Skip to content

fix(guitar-fx): ship real built-in Guitar FX through approved FX Picker and native PCM - #607

Merged
masarray merged 92 commits into
mainfrom
feat/guitar-fx-foundation-assets-ux-20261113
Nov 13, 2026
Merged

masarray merged 92 commits into
mainfrom
feat/guitar-fx-foundation-assets-ux-20261113

Conversation

@masarray

@masarray masarray commented Nov 12, 2026 •

Copy link
Copy Markdown
Owner

Tracking epic: #608

Product objective

Start ArSonMix Guitar FX as a maintainable single native FX insert program, with a clear user journey and a reproducible original hardware PNG system. Do not add a second audio callback, a separate mixer, or a browser PCM engine.

Deliverables in this foundation PR

  • 12 original hardware-look indexed PNGs (amp head, dual cab, 7 pedal bodies, rotary knob, active/bypassed LED). No copied commercial photos/branding.
  • Deterministic Node-only bitmap generator scripts/build_guitar_fx_assets.mjs.
  • Local asset-backed UX study design-references/guitar-fx/editor-prototype.html: preset categories/search, signal chain, focused editor, block bypass, bounded drag reorder (AMP/CAB paired), knob control, A/B, tuner placeholder and design JSON export.
  • Full UX/engineering contract docs/guitar-fx/GUITAR_FX_FOUNDATION_20261113.md: native host ownership, state schema, resource policy, acceptance tests, licensing and staged milestones.
  • Fast PNG/determinism/honesty contract tests.
  • Pure portable engine/builtin_fx/guitar/GuitarProgramModel.h: bounded 16-block stable-ID program validation, normalized parameter mutation and fail-closed reorder. No JUCE/PCM host changes; portable C++ test and scoped Linux CI.

Boundaries

  • Design-only. No DSP, no native picker, no live audio, no actual mic/IR processing, no ASIO FAT claims. Do not advertise the prototype as a playable plugin.
  • Preserve the one approved /preview/, native ChannelId/EdgeId graph and currently approved mixer controls/buttons.
  • No changes to production src/, existing engine/core/, web/, site/preview, device or installer. Only a new unregistered pure program-model header is added under engine/builtin_fx/guitar/.
  • Concurrent Open the FX editor automatically after picker insert #605 FX picker/editor behavior and feat(gui): contextual GAIN/FX RACK and safe Sends-on-Faders console workflow #604 mixer toolbar work need reconciliation before product editor integration.

Focused verification

node scripts/build_guitar_fx_assets.mjs
node --test tests/guitar_fx_assets_contract.test.mjs
c++ -std=c++20 -O2 -Wall -Wextra -Werror -pedantic tests/guitar_fx_program_model_test.cpp -o /tmp/guitar-program-test && /tmp/guitar-program-test

Visual review: open design-references/guitar-fx/editor-prototype.html locally; check 1280/1440/1920 desktop dimensions, browser collapse, button/focus/bypass distinction, amp vs cab vs pedal focus and realistic PNG quality. Native Windows/FAT are intentionally not dispatched for this design-only foundation.

Next product milestone

M1 playable: native GuitarRigProcessor registered once in NativeBuiltInFx, bounded gate/drive/amp/cab, versioned state, generic editor/ACK readback, genuine PCM from Hi-Z channel through existing WIRE to MAIN, and physical ASIO guitar FAT. This requires a separate cohesive vertical integration, not mere visual promotion.

Verified exact-SHA evidence (2026-11-13)

  • PR HEAD: d160885622c4f01989708abbf86cd97189fb4428.
  • Guitar FX path-scoped contract: SUCCESS, includes PNG decode/dimensions/limits, deterministic rebuild, no false PCM and portable bounded C++ model test.
  • Fast UI contracts: SUCCESS. The full Chromium job was not run; this is not visual pixel approval.
  • Automatic PR quick validation: SUCCESS.
  • Visual fidelity review pending: generated hardware skins are procedural originals, not yet approved photorealistic output. Native guitar DSP / sound / physical ASIO FAT not implemented or claimed. Do not mistake successful design CI for a playable guitar rig.

