valentine, branch HEAD

add MIT LICENSE (public release)

2026-06-02T05:10:42Z

commit e7689c3e5f768fbe404f563a85bce70b72921236 parent a1cbfd6897b236c4e5ab1c76bf5fd0515a3ef4e0 Author: Matthew Gantenbein Date: Tue, 2 Jun 2026 00:10:42 -0500 add MIT LICENSE (public release) Co-Authored-By: Claude Opus 4.8

feat: install.sh + uninstall.sh, updated README (no dmg — TUI ships as scripts)

2026-06-02T04:37:47Z

commit a1cbfd6897b236c4e5ab1c76bf5fd0515a3ef4e0 parent ea6367f204310f6e9bf3d0b960a12d7520bfac2a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 23:37:47 -0500 feat: install.sh + uninstall.sh, updated README (no dmg — TUI ships as scripts) A terminal tool doesn't benefit from a .app/.dmg (no Finder launch; needs a TTY). So Valentine ships the way TUI users expect: install.sh builds the release binary, installs it to ~/.local/bin, and deploys the 10 themes to ~/.config/valentine/themes/. uninstall.sh removes it (--purge for config). README Install section rewritten + feature list updated (routing editable, 10 shared themes, transparency). Verified install.sh end-to-end. Co-Authored-By: Claude Opus 4.8

fix: persist theme name + transparency across restarts (Hydra parity)

2026-06-02T04:29:51Z

commit ea6367f204310f6e9bf3d0b960a12d7520bfac2a parent 7fbd06301d79a498edce2c6e51a3a0370d5bebb3 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 23:29:51 -0500 fix: persist theme name + transparency across restarts (Hydra parity) Valentine had NO theme persistence — it forgot both the theme and transparency on every restart (worse than the Hydra bug, which at least kept the name). Now ~/.config/valentine/ui.toml stores theme name AND transparency together; resolve() restores both on startup (CLI --theme/--transparent override + persist). Saved on theme-picker Enter, T toggle, and explicit CLI choice. Regression tests: transparent_theme_file_loads_transparent + ui_settings round-trip. 71 tests. Co-Authored-By: Claude Opus 4.8

themes: make red-pink 'Ember' the default (drop Hydra's neutral default)

2026-06-02T04:07:29Z

commit 7fbd06301d79a498edce2c6e51a3a0370d5bebb3 parent d2841d68bc38a87ef5fde99a00c8bd1082f2a393 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 23:07:29 -0500 themes: make red-pink 'Ember' the default (drop Hydra's neutral default) default.toml is now Valentine's own plum/rose/coral palette (rose #ff5fa2 highlight) in the shared Hydra format — distinct from Hydra's neutral default, which was just Hydra's. The other 9 themes stay synced with Hydra. Updated the default-color test + docs. Co-Authored-By: Claude Opus 4.8

themes: ship exactly 10, remove Navi from the repo

2026-06-02T04:01:19Z

commit d2841d68bc38a87ef5fde99a00c8bd1082f2a393 parent aa3754980d69c830681aef68e45ee7a37a6eeb48 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 23:01:19 -0500 themes: ship exactly 10, remove Navi from the repo Drop the tracked themes/navi/ (ember/mono/slate) — personal palettes don't belong in the repo (boundary rule). The 10 bundled themes (compiled in via include_str!) are unchanged and remain the shipped set. Updated theme.rs doc to point users at ~/.config/valentine/themes/ for their own additions; the Navi files are preserved off-repo. Builds clean, navi-free. Co-Authored-By: Claude Opus 4.8 (1M context)

feat: sync 10 themes with Hydra + transparency toggle; separate Navi themes

2026-06-02T03:46:55Z

commit aa3754980d69c830681aef68e45ee7a37a6eeb48 parent 50199c69e1068614cec5a8dbfa5ed2e158a79f82 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 22:46:55 -0500 feat: sync 10 themes with Hydra + transparency toggle; separate Navi themes - Adopt Hydra's theme file format verbatim (top-level name/transparent + [palette] of 10 keys) so the SAME .toml files work in both apps. - Bundle Hydra's 10 themes: default, transparent, nord, gruvbox, dracula, rose-pine, tokyonight, catppuccin-mocha, solarized-dark, monochrome. - Valentine's own ember/slate/mono moved to themes/navi/ (separate; load only if copied to ~/.config/valentine/themes/). Repo no longer auto-bundles them. - Transparency: Theme.transparent + surface()=Color::Reset; 'T' toggles live, --transparent CLI flag, main bg uses surface(). Mirrors Hydra exactly. - Derive Valentine-only roles (border_focus/bg_selected/armed/meter_*) from the shared palette so any Hydra theme just works. 69 tests, 0 warnings. Co-Authored-By: Claude Opus 4.8

revert: remove monitor-via-mixer fader from the UI (device-safe)

2026-06-01T22:54:53Z

commit 50199c69e1068614cec5a8dbfa5ed2e158a79f82 parent d6d4dfeaf98ef73310e931708f610c309db77dc2 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 17:54:53 -0500 revert: remove monitor-via-mixer fader from the UI (device-safe) The mix-insert output fader was the only thing that could corrupt routing and wedge the device (SET_MUX 0x3). Removed it from the Monitor tab + app state: monitor_key now only does mute/dim; panel shows the read-only hardware knob + mute/dim. Confirmed with the user this device has no writable output volume and the ADAT loudness is downstream (external converter / monitor knobs), not fixable in software. Core group/route methods kept (used by spike diagnostics). Everything else intact: routing editor, presets, per-input meters, themes, input toggles, mixer + clear. 67 tests, 0 warnings. Co-Authored-By: Claude Opus 4.8

test: offline mux round-trip suite disproves 4 corruption theories

2026-06-01T21:18:08Z

commit d6d4dfeaf98ef73310e931708f610c309db77dc2 parent acf91fce8cecdf5ac8d5aabb9e20601124513252 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 16:18:08 -0500 test: offline mux round-trip suite disproves 4 corruption theories Adds MuxState::roundtrip_decode + tests proving the encoding is structurally sound (faithful round-trip on the user's real routing incl ADAT 7/8, all ADAT dests present, no dup dests, per-table lengths match assignment ≤77). The 0x3 on SET_MUX is therefore not in our encoding logic — it's live-wire framing or a device-state precondition. Documented as the key open bug. Co-Authored-By: Claude Opus 4.8

feat(spike): adatverify — fresh-process read of actual ADAT routing

2026-06-01T20:59:37Z

commit acf91fce8cecdf5ac8d5aabb9e20601124513252 parent d3394aa85a304fca36ade45a3e9e114bdcbc0ccb Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:59:37 -0500 feat(spike): adatverify — fresh-process read of actual ADAT routing Reads (no writes) what truly feeds each ADAT out, to confirm whether the device accepted ADAT<-Mix or silently kept ADAT<-PCM. Run after adatset to distinguish 'write not applied' from 'mix bus not attenuating'. Co-Authored-By: Claude Opus 4.8

feat: full TOML theme support — --theme CLI, bundled registry, in-app switcher

2026-06-01T20:48:54Z

commit d3394aa85a304fca36ade45a3e9e114bdcbc0ccb parent 59fadb844d09c033ffa3ff387ec8817cf1e6ee86 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:48:54 -0500 feat: full TOML theme support — --theme CLI, bundled registry, in-app switcher - BUILTIN_THEMES registry: ember (default), slate, mono (all TOML, compiled in). - CLI: --theme , --list-themes, --help. Loads a bundled name, a stem from ~/.config/valentine/themes/, or any .toml path. - In-app theme switcher: 't' opens a picker that live-previews on ↑↓, Enter keeps, Esc reverts. Active theme shown in the title bar. - Theme::resolve/by_name/available_themes; user themes dir ~/.config/valentine/themes/. Co-Authored-By: Claude Opus 4.8

diag(spike): adatset writes each routing table separately to find 0x3

2026-06-01T20:40:58Z

commit 59fadb844d09c033ffa3ff387ec8817cf1e6ee86 parent 3d156748656c3759baf294fc5e4197a5adc96af9 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:40:58 -0500 diag(spike): adatset writes each routing table separately to find 0x3

chore(spike): adatset reports which write throws the error

2026-06-01T20:32:48Z

commit 3d156748656c3759baf294fc5e4197a5adc96af9 parent 537036c0d3f21caf88150c652ed49b2e11559839 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:32:48 -0500 chore(spike): adatset reports which write throws the error

feat(spike): adatset — persisting ADAT attenuation listen-test

2026-06-01T20:30:13Z

commit 537036c0d3f21caf88150c652ed49b2e11559839 parent 41649b95e76d4e15572f71771b4b48abcba2f10c Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:30:13 -0500 feat(spike): adatset — persisting ADAT attenuation listen-test Sets ADAT via mixer + a given dB and LEAVES it (no revert), so we can hear whether writing the mix bus actually attenuates ADAT. 'direct' restores ADAT<-PCM. Isolates the audible effect from all TUI plumbing. Co-Authored-By: Claude Opus 4.8

fix: faders effective by default — no toggle needed

2026-06-01T20:18:38Z

commit 41649b95e76d4e15572f71771b4b48abcba2f10c parent 8291d0f2b2aa4f48658acc4421a3b08a4065f72f Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:18:38 -0500 fix: faders effective by default — no toggle needed The adatfader probe proved the device chain works (route+level land at -40dB); the failure was the TUI gating the fader behind a 'v' toggle that fought the already-via-mixer state. Now: ←→ owns the routing (auto-routes the group via the mixer on first use), Monitor-tab entry detects already-routed groups and SEEDS each fader from the device's actual current level (get_group_level), and the fader bar always shows. 'v' is now just an optional back-to-DAW. No toggling required. Co-Authored-By: Claude Opus 4.8

feat(spike): adatfader — trace the ADAT via-mixer fader chain on hardware

2026-06-01T20:12:46Z

commit 8291d0f2b2aa4f48658acc4421a3b08a4065f72f parent 9d02e362e6366d46b63e5282b34f919a9c37a2ce Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 15:12:46 -0500 feat(spike): adatfader — trace the ADAT via-mixer fader chain on hardware Runs route_group_via_mixer + set_group_level for the ADAT group and reads the device back at each step (ADAT out routing, mixer-input feeds, per-bus gains) to pinpoint why the fader has no audible effect. Restores ADAT to direct after. Co-Authored-By: Claude Opus 4.8

fix: coalesce fader writes per-frame so the fader actually works

2026-06-01T19:58:56Z

commit 9d02e362e6366d46b63e5282b34f919a9c37a2ce parent b873468a5e13f618f1058f5f4d6d5256f6229b2d Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 14:58:56 -0500 fix: coalesce fader writes per-frame so the fader actually works Previous drain dropped queued events (fader did nothing); before that, a per-keypress 8-write fader cascaded to silence. Now a fader keypress only updates the in-memory level + flags a pending write; all queued key events are processed in one batch per frame, then flush_pending_writes does ONE device write at the final value. Holding the key ramps smoothly; no cascade, stays responsive. Co-Authored-By: Claude Opus 4.8

fix: drain queued input after each key so slow writes can't cascade

2026-06-01T19:33:39Z

commit b873468a5e13f618f1058f5f4d6d5256f6229b2d parent d6965e725d2eba1c17f2bd45d4b747f4b520cb3a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 14:33:39 -0500 fix: drain queued input after each key so slow writes can't cascade A control change can be many USB writes (8 for an ADAT group fader); while that runs, held/repeated keys pile up in the terminal buffer and then replay, making one fader press cascade straight to silence. After handling a key, drain any already-queued events (non-blocking) so it's one action per redraw. The fader math itself was correct/linear — this was buffered key-repeat. Co-Authored-By: Claude Opus 4.8

feat(spike): adatcheck — diagnose hot ADAT output (routing + DAW levels)

2026-06-01T19:18:08Z

commit d6965e725d2eba1c17f2bd45d4b747f4b520cb3a parent 5a53c4fc3ff00d1e2c003db19882eb19fad6b08a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 14:18:08 -0500 feat(spike): adatcheck — diagnose hot ADAT output (routing + DAW levels) Read-only: shows what feeds each ADAT out (PCM-direct like FC, vs a mix bus that would sum the wall-of-unity mixer to a hot level) and compares DAW send levels PCM 13-20 (ADAT) vs PCM 1-2 (main/HP). Pinpoints whether the loud ADAT is induced routing, DAW levels, or the external converter. Co-Authored-By: Claude Opus 4.8

feat: generalize monitor fader to groups; add ADAT 1-8 group

2026-06-01T19:13:31Z

commit 5a53c4fc3ff00d1e2c003db19882eb19fad6b08a parent 4e9662a218da717b7626d46eaefa800582145330 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 14:13:31 -0500 feat: generalize monitor fader to groups; add ADAT 1-8 group Refactored the monitor-via-mixer fader into reusable MonitorGroup (out/bus/ mixer-in/pcm bases + count) with route_group_via_mixer/route_group_direct/ set_group_level/group_is_via_mixer. MONITOR_GROUPS = Analogue 1-2 (buses 0-1) and ADAT 1-8 (buses 2-9, PCM 13-20) — non-overlapping so both work at once. Monitor tab: [ ] switches group, v routes it via mixer, arrows are its fader; title shows the active group. ADAT is the user's main monitoring path. Co-Authored-By: Claude Opus 4.8

feat: software monitor fader via mixer-inserted output path

2026-06-01T19:02:53Z

commit 4e9662a218da717b7626d46eaefa800582145330 parent c89c962772742c013271b3b881fc1f8ba948ce51 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 14:02:53 -0500 feat: software monitor fader via mixer-inserted output path The device has no writable output volume, so to give Valentine real monitor level control: 'v' on the Monitor tab routes Analogue Out 1/2 through Mix A/B (Mixer In 1/2 <- PCM 1/2; Out <- Mix A/B), and ←→ sets Mix A/B gain as the monitor fader. Reversible (v again = direct from DAW). Core adds route_monitor_via_mixer/route_monitor_direct/set_monitor_level/ monitor_is_via_mixer. Monitor panel shows hardware knob (read-only) + the software fader. Prototype for the monitor pair; ADAT outs to follow once verified on hardware. Co-Authored-By: Claude Opus 4.8

feat: 'c' clears the mixer to silence (auto-backs-up first)

2026-06-01T18:51:45Z

commit c89c962772742c013271b3b881fc1f8ba948ce51 parent 4358539aeb5174202e6c812c014e725e629b9f93 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:51:45 -0500 feat: 'c' clears the mixer to silence (auto-backs-up first) The device sits in a wall-of-unity mixer (288 crosspoints at 0 dB = up to ~+27 dB summing per bus) — a prime loudness suspect. Mixer tab 'c' saves the current mixer to a 'mixer-backup' preset, then sets every crosspoint to silence via Scarlett::clear_mixer. Reversible: load 'mixer-backup' with p. Diagnostic step to confirm whether the mixer feeds the monitors. Co-Authored-By: Claude Opus 4.8

feat(spike): presetcheck — read-only mixer loudness diagnostic

2026-06-01T18:46:14Z

commit 4358539aeb5174202e6c812c014e725e629b9f93 parent 82fdcb2a7e26425769871c000972367b78f1fb2d Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:46:14 -0500 feat(spike): presetcheck — read-only mixer loudness diagnostic Reports non-silent mixer crosspoints, per-bus active counts, and unity-gain crosspoints, to diagnose whether 'too loud' is a hot mix bus (summing many unity sources) vs a preset round-trip issue. Read-only. Co-Authored-By: Claude Opus 4.8

fix: presets capture+apply routing; p-only picker with instant Y/N confirm

2026-06-01T18:39:39Z

commit 82fdcb2a7e26425769871c000972367b78f1fb2d parent 56f904ea99b0dbd86b063f8542ccfd4f446efa19 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:39:39 -0500 fix: presets capture+apply routing; p-only picker with instant Y/N confirm - Preset now includes full routing (mux), captured on save and applied on load — fixes 'preset doesn't switch settings' (routing was the missing, main thing). apply_preset writes routing atomically; load refreshes views. - Preset picker is now the 'p' key only (dropped L). - No staging: pick in the picker -> single yes/no confirm window -> instant apply. Removes the pending-preset stage/confirm-on-status flow. Co-Authored-By: Claude Opus 4.8

feat: multi-preset support + cross-screen picker/save modals

2026-06-01T18:23:02Z

commit 56f904ea99b0dbd86b063f8542ccfd4f446efa19 parent 3c7c75fb6710c6bfa0746b2c94c9972de0fac496 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:23:02 -0500 feat: multi-preset support + cross-screen picker/save modals Presets are now individual .json files. S opens a save-as name entry; L or p opens a preset picker (↑↓ select, Enter loads → confirm to apply) — both reachable from ANY tab via a modal overlay that captures input first. Replaces the single-slot preset.json with presets_dir/preset_file/list_presets. Co-Authored-By: Claude Opus 4.8

feat: routing grouped by stereo pairs, staged + atomic pair writes

2026-06-01T18:18:11Z

commit 3c7c75fb6710c6bfa0746b2c94c9972de0fac496 parent 273b13d1071a6d14921f41b73d491b1477c95325 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:18:11 -0500 feat: routing grouped by stereo pairs, staged + atomic pair writes Routing panel now lists destinations as stereo pairs (Analogue Out 1-2, ...); left/right stages a SOURCE PAIR, Enter commits both channels in one atomic write (set_routes) so a pair never passes through a half-routed state. Off maps both to Off. Core adds dest_pairs/source_pairs/PairRow + set_routes. Co-Authored-By: Claude Opus 4.8

feat: stage-then-confirm for routing changes and preset load

2026-06-01T18:10:32Z

commit 273b13d1071a6d14921f41b73d491b1477c95325 parent 81a1380dbaf27b6b7929e3722e592f28547b734a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:10:32 -0500 feat: stage-then-confirm for routing changes and preset load Routing: left/right now STAGE a source for the selected destination (no write); Enter applies, Esc cancels — avoids momentarily routing through every source while cycling (feedback-loop risk). Staged source shown in accent with a *. Preset load: L stages the preset, Enter/Y applies, Esc/N discards (it rewrites many controls at once). Status line + help reflect the confirm step. Co-Authored-By: Claude Opus 4.8

feat: interactive routing edit (hardware-verified write path)

2026-06-01T18:01:34Z

commit 81a1380dbaf27b6b7929e3722e592f28547b734a parent 3b1ff1bf89f035fd06526cc9de818e9d6e57a7f3 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 13:01:34 -0500 feat: interactive routing edit (hardware-verified write path) Routing panel is now editable: up/down select a destination, left/right cycle its source (Off = unrouted), written immediately via Scarlett::set_route (MuxState -> encode 3 tables -> write_routing_tables). Adds mux dest_list/ source_list/get/encode_all; Device.routing is now MuxState. The write path was proven safe first (muxcheck + --write-noop on hardware). Co-Authored-By: Claude Opus 4.8

feat(spike): muxcheck --write-noop — verify write path with a no-op rewrite

2026-06-01T16:14:43Z

commit 3b1ff1bf89f035fd06526cc9de818e9d6e57a7f3 parent 40d126c571eef6210803e1c52e6b32e2bbbbe37d Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 11:14:43 -0500 feat(spike): muxcheck --write-noop — verify write path with a no-op rewrite Opt-in flag writes the current routing back unchanged via write_routing_tables, then re-reads to confirm zero routes changed. Safest possible exercise of the real SET_MUX path on hardware before enabling actual routing edits. Co-Authored-By: Claude Opus 4.8

feat(core): routing (mux) model + write encoding + read-only safety probe

2026-06-01T16:13:03Z

commit 40d126c571eef6210803e1c52e6b32e2bbbbe37d parent 60b2ccd30be1b4caeb5a5eb16238e81b2b2d94a9 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 11:13:03 -0500 feat(core): routing (mux) model + write encoding + read-only safety probe scarlett-core/src/mux.rs: port-number<->hardware-id conversions, MuxState (flat mux[dest]=src), 3-table mux_assignment, encode_table() packing dst_id|(src_id<<12) — faithful port of scarlett2_usb_set_mux. matrix.rs adds write_routing_tables() (3 SET_MUX msgs) + dry_run_routing(). spike/muxcheck: READ-ONLY probe that re-encodes current routing and verifies it reproduces the device byte-for-byte (no-op) before any write is enabled. 60 core tests. Co-Authored-By: Claude Opus 4.8

fix: unify meters tab + inputs panel onto one dBFS scale (clip=0dB)

2026-06-01T15:56:05Z

commit 60b2ccd30be1b4caeb5a5eb16238e81b2b2d94a9 parent c83c637bb41372512d77d7153026470c8cd06b28 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 10:56:05 -0500 fix: unify meters tab + inputs panel onto one dBFS scale (clip=0dB) The two views disagreed: meters tab was linear (raw/4095) while inputs used log dBFS with full-scale 16384 (~12dB too high, never hit red). Now both use raw_to_dbfs + a shared meter_fill() window (-48..0) + shared color thresholds, and METER_FULL_SCALE = 4095 so an overloading input pegs at 0 dBFS / red. Co-Authored-By: Claude Opus 4.8

tune: hotter input meters — -48dB window, amber@-18 red@-6

2026-06-01T15:50:50Z

commit c83c637bb41372512d77d7153026470c8cd06b28 parent d31fadaa431b331e68027618f7b88c3642cb8119 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 10:50:50 -0500 tune: hotter input meters — -48dB window, amber@-18 red@-6 Narrowed the inputs meter window from -60 to -48 dBFS so program material fills more of the bar, and moved color thresholds to DAW-typical points (amber ~-18, red ~-6). dB calibration (METER_FULL_SCALE) unchanged. Co-Authored-By: Claude Opus 4.8

fix: meter each input where it's routed (PCM capture), dBFS bars

2026-06-01T15:41:44Z

commit d31fadaa431b331e68027618f7b88c3642cb8119 parent 0bcf3fd3f6d2d95b646e6d4025567e301c3eb53a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 10:41:44 -0500 fix: meter each input where it's routed (PCM capture), dBFS bars Inputs are normally routed straight to the DAW (PCM capture), not the mixer, so the old mixer-input meters showed nothing for ADAT and wrong levels. Now read each source's PCM-capture meter via live routing (source_to_pcm_capture) and draw a dBFS bar (-60..0). Verified on hardware: Analogue 3->PCM cap 3 (raw 2), ADAT 1->PCM cap 13 (raw 10). Adds pcm_capture_level/raw_to_dbfs (tested) + Source::hw_id. Co-Authored-By: Claude Opus 4.8

feat(spike): metermap v2 — dump routing + peak-hold every raw index

2026-06-01T15:25:48Z

commit 0bcf3fd3f6d2d95b646e6d4025567e301c3eb53a parent 872781d3d87cfc3d871c0c628a5ce49c9c96eb49 Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 10:25:48 -0500 feat(spike): metermap v2 — dump routing + peak-hold every raw index Prints full decoded routing (sink<-source) and peak-holds all 65 GET_METER indices for ~12s, labeling each by destination + routed source. Lets us see exactly where ADAT is routed and which raw index moves (channel mapping) plus real signal peaks (full-scale calibration vs DAW). Read-only. Co-Authored-By: Claude Opus 4.8

feat: live per-input meters on the inputs page (hardware-verified mapping)

2026-06-01T15:20:49Z

commit 872781d3d87cfc3d871c0c628a5ce49c9c96eb49 parent 406fbc57ff9927f45af34480776e35ea1cfd9e5a Author: Matthew Gantenbein Date: Mon, 1 Jun 2026 10:20:49 -0500 feat: live per-input meters on the inputs page (hardware-verified mapping) Each input row shows a level bar driven by the mixer-input meter span. Mapping confirmed on the real 18i20 (2026-06-01): mixer input N = raw GET_METER index 40+(N-1); feeding analogue input 3 lit raw 42. Inputs tab now polls meters every tick; stereo rows show the louder channel. Also fixed metermap probe labels/offset and added mixer_input_level/raw_index helpers (tested). Co-Authored-By: Claude Opus 4.8

feat(spike): read-only metermap probe to verify source->meter mapping

2026-06-01T02:30:11Z

commit 406fbc57ff9927f45af34480776e35ea1cfd9e5a parent ec11802b58ca5856299b2473146e63c30c0dadce Author: Matthew Gantenbein Date: Sun, 31 May 2026 21:30:11 -0500 feat(spike): read-only metermap probe to verify source->meter mapping Samples GET_METER 20x and prints raw levels grouped by meter_map spans (with raw index + destination), plus a peak summary, so we can empirically see which index moves when an input is fed — verifying the mixer-input meter mapping before wiring meters into the inputs UI. Reads only; changes nothing. Co-Authored-By: Claude Opus 4.8

feat(core): port kernel meter-level remap (dest->raw index)

2026-06-01T02:23:16Z

commit ec11802b58ca5856299b2473146e63c30c0dadce parent 019b2184f709572d24c47b51cbd50164d2c82761 Author: Matthew Gantenbein Date: Sun, 31 May 2026 21:23:16 -0500 feat(core): port kernel meter-level remap (dest->raw index) scarlett-core/src/meter.rs faithfully reimplements scarlett2_update_meter_level_map: build_level_map() turns the live routing (mux[dest]=src) + the device meter_map into dest->raw-index, with source dedup and Off->zero. Tested (4 cases). Foundation for per-channel meters; source->meter still needs routing inversion + hardware verification. Co-Authored-By: Claude Opus 4.8

feat: stereo-pair input view (default) with mono toggle

2026-06-01T02:06:52Z

commit 019b2184f709572d24c47b51cbd50164d2c82761 parent e9f1cc736c4236d848aa32a7dff8cf98b31a1af7 Author: Matthew Gantenbein Date: Sun, 31 May 2026 21:06:52 -0500 feat: stereo-pair input view (default) with mono toggle Inputs panel now groups odd/even same-kind sources into stereo rows by default (s toggles mono). Preamp toggles apply to both channels of a pair and stay in sync; 48V handles multi-group rows. Vertical scroll, mode shown in the title and help. Co-Authored-By: Claude Opus 4.8

fix: reconcile inputs panel with real sources API; add source catalog

2026-06-01T01:57:16Z

commit e9f1cc736c4236d848aa32a7dff8cf98b31a1af7 parent b8b9de94b30c3533c6af9f508834e31367e41dbc Author: Matthew Gantenbein Date: Sun, 31 May 2026 20:57:16 -0500 fix: reconcile inputs panel with real sources API; add source catalog The previous commit rewrote inputs.rs against a sources::{Source,SourceType} API that was never created, leaving the workspace uncompilable. Point the panel at the actual scarlett-core::sources catalog (all 39 sources: analogue incl. talkback, ADAT, S/PDIF, PCM), wire Device.sources + main.rs call sites, and drop dead code. Also: lowercase wordmark/tabs, theme loads ~/.config first. Co-Authored-By: Claude Opus 4.8

feat: Implement Inputs panel rewrite to support all input types with per-channel strips and meters.

2026-05-31T23:18:39Z

commit b8b9de94b30c3533c6af9f508834e31367e41dbc parent 35effd89312ab0a7013f798b7f5b0a1d3ea82932 Author: Matthew Gantenbein Date: Sun, 31 May 2026 18:18:39 -0500 feat: Implement Inputs panel rewrite to support all input types with per-channel strips and meters. Co-authored-by: aider (ollama/qwen2.5-coder:14b)

fix: correct capitalization in input panel title

2026-05-31T23:18:15Z

commit 35effd89312ab0a7013f798b7f5b0a1d3ea82932 parent 14b34b172794748c6d3fa9f23f7ba33fc19bb37c Author: Matthew Gantenbein Date: Sun, 31 May 2026 18:18:15 -0500 fix: correct capitalization in input panel title

Valentine: terminal control panel for Scarlett 18i20 3rd Gen

2026-05-31T22:02:20Z

commit 14b34b172794748c6d3fa9f23f7ba33fc19bb37c Author: Matthew Gantenbein Date: Sun, 31 May 2026 17:02:20 -0500 Valentine: terminal control panel for Scarlett 18i20 3rd Gen scarlett2 USB control protocol (clean-room, rusb/EP0) + ratatui TUI. Six panels (Inputs/Monitor/Mixer/Routing/Meters/Clock), presets + standalone NVRAM save, swappable themes (Ember default). Protocol layer hardware-validated on the 18i20 g3; routing read-only for now. Co-Authored-By: Claude Opus 4.8