage arrays /
DynamicStorage arena) so nisps/ stays allocation-free and the firmware's
zero-heap contract holds. It is optimiser state, not model state: excluded
from weight_count()/get_weights()/set_weights(), matching upstream, and
cleared by MLPCore::reset_optimizer_state() (upstream ResetOptimizerState).
draw_weights() deliberately does NOT clear it — upstream's DrawWeights
doesn't either.
Measured with tests/cpp/ml_bench.cpp:
D1 one geometric dislike moves the mapping 1.6e-2, up from 5.3e-5 (~295x),
and repeated presses now CONVERGE on the intended 0.5 push (0.12 at 10,
0.56 at 100) instead of creeping linearly forever.
A4 geometric-vs-Diffuse gap narrows from ~4100x to ~14x in one press.
U4 the upstream-LR positive path actually trains now (range_util 0.71 at
100 ticks/gesture, was 0.016 — it was inert under SGD).
Not fixed by this, and now tracked as ALIGNMENT defect 6d: the dose
asymmetry. lurch_max is still ~1.08 against a [0,1] output range.
Golden vector stages 2 and 3 re-captured; stages 0 and 1 are pre-training
and did not move, which is the cross-check that only the update rule
changed. manifold/public/nisps.wasm rebuilt so parity-check compares like
with like — it FAILED at up to 5e-2 against the stale artifact and PASSES at
2.4e-7 against a fresh one. parity-check.sh only builds the WASM when it is
missing, never when it is stale; noted in MAP.md and filed separately.
ALIGNMENT defect 6 resolved (moved to Recently resolved); 6b's optimiser
cross-reference updated; new defect 6d for the positive-training dose.
Gates: build-cpp-tests 138 tests / ctest 4/4, parity-check PASS, lint-cpp
clean, manifold typecheck clean.
Four items from one workflow, committed together because their build and CI
wiring genuinely interleaves — nisps/CMakeLists.txt, run-all-tests.sh and
ci.yml each carry hunks from two of them, and the stage renumbering (1/5 ->
1/6) touches every line. Splitting would produce commits that do not build,
which is worse than a commit that does four things and says so.
S26 part 2 — the curve declaration now matches reality. params[].curve stays
the mode-wide DEFAULT; a voice_spaces entry may now be {name, curve_overrides}
declaring only the slots where THAT voice space deviates. The 6 modes with one
voice space are byte-identical. The values were derived MECHANICALLY by a new
codegen/curve-audit.ts that models the four idioms a p[N]*p[N] regex misses
(alias form, memlcelium's implicit-counter sq() lambda, loop-generated indices,
smooth_params_), inlines helpers, and RAISES rather than guessing when it
cannot reduce an expression. A drift gate cross-checks 1179 (voice space x
param) slots against engine source on every run and was proved to fail loudly
on three drift classes. Application stays in the engine: nisps/engines,
nisps/pipeline and nisps/core are untouched, generated output is pure insertion
(755 insertions, 0 deletions), and the rebuilt nisps.wasm was byte-identical.
S4 / 7.2 — firmware reads the active mode's driver config at mode start, and
mic/line is real. My brief assumed the engine owns this; the code disagreed and
the code was right. sound_analysis_midi's EngineT is NoOpEngine — the mic lives
on a separately-composed AnalysisEngine member — so engine-level wiring would
have compiled, passed every gate, and left the one mic mode on line input.
Hence a mode-level seam defaulting to engine().driver_config(). Separately,
DriverConfig's defaults (line_level 0, output_volume 1.0) had drifted from
memllib's actual 3/0.8 because nothing had ever read them; wiring them as-is
would have made every silent mode louder and its line input maximally
insensitive — a behaviour change disguised as plumbing. Now pinned by a test.
Also: GetSysClockSpeed() panic()s on unsupported sample rates and runs on the
first line of setup(), so sample_rate needed a fallback ahead of clock setup.
CI's firmware env list gains soundanalysismidi — it is the only mic variant and
nothing else compiles that path.
Plan 5e — telemetry is real. A loss_history C-API entry across the full 5-layer
chain lets the browser read the per-iteration loss the core already records.
The audit named one fabrication site; there were two — wasm-iml.ts's
synchronous train() published lossHistory: [loss] as well. A third, ctx.loss,
was not merely dead but actively synthetic (fallbacks of prev * 0.82 and a
literal 0.5, rendered by nothing) and is deleted. The firmware buffer stays
untouched, per the L25 call. EngineApi.lossHistory() reads spine state rather
than the MLP handle, because trainAsync() fits on the worker's mirror net and
the handle would give a subtly-wrong second answer.
Plan 5f — engine throughput is measurable. One source compiled twice (CMake
natively, emcc for WASM) so the targets compare directly and no WASM export is
added. Sequencers are driven into a working state, and every row prints its own
working-state evidence so a number produced by an idle engine is visible rather
than plausible. Reports, never asserts: a wall-clock threshold on shared
hardware is meaningless or flaky, same call as the firmware size job.
ALIGNMENT: the telemetry defect is deleted (built, not deferred); the
performance defect is rewritten to what is actually left — these are HOST
numbers, and nothing measures the RP2350 at 150 MHz, which is the target the
mission's constraint is about. Q4 (memllib ownership) and Q5 (legacy feedback
modes) are closed.
Corrections to my own earlier claims, both found by agents contradicting the
brief: manifold/ONBOARDING.md was NOT "now accurate" — its primitives list
still named five deleted primitives and cited a seededGradient() that does not
exist. And the parity harness misses the sequencer engines because it runs 128
frames while their sequencers evaluate every 400-500 samples, NOT because
all-params-0.5 fails to trigger them (it does trigger: 0.5 maps to ratio 2,
firing three times per bar). The fix is a longer window, not different params.
Gates: run-all-tests.sh ALL GREEN — 4/4 ctest, parity PASS, lint clean, curve
drift 1179 slots ok, 39 e2e (was 33). Firmware: 5 envs built including the mic
variant.