M1 first-sound native DSP candidate — 2026-11-13

At the exact PR HEAD, this work extends beyond the initial design foundation:

  • engine/builtin_fx/guitar/GuitarRigCore.h: real no-heap wet PCM Gate / Drive / nonlinear amp / original 65-tap cabinet FIR (fixed 32-sample PDC), portable C++ proof.
  • engine/builtin_fx/GuitarRigProcessor.h: stereo JUCE AudioProcessor, bounded versioned state serialization, fail-closed malformed-state restore; no second callback.
  • src/NativeBuiltInFx.h: generic native registry descriptor; existing ChannelId, FxInstance, WIRE and hardware authority unchanged.
  • tests/guitar_fx_core_pcm_test.cpp + native_fx_registry_extension_test.cpp: wet PCM tests, cabinet latency/bypass, corrupted state, actual descriptor preparation and state roundtrip.
  • DSP is an algorithmic first-sound rig, NOT a NAM neural model nor a measured commercial speaker IR. No such audio-quality claim.
  • The in-product accepted controls/editor, actual PNG visual integration and physical ASIO guitar FAT are still PENDING. Do not call M1 complete, merge into product release, or show fake meters.

Exact-SHA asset/C++ and fast UI/native CI PASS at c548d6e.
One owner-dispatched native Windows job: https://github.com/masarray/arsonmix/actions/runs/43326382778 — initially RUNNING; inspect current job for final result.

Native live-editor milestone — 2026-11-13

  • PR exact SHA df56da63ddbd35d7b29f1b9f0edc69caa82fa908 includes merge-base reconciliation with main@cf49c56eb3bbd4b44bf7bed3a04c3c1530b28de6 (nonoverlapping paths preserved).
  • GuitarRigProcessor now supports CONTROL-prepared immutable 4-stage programs through bounded 4-slot SPSC, audio-callback ACK, monotonic request IDs, 128-sample dual-DSP crossfade, same-PDC FIR, and stale-rate packet cancellation on device restart.
  • New engine/builtin_fx/GuitarRigEditor.h is a native JUCE editor inside the existing FxInstance editor lease (not a second PCM graph/WebView/mixer). It loads ORIGINAL indexed PNG art embedded by production CMake, has one focused module panel, 4 fast tone presets, real peak meters, and keyboard/mouse sliders. Logical-only/no-callback state never falsely reports a live meter or completed edit.
  • src/NativeBuiltInFx.h registers the one existing host identity and hasEditor; browser FX picker dynamically receives the native registry so no optimistic JS-only plugin stub was added.
  • Only first-sound topology GATE > DRIVE > AMP > CAB is accepted by this editor and DSP. Compression/reverb/modulation, actual NAM models and verified capture IRs remain separate future DSP work, not fake selectable stages.
  • Existing .smix capture is gated on accepted/ACKed state. A closed editor does not strand completions; no heap, locks, file IO, model preparation, or second callback in realtime.
  • Fast/GFX/UI/native presets checks PASS on this exact SHA (see GitHub Actions checks). Windows full compile/CTest triggered once: https://github.com/masarray/arsonmix/actions/runs/43336513346. Check final result; do not interpret pre-qualification success as native/editor FAT.
  • Still pending: Windows exact-SHA qualification, hardware ASIO guitar FAT, measured latency, visual polish/asset approval, safe verified NAM/cabinet IR licenses, M2 tuner, A/B, extra processors and modular reorder.

P0 Windows regression RCA and source repair — 2026-11-13

  • Red workflow: https://github.com/masarray/arsonmix/actions/runs/43336513346, Windows MSVC Build executable and native RT tests. Root cause C2039/C2065 BinaryData::amp_head_class_a_png, cabinet_dual_12_png, pedal_overdrive_png, knob_brushed_metal_png not generated. JUCE omits hyphens when generating C++ resource member identifiers, as independently evidenced by production BinaryData::arsonmix256_png for arsonmix-256.png.
  • Fixed original resource declarations to ampheadclassa_png, cabinetdual12_png, pedalgate_png, pedaloverdrive_png, and knobbrushedmetal_png (with matching Size) in the production native editor.
  • Added fast Node source/asset regression for symbol naming, file presence, deterministic PNG and embedded ARMixAssets contract. New Gate focus illustration uses its own original asset. Sequenced native program transitions so an active crossfade cannot be interrupted by a second edit.
  • Reconciled latest approved main with no overlapping modified paths. Current candidate HEAD is 67b526ad8847a07840cc227b66c04f840ea9fcd1, mergeable draft; preserve single approved /preview/ and one native JUCE PCM callback.
  • Exact-SHA quick checks (asset, preset, UI smoke, native contracts) pass. One exact-SHA Windows native build/CTest dispatched: https://github.com/masarray/arsonmix/actions/runs/43342084263. Inspect final result before marking the blocker green.
  • Physical ASIO guitar listening, round-trip latency, photorealistic visual approval and M2 advanced guitar effects are not claimed; continue as a separate qualification after native code is green.

M1 active-audio reliability milestone — 2026-11-13

New product behavior: Previously GuitarRigProcessor::processBlock() did not consume queued tone edits when the insert was ON; the existing FX host drains native mailboxes only while fully bypassed, and EQ/COMP/Harmonice do the normal active-path drain in their own PCM processing. This stranded Guitar FX preset/knob edits during actual playing. Corrected to consume one bounded edit inside the SAME active realtime block, with no new callback or device graph.

Snapshot integrity: added pending_ atomic intent counter, incremented before SPSC publish, decremented only after accepted callback ACK or restart cancellation. stateSnapshotQuiescent() now denies capture during the queue-pop → completion-publish gap, preventing stale .smix snapshots. Added 32-frame stepped crossfade queue regression, active-PCM ACK, and device-restart cancellation coverage.

Editor lifecycle: drains previous PCM completions before displaying reopened tone, so closing a native editor window before the timer handles ACK no longer displays stale accepted settings. All new changes are covered in targeted source/native tests.

Integration: reconciled with current approved main@de7ad861930bf03ca35f783a59395e42386ca26d, preserving its mixer-fader CSS changes (no path overlap). Current exact PR HEAD c40c661f7fe1d7d31f5d7233003a9a4bfdc9d934 remains a draft.

Evidence: previous Windows exact-SHA MSVC + native CTest SUCCESS at 67b526ad8847a07840cc227b66c04f840ea9fcd1. At current HEAD, all five fast/check lanes succeed: guitar asset validation, fast native, harmony verification, native preset targeted Windows CTest, UI Chromium smoke.

Exactly one new field-preview FAT build requested: Windows FAT / verified ZIP. It includes full Windows MSVC and CTest; do not also dispatch native for the same SHA. Follow GFX_FIELD_ACCEPTANCE.md for actual physical Hi-Z/ASIO validation.

Do not mark M1 shipped until FAT job completes, its ZIP integrity/provenance is verified, and real interface/guitar listening (round-trip latency, routing, tone changes, saved state, device switch) is reported separately. No NAM or cabinet capture claim.

M1.5 professional live-rig update — 2026-11-13

Exact candidate SHA: 2f7c5ea5a8d4bfe9e7e6286bd3728bcc324133a8, reconciled with approved main@a123fd9775a9013f44f6f1df50305893c734f3d2 (no overlapping files; preserves /preview/ WIRE changes). Full 5-lane fast CI passed at this exact SHA.

Product code in this PR:

  • Centralized the four actual factory profiles in engine/builtin_fx/guitar/GuitarFactoryTones.h. Engine and native editor now share the same values and names; any actual parameter or bypass deviation correctly becomes Custom.
  • Real native A/B comparison stored on the processor as two independent bank programs. Versioned GRX1 v2 serializes both and active bank to existing .smix FX blob; old v1 single-tone state is migrated into both banks. Malformed bank/size/ID/parameter rejects atomically.
  • Selected bank, PCM SPSC request, audio callback crossfade, ACK, and accepted bank all use an explicit identity. Quick repeated clicks are deferred instead of replacing an in-flight tone. Editor reopen restores both saved banks; close queues final pending knob and A/B intent if the processor remains ready.
  • Fixed fully bypassed FX workflow: UI no longer requires observed wet PCM before enqueuing, because the existing FxInstance drains native edits in bypass callback. No fake acceptance without actual callback ACK.
  • Cabinet 65-tap synthetic original linear-phase FIR now mirrors coefficients bit-identically and pairs samples (33 rather than 65 filter multiplies/sample/channel); portable C++ tests verify symmetric impulse response and 64-frame vs 1024-frame partition-invariant output at 44.1/48/96 kHz. CPU savings are theoretical operation counts until hardware benchmarking.
  • Expanded real native registry fixtures for v1→v2 migration, both bank persistence, invalid second bank/selected bank, bypass ACK with zero PCM blocks, continued device restart and live two-step crossfade.

Research / future strategy: Professional workflow and DSP architecture. The core reference workflows are NI Guitar Rig rack, Helix Native snapshots, and MIT NeuralAmpModelerCore (library license is not a license for external model captures).

Quality evidence: the previous exact-SHA first-sound Windows FAT SUCCEEDED on c40c661f, publishing verified Windows ZIP and installer artifacts; that preview predates this A/B/FIR update. The current exact-SHA combined Windows MSVC/CTest and field ZIP run was dispatched once: FAT #43355170393; final result is not asserted until the run completes. Hardware guitar/Hi-Z/ASIO FAT, licensed measured cabinet IR and any NAM quality claims remain pending. Keep PR draft until those gates.

M1.6 Windows red-to-green recovery and seamless playing (2026-11-13)

  • Confirmed prior exact-SHA FAT #43355170393 FAILED (1 of 92 CTest): native registry fixture asserted legacy.getSize()==188 before juce::MemoryOutputStream::flush(), which leaves spare MemoryBlock capacity until flush/destruction. Fix checks getDataSize(), flushes, then checks the exact serialized 188-byte v1 payload. JUCE official docs: https://docs.juce.com/master/classjuce_1_1MemoryOutputStream.html.
  • Audio continuity: compatible Gate/Drive/Amp/FIR audio histories are transferred to a prepared program on the realtime callback, so turning Gain/Tone no longer crossfades an old warm cabinet into a zeroed filter history. Reject mismatched sample-rate or module-layout transfers. Crossfade still 128 frames; no DSP coefficient generation in PCM.
  • Bypass recovery: existing RealtimeNativeFxControl::onRealtimeBypass callback now tells Guitar FX when the host is fully bypassed. A partial transition is finalized without waiting forever for skipped wet PCM; queued A/B programs ACK while bypassed. On full bypass, stale 32-sample cabinet / gate history is reset so old notes do not replay on wet resume.
  • GUI performance: embedded amp/cab/gate/drive/knob images are decoded once through a function-local JUCE Image cache; subsequent native editor windows share image handles.
  • Tests: portable warm-history PCM identity, non-compatible rate guard, bypass-mid-fade A/B ACK, no stale FIR impulse after resume, PNG-decode-cache source contract. Preserves v1/v2 .smix migration, stable IDs and one existing PCM callback.
  • M1.6 exact HEAD after nonoverlapping reconciliation with main@2740dcf767a24f82c5c0e8a7aa747fd33f8c4165: 56b9dadf8cf1bb1080cd325ffc09e11d963db24a. Before FAT, confirm exact-SHA fast CI green. Do not confuse previous Windows FAT failure with release success. Physical ASIO audition / validated audio tone / NAM assets remain unproven.

M2 antialiasing + musical Gate milestone — 2026-11-13

New code beyond M1.6 (under this PR):

  • Studio-style stereo-linked gate (independent audio channels, one shared max-level detector/gain), adjustable Threshold −80…−30 dB, Hold 0…120 ms, Release 12…240 ms; preserves existing normalized GRX1 state and hysteresis. Native editor shows real physical dB/ms controls, no fake toggles.
  • New portable GuitarOversampling2x.h: always-on CONTROL-prepared polyphase 33-tap FIR interpolation/decimation around nonlinear Drive/Amp ONLY (no new thread/callback/heap); native-rate Gate/CAB retained. Dry preamp delay stays warm through the enabled stages.
  • Correct processing latency: 16 samples oversampling + 32 samples cabinet = 48 samples PDC, fixed with all stage bypass combinations. JUCE host uses setLatencySamples(GuitarRigCore::processingLatencySamples).
  • Exact portable spectral regression at 48kHz, high-gain 9kHz sine and CAB OFF: folded 21kHz alias from −16.6682 dBc (native-rate legacy) to −57.1979 dBc (2×), ~40.53 dB lower. Hard threshold below −35 dBc. This is one measurable alias target, not a claim of no aliasing elsewhere.
  • Non-gating Linux CI runner CPU probe at 48kHz/128 frames: 1/4/8 inserts approximately 0.75% / 2.93% / 5.77% of equivalent realtime audio wall-clock. NOT physical ASIO or customer's PC CPU.
  • Portable and source contracts verify stereo-linked PCM, audible Hold/Release timing, signal-path history, symmetry, bypass PDC, filter quality gate, 1/4/8 performance measurement and exact native latency declaration. Real PCM callback and no duplicated FX authority preserved.

Evidence: Previous M1.6 exact-SHA FAT #43358470139 SUCCEEDED with 92 native CTests and Windows installer/ZIP artifacts. Those artifacts predate the M2 2× audio processing and are NOT a current-M2 release qualification.

Current M2 candidate: PR HEAD a773a6e5a76f0ace4d8e89f64fdeb413d929773b, base main 383cf1927498d9c3f557110d71825363077b5be6. Fast checks pass or are finishing; do not claim Windows M2 success until an exact SHA FAT succeeds and publishes a verified ZIP. Remain draft pending guitar/Hi-Z/ASIO physical listening, actual xrun/CPU benchmarks, model/IR licenses and visual approval.

Research basis: JUCE Oversampling latency/filter tradeoffs, FabFilter Pro-G stereo linking and Ableton Gate hold/release. New M2 DSP architecture and qualification.

One exact-SHA M2 FAT was dispatched (NO duplicate native job): run #43363114517, candidate a773a6e5a76f0ace4d8e89f64fdeb413d929773b. Fast native preflight already SUCCESS; Windows MSVC + full 92-CTest bundle in progress at recording. Release remains blocked pending exact-SHA job SUCCESS + artifact publication + physical Hi-Z/ASIO qualification.

P0 end-user regression: Guitar FX absent from FX Picker (13 Nov 2026)

Owner field report: Guitar FX cannot be found in the installed app FX Picker. Root cause is a release/integration gap, NOT a DSP success: this complete Guitar FX work was still DRAFT/unmerged PR #607, so any main-based installed build does not contain the native GuitarRigProcessor. Separately the compiled WebView bridge's initial nativeBuiltins catalog lacked armix.builtin.guitar-rig.v1 even inside PR, making first picker opening unreliable until arMixCatalog arrived.

Fixed on this branch: web/approved-native-bridge.js now registers Guitar FX Rig / Guitar alongside EQ, Compressor, Harmonice with EXACT same identity as GuitarRigProcessor::catalogIdentity and C++ NativeBuiltInFx registry's showInPicker=true. The bridge still replaces its pre-event list with the actual native arMixCatalog.builtIns snapshot, so old binaries cannot silently claim Guitar processing. Existing setNativeCatalog→renderFastPicker and LOAD_INSERT→ACK→snapshot are reused; no fake demo audio or separate FX picker.

Regression: tests/approved_native_m2_contract.test.cjs asserts C++/bridge ID parity, visibility, and native insert ACK path; tests/approved_native_offline_logical_smoke.py actually opens original compiled Chromium FX Picker (first-open plus native catalog), searches Guitar and emits a logical ADD_FX_REFERENCE as LAST offline action without wrongly claiming a PCM INSERT. First full UI run #43375981087 verified item visibility and catalog binding but encountered a test sequencing bug: the inserted pending logical command prevented subsequent CREATE_CHANNEL test. Moved the pending click to the last workspace action, preserving both tests. One targeted retry #43376716371.

Hard delivery gate: after actual Chromium picker PASS, reconcile with latest main, run one exact-SHA Windows FAT build producing a user-test ZIP and installer. Only then promote verified code to main if allowed; do not claim the user's currently installed main build magically contains this draft. Physical Hi-Z/ASIO listening remains separately pending and no public stable release is authorized by CI alone.

GFX picker owner deliverable checkpoint — 13 Nov 2026

  • GitHub branch feat/guitar-fx-foundation-assets-ux-20261113 was never part of main-based installed builds. Owner's missing-Guitar FX observation is expected on that older binary, even though previous FAT DSP passed.
  • Corrected first-opener web/approved-native-bridge.js built-ins with armix.builtin.guitar-rig.v1, exact native descriptor parity. Browser picker shows Guitar FX Rig under Guitar, searchable by "guitar" without VST3 scan, with real LOAD_INSERT/ACK/readback.
  • User-visible picker regression in actual Chromium: initial list and authoritative native builtIns event both show Guitar, search narrows to Guitar, click dispatches logical ADD_FX_REFERENCE and waits without optimistic success. Prior full UI failure #43375981087 was test-order contamination by pending logical intent (fixed). Subsequent UI #43376716371 passed GFX/other native FX checks but later failed a pre-existing unrelated V43 WIRE SVG path-count assertion (130 generated vs historical 114). No unrequested WIRE animation edits.
  • Reconciled current main a6b8a3f4c98f7c8552442ffb1ae68f990034f2e0 without path overlap. Current exact source SHA 03ba6ecc2065a6b7e0057b4db0f674be5e78068a has 5/5 fast checks successful, production native registry tests and actual Chromium picker steps passed.
  • One exact-SHA new Windows FAT build #43377503898 is active; it must finish and publish an actual test ZIP/installer before making a claim about the new deployed binary. Do not promote stable/public release before guitar/ASIO listening proof.
  • Previous exact-SHA PR GFX Windows FAT #43363114517 already succeeded with test ZIP/installer, but predates the initial picker fallback fix. It is an optional prior test build, not current verified product release.

@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Nov 12, 2026 •

Copy link
Copy Markdown

Deploying arsonmix with  Cloudflare Pages  Cloudflare Pages

Latest commit: c904034
Status:⚡️  Build in progress...

View logs

Copy link
Copy Markdown
Owner Author

/arsonmix-ci native c548d6e

@masarray masarray changed the title feat(guitar-fx): original PNG hardware asset system and maintainable UX foundation feat(guitar-fx): original PNG assets and first-sound native PCM candidate Nov 12, 2026
@masarray masarray changed the title feat(guitar-fx): native A/B rig, warm DSP continuity, bypass-safe realtime and verified assets feat(guitar-fx): real-time stereo guitar rig with 2x antialias and durable A/B tone Nov 13, 2026

Copy link
Copy Markdown
Owner Author

/arsonmix-ci fat a773a6e

Copy link
Copy Markdown
Owner Author

/arsonmix-ci ui 5b68eea

Copy link
Copy Markdown
Owner Author

/arsonmix-ci ui 2fab34a

@masarray masarray changed the title feat(guitar-fx): real-time stereo guitar rig with 2x antialias and durable A/B tone fix(guitar-fx): ship real built-in Guitar FX through approved FX Picker and native PCM Nov 13, 2026

Copy link
Copy Markdown
Owner Author

/arsonmix-ci fat 03ba6ec

Copy link
Copy Markdown
Owner Author

GFX audit / delivery checkpoint — 13 Nov 2026 (verified from live GitHub)

Verified candidate: 03ba6ecc2065a6b7e0057b4db0f674be5e78068a, still OPEN/DRAFT, not merged into main.

Exact-SHA Windows delivery is now COMPLETE/SUCCESS, not still running:

Important delivery boundary: the installed main-based application still lacks Guitar FX because this branch is unmerged. The two links above are source-qualified field-preview builds, not public/stable release proof. The GitHub workflow's dispatch metadata refers to main but logs demonstrate checkout/packaging from the exact GFX PR SHA.

Remaining hard gates:

  1. Owner's real physical guitar/Hi-Z→ASIO→mixer WIRE→MAIN audible FAT, compare the four tones and CAB bypass, active knobs/bank A/B, .smix reopen, device restart; report actual 44.1/48/96 kHz and 64/128/256 settings when supported, CPU/xrun and measured roundtrip latency. Record results against GFX_FIELD_ACCEPTANCE.md.
  2. Reconcile the latest main 1f77b89f (one commit ahead of the PR's integrated base a6b8a3f4; unrelated WIRE GUI/test paths) when promoting. Avoid redundant native recompiles merely to restate already-verified 03ba6ecc provenance; verify the final integration SHA with proportionate CI.
  3. UX/artist acceptance: evaluate procedural original amp/cab/pedal artwork and compare musical feel; do not claim photorealism/commercial-amp authenticity.
  4. Subsequent M2/M3 work: actual tuner, chorus/modulation/delay/reverb/compressor production DSP and reorder UX; approved IR/cab/model licensing and NAM runtime only after resource/performance/license validation. The 16-block model/prototype is not equivalent to playable effects.
  5. Keep the PR draft and public stable release blocked until field evidence; do not claim hardware qualification from CI or assume current main installer includes this branch.

This checkpoint closes the Windows FAT result verification task; physical qualification and product promotion remain open.

…ar FX P0

No overlapping files with native Guitar FX integration. Preserve both latest approved V66 WIRE UI/tests from main and verified GuitarRigProcessor/FX Picker from #607.

Source FAT #43377503898 and 92/92 CTest were verified at parent 03ba6ec; new merge SHA still requires targeted CI. Physical Hi-Z/ASIO FAT is a separate field gate, not claimed.
…eview branch

Reconcile main #646 native startup foreground repair and tests without overwriting Guitar FX #607. No path overlap. Parent source Guitar FX exact-SHA Windows FAT and current branch quick checks previously green; requalify merge SHA before product integration.
@masarray
masarray marked this pull request as ready for review November 13, 2026 07:43
@masarray
masarray merged commit 977b622 into main Nov 13, 2026
16 of 17 checks passed
@chatgpt-codex-connector

Copy link
Copy Markdown
Codex Review Summary: ❌ Failed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ❌ Failed 2026-11-13T07:43:46.237503Z
We couldn’t complete this review. Comment @codex review to try again.
c904034 Draft marked ready
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant