Skip to content

ARD: All-channel native FX Rack + safe drag reorder + yellow-green toolbar + recessed fader #95

Closed as duplicate

Description

@masarray

FACTORY FX / SIX FIRST-PARTY VST3 (2026-10-30): Factory FX ADR · Factory FX PRD · manifest schema · merged architecture PR #187 · epic #185. Six original bundled VST3 targets: ArSonEQ, ArSonComp, ArSonVerb, ArSonHarmony, ArSonColor, ArSonKuPik Enhancer. NO binary FactoryFX modules are currently shipped by this docs change; first qualify ASKPVST public v0.5.26 + ArSonEQ feature branch artifacts and obtain product/AGPL/JUCE/EULA approval; ArSonVerb license and Comp/Harmony/Color DSP are later gates. ASKPVST #60, EQ #6, Verb #4. G5A #178 and MS1 #184 are already MERGED. Factory immutable VST3 modules must use existing native FX graph while USER vault remains separate; protect L2 #176, PDC #95, actual hardware #15. Do NOT mark a plugin loadable from a roadmap card.

R6 CURRENT EXECUTION / L2 PRESERVATION — 2026-10-30: Code-audited architecture and priorities · Operator GUI contract · Merged PR #181 · UX implementation #182 · L2 immutable GUI #176. Source baseline main@8251bd6 (re-fetch HEAD before editing): G3 and L2 merged; channelFormat MS1 #173, typed bus MS2 #174, independent hardware outputs MS3 #175 and uniform FX/PDC #95 are NOT completed. #15 real zero-FX audio hardware blocker remains unresolved. Protect ONE seven-row grid and fader alignment for CH1–10+MASTER, FX/ROUTE/SENDS left rail, top MIXER/FX RACK/PATCH, CH1 VST+Orbit, native epoch/revision ACK/MUTE. No old PR bulk-merge, fake Master FX/monitor pins or GUI-only LIVE.

FORMAT-AWARE VST3 / PDC DEPENDENCY (2026-10-30): ADR R5 applies to this all-channel FX issue. Channel format cannot be faked by duplicating stereo inputs or by cosmetic FX slots. Plan MS1 #173 before format-dependent FX, MS2 #174 for FX AUX/GROUP buses and VST mono/stereo negotiation, MS3 #175 for independent output patch. Existing CH1 multi-VST/Orbit and G3 AUX editor must survive; PDC required for latency-aware parallel sends; negative ACK preserves old graph. Older issue sections below are historical. #15 real hardware crackle still open.

ArSonMix — FX Rack all-channel + realtime-safe reorder + tactile mixer refinement (engineering ARD, 2026-10-27)

Status: architecture/audit and acceptance proposal, NOT implemented, NOT hardware-certified. Baseline: main@0dc201c556623a55348abbb93d8dbd0d1cf08cb4 (verify fresh SHA before implementation). Screenshots supplied by owner in ChatGPT on 2026-10-27. Preserve latest approved 10-channel + MASTER GUI, original ARI PNGs, current CSS ownership, native audio/Orbit, and all existing P0 protections. Related: #15 hardware P0, #56 pointer performance, #37 CSS ownership, #85 portable studio contract.

Source-confirmed root causes

  1. web/app.js:167-176, 489-565: 10 GUI channel racks are built from mock channels[].inserts and localStorage; normalChain()/movePlugin() reorder a demo chain and slot-index keyed bypass/parameter metadata. web/bridge.js:159-187,514-560 replaces native FX RACK with CH1-only live rack and patches only the first strip; MASTER and CH2–10 are left preview. web/bridge.js:659-775 intercepts all insert clicks, rejects CH!=1 and offers no native drag reorder.
  2. src/Main.cpp:543-555 only applies channel fader/PAN/mute when channel == 1. Main.cpp:965-1067 sends FX operations by slot without channelId; LiveAudioEngine.h contains a single live graph (graphOwned_, chainSize(), chainState()) and a single mic->FX->master processing path. Multi-channel mixer DSP is NOT implemented; merely enabling GUI buttons is deceptive.
  3. LiveAudioEngine.h:500-549,701-741 uses detach()/attach() during add/remove/move. Main.cpp explicitly blocks structural edits in independent-clock mode. Repeated native moves while dragging would cause dropout and unsafe graph churn. restoreSnapshot() has an existing prepared graph handoff path; adapt its lifecycle patterns, do not bulk import old draft branches.
  4. web/styles-src/20-button-180.css:147-149 sets selected toolbar navigation to near-white #f0f1f3, contrary to owner's new yellow-leaning green request.
  5. web/styles-src/10-console-legacy.css paints a narrow dark .fader:before vertical slot; 42-mixer-premium.css:146-158 calibrates top/bottom to 38px/39px. The deeper wide recessed bed under the cap seen in owner's hardware reference is missing/insufficient. Existing 36×77 ARI silver & gold fader artwork must remain intact.