Roughly 20k lines were deleted from this repo in the last week and much of the
corpus still described the pre-deletion world in the present tense. Executes
the §8 table: archive the retired, reclassify the executed, prune the stale.
aimmersive-clone-spec -> _archive/ with a deprecated-by note
feedback-modes-port-spec -> plans/, kind: plan, status: executed
manifold-parity-features -> plans/, kind: plan, status: active
playground-2.0-rewrite -> status: superseded
engine-architecture 434 -> ~120 lines; seam + spine kept, rewritten
present-tense against the shipped engine/
MAIN.md six contradicted claims fixed; registry resynced
vcv-module.md pruned to the current 8->16 contract and made the
single .nisps format spec
vcv/NISPS-FORMAT.md DELETED — documented a v1 format that no longer loads
vcv/README.md, BUILDING.md rewritten to the real contract, menu, OSC table
inputs/backends/dock trio grounding sections marked historical, dead cites fixed
Two rows of the §8 table were themselves wrong, corrected here: the deleted
full-state sync lives in backends-spec.md §6.3, not vcv-module.md (which has no
§6.3), and codegen/README.md was already a MAP pointer with no port-solidjs
trigger left to remove.
Beyond the table — found by sweeping every backticked path in the changed docs
against `git ls-files`, which is how these should have been caught before:
manifold/ONBOARDING.md documented a UI that Phase 1 deleted, as if current:
SplitStage, ReadoutStrip, InputMini, BackendAdvanced, AltitudeNav, and a
shot.spec.ts that does not exist. The whole stage table was keyed on a `focus`
axis that no longer exists — selection is now sandwich > particles >
composite. This matters more than the rest: CLAUDE.md tells every agent to
read ONBOARDING.md first for Manifold work, so it was actively teaching a
fiction. Rewritten against ConsoleApp.tsx.
MAP.md claimed the input layer reduces axes to the engine arity with an
"even/odd blend". input-layer.ts says the opposite in its own header: one
dedicated slot per axis, 1:1, into a 32-input over-provisioned head, and
mean-blending was removed deliberately because it diluted every source.
AGENT-REFERENCE.md still promised TS emission "returns at P5" (landed),
per-mode dims "become schema-real at P5" (landed at P5.3), and pointed at
nisps::FixedBuffer (deleted).
Doc-right/code-suspect, filed rather than fixed: VCV computes derivedMean/Std/
Delta and cachedNovelty behind a live context-menu toggle that nothing reads;
vcv/plugin.json points at the MusicallyEmbodiedML org rather than this repo's
origin; and the module defaults to UDP 7001+id%64 while bridge.ts defaults to
9000, so out of the box they do not meet.
Firmware-build docs are deliberately untouched — the PlatformIO migration
lands next and rewrites all of them.
The operator's call: "there should be one default learning rate and one
default max iterations and they should both be configurable at runtime."
There were SIX copies, not the four the audit described, and they did not
agree:
nisps/ml/mlp.hpp no-arg train() hardcoding 1.f / 1000u / 0.001f —
and firmware's ONLY training path calls exactly
this, so firmware had no runtime knob at all
wasm-iml.ts train() and trainAsync() TS default params (x2)
engine-api.ts learningRate ?? 1.0, with no maxIterations knob
vcv/src/iml.hpp 200 / 0.1 / 0.00001 — silently divergent
external_synth_midi.hpp its own kDefaultLearningRate/kDefaultMaxIterations
schemas/modes/*.json x9, identical, read by nobody at runtime
Now: schemas/ml_defaults.json is the single declaration (validated against a
sibling meta-schema, matching the midi_device.schema.json convention), codegen
emits it to C++ and TS in the same run, and MLPCore carries a TrainConfig whose
default member initialisers read the generated constant.
set_train_config()/nisps_ml_set_train_config() make it runtime-overridable on
every target; the explicit-argument train() overload is untouched. min_error
joins the tuple — it was duplicated identically and belongs with the other two.
The per-mode ml block loses default_learning_rate/default_max_iterations.
default_spread stays (genuinely wired on both targets) and input_channels stays
(codegen-time validated, real information for sound_analysis_midi).
VCV BEHAVIOUR CHANGE, deliberate: MEMLNaut.cpp constructs IML positionally and
relies on those defaults, so the module moves to 1000/1.0/0.001 — 5x the max
iterations, 10x the learning rate, and a 100x looser early-stop threshold. The
old values were never justified anywhere; they arrived with fbc68eb alongside
an unrelated module rewrite and no tuning rationale. Firmware and WASM have
shipped 1.0/1000 all along. It is now runtime-settable if this turns out worse.
The generated header lands in nisps/ml/generated/, not nisps/modes/generated/
where the rest of codegen output lives: training hyperparameters are an ML
fact, and nisps/ml sits below nisps/modes, so emitting them there would make
mlp.hpp include upward. The agent that built this flagged the directory-crossing
rather than hiding it; this is the fix. CI's generated-freshness gate learns the
new directory.
Gates: run-all-tests.sh ALL GREEN — 4/4 ctest, parity PASS (max delta 2.38e-7),
lint clean, manifold typecheck + 17 unit + 33 e2e (which exercise train() and
trainAsync() through a real browser).
feedback_/fbzm1_/fb_smooth_alpha_ were read every process() call but nothing
ever wrote feedback_, so the path was silently inert. Git history shows this
was a *muted* feature, not dead code: `feedback = v0 * feedbackGain` with
feedbackGain 0.1f (MEMLCeliumAudioApp.hpp @ d095688), then "0; //0.1f" —
zeroed with the value kept in the comment — then the write was dropped
entirely when the engine was ported to nisps/engines/ at the 2026-04-29
rewrite (8d0d47b). Operator decision: port it back, live at 0.1f.
The tap is voice 0's own enveloped output, post-envelope and pre voice-1/mix/
shape — NOT the final output. This differs from PAFSynth's feedback shape;
where the sibling and the reference disagreed, the reference won.
Not schema-exposed: the reference hardcoded it, and MEMLCelium has no voice
spaces to vary it by. If feedback depth becomes a research axis it wants to be
a runtime parameter — noted, not built.
Verification, and a gap worth recording: parity-check.sh PASSES (max delta
2.38e-7) but that is NOT evidence for this change — its harness only exercises
PAFSynth and ChannelStrip, and the impulse baseline runs all-params-0.5, under
which voice 0's sequencer never triggers, so v0_env stays 0 and the new write
is 0 either way. That baseline is blind to this path, not a check on it. So it
was cross-checked directly instead: sequencer params forced to trigger early,
4800 samples native vs WASM through the nisps_engine_* C ABI —
native vs WASM max abs diff 1.19e-7 (tolerance 1e-5)
all 4800 samples nonzero, finite, max amplitude ~0.99 (tanh-bounded)
same run with gain forced to 0 diverges by up to 1.49 — not a no-op
No fixture regenerated. nisps.wasm ships here because it is a tracked artifact
under the Phase 0 freshness gate.
Follow-up: the parity harness covers 2 of 8 engines. "Firmware and WASM share
the same engines" is asserted repo-wide but tested narrowly.
Phase 3 (L38, ST4).
L38 is a REAL, VISIBLE behaviour change, and it is a bug fix rather than a wash.
There were two divergent applyCurve implementations: model.ts's
(`e = 0.25 + c*1.75`, linear at c≈0.43) drove the on-screen bars, while
mapping.ts's spec-anchored version (0.5 = exactly linear, symmetric in
log-exponent space) drove `mapOutput` — i.e. the value actually sent to MIDI/OSC/
VCV. So the display disagreed with the signal at every curve setting except the
shared endpoint. mapping.ts's survives; model.ts's is deleted and re-exported.
Displayed values now move visibly, and they now match what is transmitted.
That unification exposed a third copy the finding did not mention: CurvePad.tsx
inlined the same deleted formula to draw the curve-knob preview and its numeric
readout. Left alone it would have matched NEITHER implementation — the knob
would have drawn one response while the bars and the output used another.
CurvePad now calls the same applyCurve, so knob, bars and transmitted value are
one thing. The readout shows the actual exponent.
ST4: GROUP_COLOR was duplicated across four files (Phase 1's ReadoutStrip
deletion removed a fifth). All four now import the single map in model.ts —
including OutputStage.tsx's OUT_GROUP_COLOR, which was outside the sweep's scope
and would have been the last survivor.
NOT done, deliberately: ST4 also invited replacing the fixed map with a
hash-to-palette function, because its keys (formant/pitch/amp/filter/fx/mod) are
JSX-era placeholders that do not match the real schema group strings
(verb/sequencer/general/filterbank/delays/eq/midi/...), so most groups currently
fall through to the accent colour. That is a visual redesign, not a
deduplication, and it should be one deliberate change now that every consumer
reads one map.
Gates: typecheck clean, 17/17 unit tests, 33 e2e specs.
Phase 3 (L17). The midi/osc/vcv transports each carried their own copy of the
status plumbing, throttle gate and lastSent management; the inputs layer already
had this factored out as BaseSource, so BaseBackend mirrors its shape rather
than inventing a second convention.
AUDIT CORRECTION: L17 says "all four" backends duplicate lastSent. Only three
do. The uSEQ CV backend has a genuinely different frame-level dead-zone (an
Int32Array over 14 channels plus gate bits) and no per-output lastSent, so it
takes the status half only — forcing it into the shared per-output path would
have needed a special case, and a base class with a per-subclass escape hatch is
worse than three backends sharing one. Also: BaseSource itself has no throttle
or lastSent (it is status + action plumbing only), so the mirror is partial by
nature; the throttle/lastSent half is the duplication the backends actually had.
Three internal representation changes, none observable: the MIDI backend's
lastSent moves Int16Array -> Float32Array (stored values are integers -1..127,
exact in f32, so every dead-zone comparison is bit-identical); the CV backend
now inherits a setContext that also allocates a per-output buffer it never reads
(cold path, no effect); and onStatusChange's unsubscribe returns void rather
than Set.delete's boolean (the declared type was already `() => void` and no
caller used it).
Phase 2's fix to vcv-backend.ts — tracking and dead-zoning the full N-dim input
vector rather than 2 — is preserved.
Gates: typecheck clean, 17/17 unit tests.
Phase 3 (S1, S5, S6, S25, L11, L37, ST11, ST12). Behaviour-preserving by
construction: the diff on the generated directories is PURELY ADDITIVE (217
insertions, 0 deletions), so no emitted constant changed value. This moves where
truth lives; it does not change what truth says.
- S5: nine mode headers each carried a mechanically identical 12-field
positional ParamSchema aggregate. codegen now emits one
`inline constexpr ParamSchema k<Mode>Schema` per mode, and the struct itself
moved into generated/schema_types.hpp. Each param_schema() is a one-line
return.
- S6 + S25: every mode hand-typed its net shape a second time as MLP template
args, duplicating the schema's own dims. codegen emits a `<Mode>MLP` alias
built from the already-emitted constants (not re-literalled), and all nine
modes use it. NMaxExamples still defaults from kDefaultMaxExamples (Phase 2).
- S1: model.ts hand-imported all nine schemas by name and hand-paired each with
its overlay — so the SET of modes was hand-maintained and could silently drift
from codegen. codegen now emits ALL_MODE_SCHEMAS; SCHEMA_MODE_OVERLAYS is
purely display truth (label/glyph/css/order), which stays hand-curated.
- L37: deleted the hand-written modeEngineId switch, which duplicated
schema.engine_id and silently defaulted unknown modes to 'thru'. Routes on
MFMode.engineId with exactly one documented exception: sound_analysis_midi
declares engine_id 'thru' because its ModeBase audio slot really is
NoOpEngine, while it separately drives the real AnalysisEngine.
- L11: ExternalSynthMIDIMode's shape was literal in two places; now named once
in an ext_synth_defaults namespace with an ExtSynthMIDIMLP alias. Full folding
into the JSON pipeline is NOT done and the reason is recorded in-file: it is a
template family over an externally-supplied Device and variable NOut, with no
single (device, NOut) schema to author.
- ST12: extracted codegen/lib.ts for the helpers both generators duplicated, and
corrected the comment that claimed they had to be separate. Proof the
extraction was behaviour-free: regenerating the MIDI-device outputs produces a
byte-identical tree.
- ST11: deleted codegen/templates/ — dead "reference" files no generator reads,
already drifted from the real emitters.
Gates: run-all-tests.sh ALL GREEN; codegen idempotent (re-running both
generators yields no further diff), which is what CI's dirty-diff gate checks.
Phase 2 (S10, L19, L18, L24).
- S10: EngineApi.inputVector() returned a freshly allocated [lastRawX, lastRawY]
on every spine tick, so VCV bridged mode silently truncated gamepad/MIDI input
to 2-D while the spine already held the full N-dim raw vector. It now returns
spine.lastRawInputs (ArrayLike<number>, documented as a live reused buffer —
copy, don't retain; VcvBackend already copies), and VcvBackend tracks and
dead-zones the full length. Audit correction: "32-input head" is not a
constant — 32 is DEFAULT_MODE_ML.inputSize, the over-provisioned default
before any mode is chosen; real per-mode widths come from the schemas.
- L19: BackendManager.setActive silently dropped a switch requested while
another was in flight. Now stores the latest requested id and re-runs it in
the finally block (latest-caller-wins).
- L18: MIDI CC messages triggered a React state update plus a snapshot
allocation each. notifyBindings now fires only when the binding LIST changes.
- L24: two ConsoleApp global-listener effects had no dependency array and so
re-subscribed on every render, including every pointer frame. Both now read
through a single ref assigned in the render body, matching the existing
onMoveRef pattern. Audit correction: its suggested `[inputs]` dep would not
have worked — useInputLayer returns a fresh object literal each call, so that
dep changes every render too.
Regression tests: input-vector-truncation.test.ts, backend-manager-switch.test.ts
(a fake backend whose start() is held open, to make the in-flight switch real),
midi-notify-churn.test.ts (fail-before confirmed: 51 notifications vs 1).
L24 has no test — this repo has no DOM render harness to count re-subscriptions
against a mounted component; verified by reading and reference-stability tracing.
ALSO: manifold/package.json's test script named its test files explicitly
("bun test src tests/pipeline-golden.test.ts"), so the three new files were not
run by `bun run test` or CI — regression tests that never execute. Now a glob.
Deliberately `tests/*.test.ts` rather than `tests`: bun's discovery matches
*.spec.ts too, which would drag the Playwright e2e specs into the unit run
(verified — it fails). Unit tests go 9 -> 17.
Gates: run-all-tests.sh ALL GREEN.
Phase 2, S35. Two real defects from one root cause, both confirmed by trace
rather than taken from the audit:
1. Divergence. WasmIML built its TS Dataset mirror with a cap of 100 while
every addExample() ALSO pushed into the C++ FIFO ring, capped at 128. Since
train() reads the C++ ring and trainAsync() reads the TS mirror, past 100
examples the two trained on different datasets — silently.
2. Latent OOB read. nisps_ml_train sizes its sample-weight span by the C++
side's example_count() (up to 128), but wasm-iml.ts allocates that heap
buffer from the TS dataset's size (<=100). Once the ring exceeds the mirror,
the span reads past the end of the caller's allocation.
Fix: name the capacity ONCE as nisps::ml::kDefaultMaxExamples = 128, used by
FixedStorage's default template arg, DynamicStorage's default ctor arg, and the
MLP<> alias (which is the only real FixedStorage instantiation path and carried
its own independent 128 literal — the last copy of this dual truth). Expose it
through nisps_ml_describe and have the TS side read it instead of hardcoding.
Dataset's constructor default is removed entirely: a default was what invited
this bug class, and the sole call site now always supplies the describe() value.
ABI NOTE: this extends nisps_ml_describe from a 6-int to a 7-int descriptor.
nisps_ml_describe always writes 7 ints regardless of the caller's buffer, so
every call site had to grow in the same change or it would overflow the WASM
heap by 4 bytes per call. All five sites updated: three in wasm-iml.ts (init
defaults, init per-instance, reshape re-describe — the finding said there were
two), one in wasm-worker.ts, one in tests/cpp/parity_wasm.mjs. The parity
harness's expected-dims check now also pins the new max_examples slot.
Regression test: tests/cpp/test_mlp_storage_defaults.cpp — pins the two storage
policies to one constant, and drives MLPCore<DynamicStorage> exactly as
bindings.cpp does past the old TS cap, asserting it saturates at 128 and not at
100. Fail-before/pass-after confirmed by temporarily setting the constant to
100: 2 failures, named. Reverted: green.
Audit correction: the cited dataset.ts:81 is the FIFO eviction check; the
hardcoded default was at dataset.ts:45.
Gates: run-all-tests.sh ALL GREEN, parity PASS.
Phase 1 group 6 (S11, L16).
- S11: deleted osc-bridge/bridge.mjs. It was not compiled output but a separate
hand-written Node port of bridge.ts (node:dgram + ws vs Deno.listenDatagram).
The completeness critic settled which twin survives:
.github/workflows/osc-bridge.yml deno-compiles ONLY bridge.ts into the
released cross-platform binaries, so bridge.ts plus those binaries are the
distribution and the .mjs had no consumer in any workflow.
- L16: dead protocol legs left over from the retired playground —
sendState/sendWeights in osc-client.ts, the legacy bare-array branch in the
surviving bridge, the unreceivable /nisps/state path, and the unused
module-output listeners.
Note for the docs phase: docs/specs/backends-spec.md still calls bridge.mjs
"already compiled" (doubly false now), and vcv/ still pushes /nisps/state via
OscServer::sendState with no manifold-side counterpart — flagged, not touched.
Gates: run-all-tests.sh ALL GREEN.
Phase 1 group 5 (S15, S16, S18, S19, L22, L23, L20, L21, L1 delete-half).
- S15: the four-way focus/altitude system. setFocus was never called anywhere,
so only the 'composite' branch was reachable. Deleted SplitStage,
ReadoutStrip, InputMini, AltitudeNav, CompactAxis, the UI Focus type/prop,
the focus branches, stripPinned and the vacuous keyboard gates. MiniMeters
kept; engine.feedback.setFocus (a different, live thing) untouched.
- S16 + L1: the decorative stratum that rendered real-looking controls driving
nothing — A/B machinery, the fake seed, seededGradient + weightsRevision,
snapshots, master volume, bpm, and the learningRate/decay/tame/spreadLevel
sliders with their Drawers rows. Each was confirmed self-referential first.
NOTE a real behaviour change falls out of dropping `snapshots`: Undo outside
an active explore-and-place session used to pop a UI-only snapshot that
restored noiseCap/seed. It is now simply inactive unless a genuine
core-backed scratchpad undo exists. Geometric-dislike mode never had a real
undo primitive in the core, so only the fake path is gone.
- S18: BackendAdvanced.tsx and its Drawers block. It self-described as a
duplicate of the inline OutputsBackendConfig editor and BOTH rendered in the
same expanded drawer. OutputsBackendConfig already covers every backend.
- S19: pruned ConsoleCtx to the fields Dock/Drawers/OutputsBackendConfig
actually read; deleted the Axes type + axes/setAxis and the
preset/setPreset/offsetActive chain (permanently 'Sculpt'/false).
KEPT ctx.modes and ctx.setModeId despite having no reader today — the
Phase 5 instrument picker (§7.6, adopted) is built on exactly that plumbing.
- L22: the 5 dead primitives (Panel, StatusLine, ControlAxis, CurvePlot,
Sparkline) and their barrel exports, plus the now-dead .mf-axis-input CSS.
- L23: OutputControl/toOutputControl, ModeIconComponent and the BACKENDS
catalogue; Drawers now reads modeDesc.label/description from OUTPUT_MODES,
the surviving single catalogue.
- L20: the solo-mode selector's two unimplemented options no longer pretend to
be selectable.
- L21: FeedbackController vestiges — seed/undoDepth options, maxUndo, and six
ControllerEngine members nothing called (the finding named three; the other
three are used on the real EngineApi by debug/probe.ts, a different
interface, so removing them from ControllerEngine is safe).
Gates: run-all-tests.sh ALL GREEN (typecheck, 33 Playwright specs).
Phase 1 group 4 (S33, S34, L54).
- S33: removed 12 dead entries across the full 5-layer registration chain
(bindings.cpp KEEPALIVE -> EXPORTED_FUNCTIONS -> the NispsModule declaration
table -> the WasmIML wrapper -> the EngineApi facade): nisps_ml_reset,
example_count, move_weights, feedback_learning_paused, feedback_drag,
jolt_lr_scale, jolt_tick_lr_ramp, pipeline_state_size/save_state/load_state,
feedback_placing and feedback_state. Each was grepped against manifold/src,
manifold/tests, the e2e specs, manifold/tests/wasm-load.ts and
tests/cpp/parity_wasm.mjs — the parity gate builds its own API via cwrap and
is a real consumer, so it counts.
KEPT deliberately: EXPORTED_RUNTIME's heap views + ccall/cwrap (the parity
harness and wasm-load.ts depend on them), and nisps_ml_feedback_static_output,
whose C export IS called directly by parity_wasm.mjs even though no TS
wrapper reaches it. Also dropped parity_wasm.mjs's moveWeights cwrap, which
was declared but never invoked.
- S34: deleted the publishWeights_ channel — EngineSink.setWeights,
Spine.setWeights/weights()/liveWeights and every call site. It ran a C->heap
copy plus a fresh Float32Array allocation at up to 200 Hz into a field
nothing read. getWeights survives for persistence and the debug probe.
- L54: worklet loader — deleted the unused imports object, the 'c' branch,
exMap and the duplicate second loop, and replaced the silent `() => 0` stub
with one that throws, so a missing import fails loudly instead of returning
plausible zeros into the audio path.
manifold/public/nisps.{js,wasm} rebuilt with the trimmed export list (emcc
3.1.69, the CI-pinned version) and committed — the freshness gate added in
Phase 0 requires it, and the webhook ships this artifact to production.
Gates: run-all-tests.sh ALL GREEN, parity 1273 floats within 1e-5.
Phase 0 of the 2026-07 simplification audit (plan §1). CI has been 100% red on
main since 2026-07-13 and every "gates green" claim since rested on local runs.
- S7 / critic gap 2: push memllib `feat/nisps-core-swap` (3 commits incl. the
pin b37fc53) to monkey-w1n5t0n/memllib and repoint .gitmodules at the fork.
Those commits existed on exactly one disk; `git ls-remote` now resolves the
pin, so `submodules: recursive` checkout and fresh clones work again. Drops
the compensating unreachable-pin error paragraph in build-firmware-arch.sh.
- S24 / S31: the manifold-tests job regenerates from schemas/, runs the codegen
golden test, and fails on a dirty diff — the "schema changes ship with both
generated outputs" rule is now enforced rather than assumed.
- S32: a WASM freshness gate runs the parity harness against the *committed*
manifold/public/nisps.{js,wasm} before the CI rebuild overwrites it. That
artifact is what the webhook ships to production, so a stale commit now fails
loudly instead of shipping.
- critic gap 3 / operator decision §7.4: the VPS webhook
(~/.config/webhooks/meml-deploy.sh, not in this repo) waits for the `CI`
workflow to conclude success on the pushed SHA before building. Fail-closed;
MEML_SKIP_CI_GATE=1 for an emergency hand-deploy. Verified the gate query
returns `failure` for fa37047, i.e. it would have blocked that deploy.
- S31: corrected run-all-tests.sh's false "single command CI invokes" header.
Docs moved with the code: ALIGNMENT defect 1 deleted (resolved) and the rest
renumbered; MAP.md's unreachable-pin warning replaced with the fork pin and a
pointer to the §7.5 vendoring decision; ONBOARDING documents the deploy gate
and the tracked-WASM-ships-to-prod hazard; plan §1 marked burned down.
Gates: scripts/run-all-tests.sh ALL GREEN (ctest 4/4, parity 1273 floats within
1e-5, lint, typecheck, 33 Playwright specs).
Schema-backed modes in console/model.ts are now DERIVED from the codegen
schemas in src/modes/generated/ (source of truth): real param names/groups/
count, plus each mode's ml net shape (MFMode.ml) and schema engine_id. A thin
manifold OVERLAY supplies only label/glyph/ModeClass/input/ordering. New
browser-viable modes xiasri + slp_workshop get derived entries; schema-less
visualizer + c15 stay hand-written on DEFAULT_MODE_ML. Schema min/max/default/
label/curve surface as engine-unit metadata (schemaMin/... on MFParam) without
touching the 0..1 routing semantics.
Switching instrument mode reshapes the runtime-shaped WASM net to the mode's
schema ml config (ConsoleApp effect keyed on [engine, modeId]; no confirm
modal). Boot lands paf_synth dims (4->[10,10,14]->33) once WASM is ready. The
P2.3 axis-count reshape offer still reads the engine's live inputSize and does
not spuriously prompt on a mode switch.
Adds schema-modes.spec.ts (P5 gate): drives switches via a new window.__mf
debug seam and asserts describe() dims, getWeights count, output length/bounds,
and UI param count FROM the imported schemas; spot-checks trainAsync after a
switch. Updates reshape/probe-api/geo-dislike specs to assert from the boot
mode schema instead of the retired fixed 32/126 shape.
All gates green: typecheck, unit (9), build, e2e (33).
- generate.ts re-activates the TS emitters against manifold (types.ts,
per-mode <id>_schema.ts, index.ts); 9 schemas emitted
- firmware-fit check: exactly 3 hidden layers (fixed 4-layer topology) and
every dim within (0, 4096] (the browser kMaxDim)
- golden test restores the TS case against the manifold path (byte-
identical to the retained P1-era snapshot)
One-core-engine P4.3/P4.4: the input/output pipeline processing and the curve
catalog now live in the C++/WASM core (nisps/pipeline/*, nisps/core/math.hpp).
The TS ports are deleted and the browser drives the WASM chains.
Engine:
- WasmIML owns a nisps_pipeline_create handle + bridge buffers and exposes
setInputConfig (TS InputConfig → 15-float wire), processInput, resetInput,
setOutputConfig (Infinity slew → 0), setOutputFreezeMask, processOutput
(in place), resetOutput, curveApply, curveApplyBatch (chunked). Handle +
buffers created in init_, freed in dispose, output-sized buffers realloc'd
on reshape.
- Spine routes setInputs through iml.processInput/processOutput (state lives
C++-side); config source-of-truth stays TS-side and is pushed on attach /
setInputConfig / setOutputConfig. Preserves ?debug=1 fixed-dt determinism
(same dt fed to the WASM calls). EngineApi gains setInputConfig/
setOutputConfig/curveApply/curveApplyBatch.
- New types-only modules: pipeline-types.ts (InputConfig/OutputConfig +
defaults + wire int mappers) and curve-catalog.ts (CurveName + name→id).
types.ts declares the pipeline/curve C ABI. engine barrel updated.
- DELETED src/engine/{input-pipeline,output-pipeline,curves}.ts.
Tests (P4.4 gate — recorded-gesture regression):
- pipeline-golden.test.ts now loads the built WASM (indirect-eval shim,
tests/wasm-load.ts) and drives the frozen gesture/output fixtures through the
C++ chains, honouring the per-event dt clock contract. Tolerance 1e-5
(non-momentum drift <5e-7). The 3 momentum configs carry 1e-2: proven-inherent
f32 drift (a byte-faithful f32 port of the exact original algorithm matches
the WASM to <6e-8 while both diverge from the f64 capture by ~7-9e-3), NOT a
core bug.
- curves-golden.json: linear/square/sqrt/centered_power kept as the original
f64 captures (C++ matches within <3e-8); exp/log/sigmoid/cubic RE-BASELINED
from the WASM (deliberate switch to firmware-exact k=1 exp/log, slope-6
sigmoid, true cubic x^3). Provenance recorded in-file.
- _generate.ts rebuilt as the WASM curve re-baseline tool; pipeline-golden-lib
trimmed to pure data builders.
Docs: fixtures/README.md + manifold/ONBOARDING.md updated.
Gates: typecheck, bun test (9), vite build, playwright e2e (27) all green.
TS half of P3.1/P3.2/P3.3: wire the new WASM exports and delete the TS
approximations now that geometric-dislike, jolt, OU, and the seeded RNG live
in the C++/WASM core.
- types.ts/wasm-iml.ts: bind + wrap dislike_geometric, store_positive,
positive/negative_count, set_avoid_style, jolt_press/step/release/active/
lr_scale/tick_lr_ramp, explore_intensity/get/apply. Weight-mutating wrappers
republish weights (version bump); explore_apply reuses feedbackBuf.
- engine-api: extend .feedback (dislikeGeometric/storePositive/counts/
setAvoidStyle) + new .explore facade. thumbsDown now passes the HEARD
(routed) vector, not the raw MLP output (raw == net output => inert cold-start).
- feedback/controller.ts: delete dislikes[] + applyDislikeBias + both C++ GAP
blocks + the SeededRng field; dislike() -> core dislikeGeometric (returns
action, 15 => cold-start prompt); like() feeds the centroid via the core
thumbsUp; getState() exposes positive/negative counts. Delete feedback/rng.ts.
- engine/exploration.ts: execute the P3 SWAP POINT -> drive engine.explore.*;
delete engine/jolt.ts + ou-explore.ts. UI surface unchanged.
- ConsoleApp: one-time cold-start banner (British spelling), routed heard vector
at the dislike call site.
- App.tsx/spine.ts: under ?debug=1 pin a fixed seed + fixed per-tick dt so the
probe/e2e are deterministic (production keeps time-seeded, real-time dt).
- tests: new geo-dislike.spec.ts; probe gains dislikeGeometric/storePositive/
feedbackCounts/setAvoidStyle; the thumbsDown probe test now drives a real
distinct-heard-vector dislike (bare thumbsDown on the net's own output is
correctly inert under the geometric core). 27 e2e + 9 unit green.
Docs: manifold/ONBOARDING.md engine+feedback sections synced.
Wire the runtime-shaped WASM MLP (one-core-engine P2) through the manifold:
- WasmIML.reshape(dims, spread): calls nisps_ml_reshape, re-describes the
instance, reallocates every dim-dependent heap buffer, refreshes weightCount,
clears the TS Dataset mirror (C-side resets), drops the stale training worker,
and pushes the new shape through the sink.
- EngineApi.reshape exposes it and re-ticks the spine so outputs/audio reflect
the new net. Spine already tolerates the arity change (buffers resize, version
bumps); documented.
- Training worker protocol carries the current hidden dims; the worker
ensureNet()s its mirror net to match after a reshape.
- ConsoleApp offers the reshape behind ReshapeModal on an active-layout CHANGE
(never on load; default 32-input over-provisioned head + zero-padding
preserved when declined). British copy, reset-on-reshape.
- Drawers: delete the stale even/odd blending note; honest dedicated-dimensions
line + net-arity chip.
- Probe: __nisps.reshape(nIn) / .describe(); e2e reshape.spec (default 32/126,
reshape to 4, describe reports 4, bounded outputs, weight count 3148→2868,
spine still propagates). All 25 e2e pass (20 existing + 5 new).
- ONBOARDING: refresh the reshape status + stale hardwired-arity gotcha.
Operator-approved ABI change (P2 stop-point). The WASM MLP is now
MLPCore<DynamicStorage>:
- nisps_ml_create(input, output, hidden[3], n, seed) honours its args;
non-positive/null fall back to the historical 32→[10,14,18]→126, so
every pre-P2 caller (manifold, worker, parity harness) stays
bit-identical. Invalid/oversized dims (>4096) → null.
- NEW nisps_ml_reshape(ml, in, out, hidden, n, spread): fresh net at the
new dims, warm-started via nisps/ml/warm_start.hpp (overlapping region
copied; rest keeps spread init); feedback controller re-created (state
resets — reset-on-reshape modal is the front-end contract). Failure
leaves the old net untouched.
- nisps_ml_describe(ml, out): takes the handle; null reports defaults.
- FeedbackController got the same storage split: algorithms in
FeedbackControllerCore<FbStorage>; FixedFeedbackStorage keeps firmware/
tests source-identical via the old alias; DynamicFeedbackStorage (one
arena) sizes to the runtime net. Firmware .text unchanged (122692).
- MLHandle: per-instance scratch vectors; dropped the dead 2MB
batch_out_scratch.
- TS: types.ts decls (+_nisps_ml_reshape), wasm-iml re-describes the
created instance, worker carries a shape-contract note for P2.3.
Verified: ctest 4/4 incl. new warm-start grow/shrink test; reshape ABI
smoke (dims honoured, overlap survives, invalid rejected, outputs
bounded); parity PASS unchanged (2.4e-7); lint clean; manifold 9 unit +
20 e2e green; firmware .text 122692 (+0.30% vs pre-P2 baseline).
Algorithms (forward, backprop/SGD, init, move_weights, diagnostics) now live
once in MLPCore<Storage> (nisps/ml/mlp.hpp). Storage models:
- FixedStorage (storage.hpp): template-sized std::array, zero heap. The
classic MLP<NIn,H1,H2,H3,NOut,...> is an alias preserving kInput/kHidden*/
kOutput/kNumLayers/weight_count() constexpr — firmware + bindings + modes
compile unchanged.
- DynamicStorage (dynamic_storage.hpp): runtime dims, ONE arena allocation
at construction, nothing per-call. #error under NISPS_TARGET_EMBEDDED
(new macro in core/perf.hpp); sole lint-cpp.sh heap-allowlist entry, plus
a lint check that fails if the #error guard disappears.
Verification:
- new ctest test_mlp_storage_parity: fixed↔dynamic BIT-identical across
init/draw/inference/train(FIFO)/move_weights(pin mask)/eval_loss/
layer_stats/set_weights/infer_batch/reset; invalid+moved-from inert
- golden ML vectors (pre-refactor constants) pass → bit-stable refactor
- native↔WASM parity PASS, max delta unchanged (2.4e-7)
- chokepoint B compile: PAFSynth .text 122324→122692 (+0.30%, ±1% budget);
RAM +416B (eval scratch)
- fix: firmware-common.sh used bare 'python' (absent here) → ${PYTHON:-python3}
Part of one-core-engine-refactor P2. nisps_ml_create ABI untouched (P2.2 is
an operator stop-point).
Bring the two playground-only exploration UIs into manifold ahead of the
playground's retirement (one-core-engine refactor §P1):
- Jolt: press-and-hold continuous weight-morph, release to freeze.
- Explore: Ornstein-Uhlenbeck exploration intensity on the output vector.
Interim TS maths ported verbatim from the retired playground modules
(engine/jolt.ts, engine/ou-explore.ts). The ExplorationController
(engine/exploration.ts) is the single P3 swap boundary: it drives Jolt via the
existing EngineApi get/set-weights + process route, and OU via a new inert-by-
default output-morph hook on the spine. In §P3 only that module changes to call
nisps_ml_jolt_press/release + nisps_ml_explore_intensity.
UI lands in the Learning drawer (Jolt hold-button + Explore slider), monochrome-
consistent, British copy. Gates green: typecheck, build, Playwright smoke.
Migrate the specs worth keeping from playground/tests/e2e (retired in P1) into
manifold/tests/e2e, adapted to Manifold's probe surface:
- probe-api.spec.ts: the window.__nisps debug-probe contract (ready, bounded
outputs, example count, randomise, setInputs inference, thumbsUp/Down,
addExample, train loss non-increasing, async train, clearExamples, evalLoss,
inferBatch, getLayerStats, getWeights). Retargeted to MLP<32,10,14,18,126>
(weight_count 3148) and Manifold's direct addExample/routedOutputs surface;
dropped the playground's __init/iml-poke escape hatches and stream-pending
skips.
- spine.spec.ts: the spine invariant — setInputs -> processed -> ml -> routed
yields bounded, consistent routed outputs; the probe stays alive across dock
output-mode switches (driven via the real selector UI, replacing the
playground's localStorage-reload mode cycling).
- helpers.ts: loadProbe(?debug=1 + cleared storage + __ready wait), settleInputs
for EMA convergence, bounded/changed assertions.
Dropped playground UI specs (ui-interactions, persistence, mode-registry list)
that die with the playground chrome. No probe.ts changes needed.
Records canonical gesture trace, curve catalog samples, and input/output
pipeline outputs from the current TS implementations, plus a bun-test drift
guard that re-runs them against the fixtures within 1e-9. Serves the P4
one-core-engine gate: same pointer trace -> same routed output pre/post the
C++/WASM migration.
- Double-click the input mark to enter follow-mouse mode: a window-level
pointermove listener maps the whole viewport onto the surface's [0,1]²
space so the knob tracks the cursor across the entire UI. Esc or a second
double-click exits; a badge + hidden cursor signal the active state. Local
pan/long-press and auto-drift are suppressed while following.
- Keys 1 = thumbs-down/explore (perturb), 2 = thumbs-up (commit), on the
window keydown handler so they fire even in follow-mouse mode. Drawers
learn/inputs lose their 1/2 shortcuts (3-5 keep route/settings/help).
- Sync help keymap (Drawers) + ONBOARDING stages table.
Geometric-dislike ('Push away', Mode 1) is now the default feedback mode
instead of explore-and-place, which works poorly. Add a Clear button to
the Learning drawer that forgets all recorded examples and wipes the
on-map visuals (feedback markers + placed-anchor pins) via the existing
ctx.onClear, now also resetting pins.
The on-screen circular input disc let both the mouse and the gamepad drive
the knob outside the visible circle.
- Gamepad: each stick axis was clamped to [0,1] independently, so a full
diagonal push reached the square corner. Clamp the stick *vector* to the
unit disc (radially symmetric) before mapping to [0,1]; covers both sticks.
- Mouse: the circular variant clamped in normalised [0,1]^2 but the canvas
drew the knob across the full non-square panel against the inscribed circle,
so the disc rendered as an ellipse that spilled past the rim. Map both the
pointer and every drawn position (knob/pins/markers/trail/flash) through the
inscribed-circle geometry, and disc-clamp the auto-drift.
The painter's-algorithm sort was descending, but z2 (depth) increases
toward the camera in both the top-down and tilted-from-above views, so
nearest was painted first (underneath). This showed the last layer on
top in the default top-down view and made the stack occlude inside-out
when tilted back. Sort ascending: farthest first, nearest last.
Restores the April-2026 "uSEQ-Celium" functionality (browser → uSEQ
hardware + CV expander over USB Web Serial) as a first-class Manifold
Outputs backend, and re-vendors the RP2040 firmware into the repo.
- protocol v2 (uSEQ-CV): firmware/useq-celium/shared/protocol.h is the
single source of truth, mirrored by manifold/src/backends/useq-protocol.ts.
26-byte OUTPUT frame, 11×u16 CV (12-bit) + 3-gate bitfield + XOR; fixed
topology; host-agnostic so the MEMLNaut RP2350 can emit identical bytes.
Spec in docs/useq-celium/protocol.md.
- firmware/useq-celium/{main,expander}: PlatformIO RP2040 firmware rewritten
to v2 from the real April pin maps (expander I2C addr 0x10).
- UseqCvBackend (id cvgate): Web Serial connect/identify/disconnect, 100 Hz
stream, per-channel dead-zone, gate thresholding; modeled on midi-backend.
Per-output CvSpec (channel + gateThreshold) on MFParam; config UI in
OutputsBackendConfig + BackendAdvanced; new "CV / uSEQ" top-dock mode.
- bun-test for the protocol frame layout; MAP.md updated.
The OutputEditor popover was placed with a fixed top + an index-based
left/right heuristic and never measured the viewport, so it clipped the
screen edges when the window wasn't full-screen. It now measures its
anchored rect on layout and translates itself back inside the viewport
(8px gutter), re-checking on resize.
Single-file map of the Manifold front-end so an agent can make a tweak/fix
without re-grepping the tree or reading the 40KB design specs: run/build/
deploy/test commands, the three-layer architecture (UI / engine spine / WASM),
the convertible Stage system, the Dock + drawers, engine/inputs/feedback/
backends layout, and the non-obvious gotchas (imperative output reads,
baseURI asset URLs, hardwired WASM arch, curves↔C++ lockstep, no 'C15').
- mf-slider-input fill used calc(var(--mf-pct)*100%) but three consumers
(OutputControlRow held slider, OutputEditor, shared-ui Field) never set
--mf-pct, so the coloured track-fill stayed empty while only the native
thumb moved. Set --mf-pct inline on all three; harden the base class
(appearance:none + transparent bg + --mf-pct fallback + moz-range-progress).
- hide the two concentric noise-cap exploration rings around the input dot.
Fuse the /learn parameter-landscape sandbox into the main console:
- New SandwichStage: ports the 3D landscape renderer to React, driven by the
REAL EngineApi (inferBatch samples each output channel over the 2D input grid;
getOutputs feeds the live probe). Re-samples on engine version bumps; orbit
camera (top-down = flat 2D, tilt to reveal the stack), centred + grows to fill.
- Dock: new bottom toggle (SandwichIcon) → ctx.sandwich.
- ConsoleApp: when on, the sandwich replaces the centre stage in a 3-zone flex —
shrunken Manifold (left) · sandwich (centre, fills) · compact OutputStage
(right) — reusing pos/markers/verdicts. Opening it auto-closes the drawer so
the output zone is visible. Layers = the mode's params (capped 8).
typecheck + build green; verified headless (toggle on/off, sample, tilt).
The default (Particle) view no longer renders the 'MEMLNaut' wordmark
since the particle top bar became a heatmap strip; assert on the always-
present 'Learning' dock drawer label instead so the smoke test reflects
the current Console chrome.