Hard requirements

A. One consistent FX Rack contract across all strips INCLUDING master

  • The MIX view must show six visible inserts and an Add FX affordance for CH1–10 + MASTER; native FX RACK should use one reusable channel-aware component, not separate cosmetic and CH1-only versions.
  • Channel select always selects actual channelId, including MASTER bus. Picker, Editor, Bypass, Add, Remove, Move, Save/Recall all route to selected real bus. Disable/label ONLY capabilities not yet connected with truthful diagnostics; never claim an insert processed audio if no bus/input assigned.
  • Native response protocol: requestId, channelId, insertId (stable per instance), expectedGraphRevision, targetIndex, status/error. Engine authoritative, UI optimistic only during pointer preview; success ACK commits, failure rolls back. Present per-channel chain arrays, not global state.chain.
  • insertId carries VST identity + state + bypass + Orbit association through moves; no slot-index metadata swapping. Persistence saves channel graph, stable plugin IDs, ordered FX state and source mapping; integrate with [P1][BACKLOG] Portable VST Vault — user-owned state and fast session reopen #85 Workspace/PerformanceStore (no absolute-path portability regressions).
  • Mono/stereo/FX Return need explicit routing semantics; input/channel map from actual devices or user-configured source. Do not fabricate native meter or audible support for unassigned tracks.

B. Realtime-safe, minimal-overhead FX drag-and-drop

  • Pointer drag handle on FX row + drop indicator, autoscroll confined to chain. Drag DOM preview using only transforms/order markers; no native event during pointermove. One moveFx command on pointerup, cancel with ESC; retain ↑/↓ keyboard-accessible move.
  • Engine implements single transactional move (including arbitrary from/to) at block boundary via immutable processing order plan or equivalent. Preserve processor objects/editor/vendor binary state; DO NOT instantiate VST3, call prepareToPlay, releaseResources, detach()/attach(), allocate, lock, log or perform I/O in the realtime callback. Never process the same AudioPluginInstance concurrently in 2 graphs.
  • Non-RT control thread prepares new order/PDC data then atomically publishes; callback ACK + epoch/revision fence; stale/duplicate command rejected. Reclaim retired plan only on non-RT thread after quiescence. Short click-free transition when audible graph changes, preserving sensible latency; no promised seamless DSP response without measurements.
  • ADD/REMOVE prepare/cancel off RT and are published safely; owner life-cycle, failure rollback and independent-WASAPI mode safety must be handled explicitly (no unlock by simply deleting the existing guard). Active DearVR Orbit follows its stable plugin instance when other FX are added, removed or moved and never resets due to slot shift. Preserve valid editor references and all plugin state.
  • Avoid full mixer DOM rerender or RAF-per-pointer audio commits. Use event delegation + cheap visual marker and one commit. Realtime callback cost bounded by active instances/channels only.

C. Actual multi-channel audio graph

  • Separate SourceRegistry/InputRouter (physical inputs / supported loopback / mapped stereo bus), ChannelRuntime[10] each gain, pan, mute, solo, per-channel FX chain & meters, MasterBus with own FX chain, output selection. Empty/unassigned input = silent and labeled. Do not assume 10 independent hardware inputs exist.
  • Delineate mono vs stereo processing; explicit mapping for CH3/4 music L/R and FX Return; no surprise doubled gain/feedback. Sample rate/buffer contract from active device. PDC and latency exposure per bus; active VST capacity governed by measured CPU/driver deadline, not arbitrary catalog limits.
  • Must respect P0 — Crackling with 0 VST3: diagnose WASAPI clock/input/output buffering before engine expansion #15 real Windows output-crackle P0 and existing ASIO/direct/independent-WASAPI recovery fences. Stage with a deterministic in-memory fake audio device first; then measured real hardware gates. Never claim production readiness from CI alone.

D. UI color + real recessed fader

  • ONLY selected/on toolbar action pads switch from white to a restrained yellow-green with dark legible ink; neutral off buttons remain unchanged. Proposed starting swatch #B9CF77 (tune visually against owner's screenshot). Reuse existing seven-layer BTN-180 pad and 1px momentary press; keep semantic rail, MUTE/SOLO, device and MASTER palettes intact. Edit source 20-button-180.css, not compiled CSS.
  • One shared fader module CH1–10 + MASTER: wider dark recessed bed with crisp bevel and separate narrow black vertical slot, cap PNG visually passing above it; depth via two/three inexpensive static gradients/inset edges, NO blur, CSS filter, glow, animated box shadow, backdrop-filter. Do not add event listeners to decorative layers; pointer-events:none.
  • Preserve current 36×77 ARI PNG assets and --pos, detent center y=38px, +10..−60 dB tick and meter calibration, true relative drag/no-jump, native fader parameter mapping. Keep width responsive and no overlap at 1680×920, 1440×820 and Windows 125%/150% DPI.
  • web/styles-src/42-mixer-premium.css owns fader geometry; 41-mixer-performance.css owns zero-hover-filter performance overrides; 20-button-180.css owns palette; compile using scripts/build_ui_css.py and maintain standalone/native asset parity.

Incremental delivery (separate gates, never stale branch wholesale)

  • Gate 0 – Truthful UI fix: parity: native CH1/other channels/master show consistent Add FX without pretending unsupported DSP. Remove divergent native rack UI ownership; explicit unavailable/unassigned labels until each channel has real DSP. Fix selected toolbar tint and recessed fader. Tests/screenshots in Chromium/Firefox/WebView2.
  • Gate 1 – Engine foundation: explicit channelId protocol, stable insert IDs, channel/source registry and per-channel state; preserve working CH1 and Master output; deterministic routing/fader/mute tests. No real arbitrary multi-input promises yet.
  • Gate 2 – Live-safe mutation: atomic move order plan + lifecycle/ACK + rollback, ADD/REMOVE transactional, Orbit/editor continuity; stress test 1,000+ drag gestures with one transaction each, interleaved bypass/recall + device changes. Assert no detach/allocations in RT structural editing.
  • Gate 3 – Real multi-channel: active CH1–10 + MASTER real routing/FX, mono/stereo and solo/mute, true pre/post meters, multiple audio sources, PDC/CPU monitoring, snapshot/workspace integration. Hardware FAT after integrated and CI green, not after every trivial patch.

Acceptance / regressions

  • CH1–10 + MASTER Add opens same real catalog on supported configured buses, success is audible on intended source and nowhere else; unassigned input tells user exact status.
  • Drag any FX from position N to M with pointer and keyboard; preserves plugin instance, bypass, Orbit parameter, editor and vendor state; after relaunch order is identical. ESC, failure, stale revision and fast repeated drags remain unchanged, no silent state divergence.
  • No repeated graph modifications while dragging, no loss of audio device, no click/dropout from host graph commit in required tested configurations. Any remaining third-party plugin/driver issue is documented as external/hardware-unverified.
  • Visual fader bed + slot extends behind silver/gold cap, no change in thumb travel/tick calibration, no expensive mouse-hover effects; green-yellow selected toolbar with contrast check.
  • JS native bridge contract, DSP unit/integration and stress tests, CSS source/compiled/standalone parity, Playwright Chromium + Firefox, Windows MSVC/CTest packaging attestation; preserve P0: Cross-browser pointer stutter on standard Windows GPU (WebView2 + web mixer) #56 performance A/B conditions.
  • Preserve source tree single owner per component, one meaningful PR/gate, no unnecessary GitHub Actions reruns.

Activity

  1. masarray commented on Oct 27, 2026

    @masarray
    OwnerAuthor

    Progress P1A (2026-10-28): Draft PR #108 starts a clean, tested native 10-strip + MASTER source routing DSP core from main, separate from unmergeable #102. It handles explicitly mapped mono/stereo physical device pins, prepared gain/PAN coefficients, mute/solo, master mono/gain, silent unassigned channels, fail-closed invalid maps and 10-source synthetic audio tests. It is not yet wired into a live JUCE callback, therefore CH2–10 VST3 and MASTER FX remain unavailable. Release gate #15: real hardware crackle with zero FX must pass before P2.6 multi-VST activation. No wholesale stale branch merge.

  2. masarray commented on Oct 27, 2026

    @masarray
    OwnerAuthor

    HANDOFF FOR SUCCESSOR THREAD (2026-10-28): https://github.com/masarray/arsonmix/blob/docs/handoff-arsonmix-routing-fx-20261028/docs/HANDOFF_ARSONMIX_FX_ROUTING_20261028.md (PR #113, documentation-only). NOT ALL WORK IS COMPLETE. PR #108 P1A native 10-strip/master static routing core has Windows Build/CTest and GUI smoke PASS (runs 40593457359 / 40593457208), but it is not merged and currently dirty vs main. PR #102 DAW FX picker/ghost drag, orbit markers, right-click delete, atomic reorder is also unmerged and dirty; hardware de-click still unverified. Main has meanwhile merged #105 DearVR POLAR/capture triage, #106 meter ballistics, and #107 sample-accurate bypass de-click; don't overwrite. Issue #15 zero-FX hardware crackle remains OPEN. Next thread MUST read latest main before editing, reconcile P1A narrowly, keep #107 code, then finish P0 hardware FAT, real input/source registry and per-channel VST3 graph; no fake UI and no force-merging stale branches.

  3. masarray commented on Oct 28, 2026

    @masarray
    OwnerAuthor

    Cross-thread reconciliation — 2026-10-28, source of truth main@d30cf35dc74991286547765ec9927e356273785b

    User has approved new meter feel (“lumayan bagus”) and explicitly requests source-of-truth reconciliation, integration into main and forward-only progress across threads.

    Already on main — DO NOT regress: fast-drop/ease-out meter PR #106, true MASTER L/R visibility PR #104, dearVR Orbit/capture PR #105, sample-accurate FX de-click PR #107, latest ten-channel offline DSP P1A–P1D merge and lean CI PR #101. Browser/Windows/Pages exact-main CI runs 40604719342/40604719325/40604852202 all PASS.

    Next integration priority: PR #118 is a narrow follow-up to P1D, built from exact main, correcting true raw pre-fader sources at -∞ and exact directed DAG reachability. Windows CI 40605406924 not yet completed when reconciled: merge only after PASS and refreshed main/review check; NEVER force-merge old PR #112 (old CMake drops P1D CTests).

    Parallel threads / never bulk merge: PR #116 portable Studio BACKUP one-click, UI CI PASS and Windows native build pending (40605311929), independent user functionality; PR #102 FX reorder currently dirty/conflicted, must rebase/repair on actual latest main rather than overwrite newer Orbit/GUI/core; #97 historical GUI styling, #60 experimental Canvas. Stacked PR #110/#111/#115/#117 duplicate/refer to older P1D lineage and should not be merged wholesale without diffing all current contracts. P1D offline 10-strip/sends/bus processing is NOT proof of live CH2–10 native signal nor hardware FAT.

    Physical acceptance: issue #15 zero-FX crackle remains open; validate actual routing, latency/XRUNs, Windows audio driver and 10-channel live sound before claiming production readiness. Issue #56 is currently closed by another thread; do not reopen/close it from this reconciliation. Hardware pointer p95 remains separate from CI. Keep owner's existing selected BTN-180, LEDs, Orbit latest, faders, and no fake native DSP behavior.

    Release discipline: all new native artifacts should be direct GitHub Actions artifact URLs, not sandbox links. Do not publish a new release until exact-main Windows packaging gates green.

  4. masarray commented on Oct 28, 2026

    @masarray
    OwnerAuthor

    2026-10-28 milestone: LIVE 10-channel DRY Routing Lab merged, preserving all GUI/engine progress

    New main: e5c95a90441471cfd5d866c106a3ee9dba4c8fb3. PR #121 merged and old stale PR #119 closed as superseded. Scope is REAL hardware opt-in DRY LAB, NOT independent 10× VST3 production mode. Actual ASIO/WASAPI enabled hardware channels can map to ten separate mono/stereo dry paths in preallocated P1D audio callback. Device epoch/revision ACK and stale mapping fail-closed; CH1 original VST3 remains default when lab OFF. Native STOP-race output-clear and mute-before-DSP/256-sample zero-origin unmute fade are code-reviewed safety fixes.

    Regression trace: baseline main 61b2971 including #120 folded-metal fader, #106 fast-fall/ease-out master/input LED, #104 true MASTER L/R, #105 Orbit/dearVR, #107 sample-accurate FX bypass de-click, #116 backup, #118 pre-fader DAG hardening all preserved via exact changed-file blob diff; no unrelated GUI file replaced. Exact PR-head Chromium/Firefox/Orbit PASS and Windows MSVC CTest PASS. Postmerge exact-main browser PASS; Windows build in progress.

    Next gates: hardware FAT per docs/P1E_NATIVE_DRY_LAB_FAT_20261028.md on physical 8/10-channel interface with low headphone gain, actual port mask/epoch ACK and 0-FX crackle/latency reporting (#15). PR #102 FX drag/reorder is stale against latest main, touches DSP and visual components, MUST NOT merge wholesale. Multichannel per-strip live VST3, AUX live sends and atomic safe FX graph remain open. No claim performance-ready from CI.

  5. masarray commented on Oct 28, 2026

    @masarray
    OwnerAuthor

    Current GUI/FX cross-thread handoff: #125 — main@3b327bd5 has passed exact-main Windows/Chromium/Pages CI, including accepted 10CH+MASTER uniform GUI, bright BTN-180 palette, new fader recess, LCD/LED meter with 18ms/165ms native motion, real MASTER L/R. FX drag/reorder + atomic bypass/de-click from #122 already merged; stale PR #102/#97 are closed unmerged. Don't replay old styling/FX branches. Hardware audio #15 still OPEN.

  6. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    2026-10-30 G5A implementation in PR #178 (exact-head gates pending): LiveChannelFxGraph changed from 1-per-lane to immutable 8 sequential instances × 10 CH + 4 internal AUX, with stable per-instance shared ownership on control thread, append-only installAt, reuseAt allowing same-lane reorder without recreation, fixed per-lane counts and no callback allocation/refcount. Added actual JUCE CH2 2×VST test, response/identity and legacy compatibility checks. UI fixes: star-only FAVORITES, COMP label mapped to native Dynamics, concise portable import vs linked path, loaded insert green alignment, no fake active Master ADD.

    Do not close #95 or enable CH2–10 second ADD yet. G5B work still required: replace acceptedRoute one fxId per lane with bounded ordered stable inserts[], control-thread preflight/prepare of entire chain, epoch/revision/immutable graph swap and exact per-slot ACK, vendor editor/bypass/order/removal/portable state routing for CH2–10/AUX, coexist with CH1 existing 8 VST/Orbit, monotonic rollback and no autounmute; format-aware mono/stereo ADR R5 (#173/#174), parallel-bus PDC, MAIN actual post-sum FX before claiming full parity. Existing #15 crackle FAT remains unresolved. Keep one native callback and one graph authority; never draw fake LIVE slots.

  7. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    G5B integrated — 2026-10-30 (verified current main)

    MERGED: PR #186 into main@928dfdfd793bf69398c0c62f8a585d436efc6333. Exact-head source fcfe90f passed both Chromium UI smoke run 40999027779 and Windows MSVC + native CTest + ZIP verification run 40999027787. The GitHub push workflows on the combined merge commit remain separate verification; hardware FAT is not yet certified.

    Actually available: native 8-slot ordered G5B VST3 graph for the scheduled CH2–10/AUX lanes, FX RACK choose/add/remove/move draft, one muted routing APPLY with epoch/revision validation and positive native ACK. Six real MIX insert rows. Keeps CH1 legacy rack/Orbit and MS1 independent source/channel mono/stereo; factory-FX docs PR #187 merged in same main history. Latent plugins are currently rejected (no PDC); empty/unknown/invalid instances rejected before graph publication.

    Next non-cosmetic FX milestones (issue #95):

    1. G5C native lane identity + BYPASS + vendor editor: design stable per-instance IDs, audio-thread-safe bypass envelope and control-plane command/ACK; do not use slot-only identity for effects after MOVE, do not make GUI-only clickable bypass controls. Keep current 0-latency/format/epoch guards and no callback allocations.
    2. G5D genuine persistence: owner-controlled quiescent vendor-state capture and apply for every routed insert in portable Show; stable order/state, full hash/catalog verification, rollback if missing module; migrate old V4/legacy shows safely. Preserve [P1][BACKLOG] Portable VST Vault — user-owned state and fast session reopen #85 separation between user vault and shipped factory bundles [P2][OPTIONAL] Factory VST3 bonus — qualify available plugin modules per OS #185.
    3. PDC + mixer-wide format agreement: integrate MS2 typed buses [MS2 P0] Format-aware MONO/STEREO FX AUX/GROUP sends and VST bus layout — native+console #174 and latency compensation for PRE/POST/AUX parallel paths before allowing latent VSTs, then native master processing. No fake live Master FX and no second DSP graph.
    4. Hardware FAT P0 — Crackling with 0 VST3: diagnose WASAPI clock/input/output buffering before engine expansion #15: independently verify 2 VSTs on CH2, reorder/removal with audio running under controlled muted-apply policy, no CH1 state change, crackle at 0FX and populated rack, ASIO/WASAPI reopen, CPU deadline, no white flash/minimize restore.

    Known operator caveat: route-row FX power/bypass has not been wired to G5B native graph; do not claim it works just because the rack shows eight inserts. Legacy CH1 remains a separate native path.

    Avoid reopening completed G5A/G5B work or overwriting latest L2 or MS1 UI/routing controllers. Use atomic control-thread graph commits; run Windows build/CI once per integrated code milestone, not per cosmetic tweak.

  8. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    G5C1 implementation update (2026-10-30)

    Implementation: draft PR #190 at ed726f62da6a4318dda6efafc631810649011f based on merged G5B main 928dfdf.

  9. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    Dynamic channel integration requirement added 2026-10-30: #192. G5B #186 eight-slot graph is a useful foundation but processLegacyChannelOne makes CH1 a special case and stripCount=10 prevents reusable DAW-style +CHANNEL. DYN0 #193 adds only a stable-ID control roster; follow with one shared CH1..N native VST3 processor lifecycle and validated migration of existing CH1 Orbit/vendor state, then native-ACKed user ADD/REMOVE/CLONE. Do not duplicate JUCE processor objects on cloning or allocate in realtime. Existing current product remains 10CH until actual native PCM and GUI tests pass; do not close #95 or #192 because a control model exists.

  10. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    G5C merged; G5D native vendor-editor integration staged — 30 Oct 2026

    Completed: G5C PR #190 merged into main@593427d3bcbff9ebbb3ce7e2d0bd3e38333bbbba with exact-head Chromium/Windows + CTest PASS.

    Current major milestone: G5D PR #194 implements real native VST3 editor on each ACK-accepted routed input, replacing missing/no-op editor controls. A stable G5C per-instance token plus (epoch, revision, channel, slot) fences stale requests. One JUCE window retains shared_ptr<FxInstance> on the control thread, with the editor destroyed before lease release; structurally replacing/clearing routing and device changes close it, and existing 12Hz status tick closes orphaned editors on unexpected hotplug. No additional DSP graph, OS thread, timer, callback refcount, fake Master FX, or browser editor clone. Tests cover vendor ownership after old snapshot retirement, negative stale token, and real browser ACK command.

    Validation status: G5D initial native CI detected a pre-existing Node source-contract regex tied to the exact legacy FxRack=std::array<std::shared_ptr<...>> declaration, not a compiler error. PR #194 updated to retain that stable declaration; exact new HEAD 62cf72e Windows 41044963131 and Chromium 41044963587 underway. Do not merge without both green, and do not confuse CI with hardware-qualified vendor windows.

    Other active owners: #193 dynamic logical channel roster (not an actual unbounded PCM renderer yet), #189 MS2A format-aware send arithmetic/PCM failing Windows earlier. Do not overwrite them in G5D or claim VST3 PDC and portable vendor Show state, physical output multiport, or no-crackle FAT #15.

  11. masarray commented on Oct 31, 2026

    @masarray
    OwnerAuthor

    2026-10-31 latest-main FX parity blocker, detailed current-code audit: #192 issue comment. At main@d3293af72eef, web/bridge.js:202-305,691-785 still renders/commands CH1 legacy FX differently from routed CH2–10; LiveChannelFxGraph.h:264-287 uses CH1-only externalFx/LegacyChannelOneLease versus shared native processors; web/monitor.js:1016-1017 rejects duplicate legacy CH1 rack, and MAIN insert slots are explicitly disabled. This is NOT fixed by matching CSS. Require one actual native per-ChannelId multi-insert FX lease/lifecycle/editor/state/automation for CH1..N, AUX, group/FX bus and MAIN, no special branch, same compact strip presenter and native ACK, with no PDC false claims. Scope and migration acceptance integrated with #192 rather than a second competing FX host. Actual hardware #15 FAT remains independent.

  12. masarray commented on Oct 31, 2026

    @masarray
    OwnerAuthor

    ISSUE RECONCILIATION — historical/overlap closed, NOT declared solved. All-channel VST3 insert, drag/reorder, bus layout, latency/PDC and non-CH1 privilege are now MUST-PASS acceptance under engine #192 and ARD v5. Old issue also mixes cosmetic toolbar/fader requirements that remain protected by GUI baseline #176. The VST3 work is NOT finished by closing this duplicate.

    Current authority: PRD v5, ARD v5, native epic #192, GUI L2 #176, and hardware #15. Existing original comments and links remain.

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions