Pre-compact cleanup: CHANGELOG + archive + memory

CHANGELOG.md

+# Changelog — ParVagues rig
+
+Sprint entries, newest first. Player-facing: what changed about *playing*, not
+about the code. Task IDs reference the L'Armada board; `n/a` where the work was
+unplanned (which, on a gig night, is most of it).
+
+## Sprint 1 — 2026-09-06 · rose_rouge plays on a new laptop
+
+The portable-rig thesis met a real box for the first time. Five faults stood
+between a fresh XPS24 and a track, and every one of them was invisible in the
+place you would look first: every check was green, every service `active`, every
+name correct.
+
+### Fixed
+- **The tray ignored a left click** (n/a). GNOME turns button-1 into a
+  StatusNotifierItem `Activate()`, Qt re-emits `activated(Trigger)`, and nothing
+  was connected — so the rig's only mid-set UI did nothing on its most obvious
+  gesture. Its D-Bus menu was provably healthy the whole time (19 items,
+  `AboutToShow` in 0.1s). Left and middle click now open it.
+- **midiviz had never been able to start on this box** (n/a). It imports PySide6,
+  which was absent from every interpreter here. Installed 6.11.2.
+- **The LCXL3 went dark on every unplug/replug** (n/a) — and PLN replugs
+  constantly while setting up. It now re-subscribes, re-asserts DAW mode and
+  repaints within one reconcile tick.
+- **Ardour's "Missing File" modal cost a launch** (n/a). 294 dead sources, so
+  Ardour sat 13 minutes with 0.08s of CPU and *no window at all*. Cleared: 270
+  verified alive on the 53G Freebox mirror, 24 = Take101, already written off.
+  The session is now a clean jig — 0 sources, 0 regions, 15 routes and 14
+  playlists intact.
+- **No sound anywhere, Tidal and Zoom alike** (n/a). One cause, three symptoms:
+  Ardour's ALSA backend had reserved BOTH cards over D-Bus, PipeWire released
+  them, and the laptop fell back to a single "Dummy Output". Then the JACK
+  backend started a *second* server, jackd2 on `hw:NVidia,3`, sending Ardour's
+  audio out an unplugged HDMI port while scsynth sat in PipeWire's graph. Now one
+  server, verified end to end.
+- **Ardour GUI was at 150%** (n/a) — `font-scale` back to 100%.
+
+### Changed
+- `gig-up.sh` launches Ardour through **pw-jack**, so its JACK backend takes
+  PipeWire's libjack instead of jackd2's. `launch_bin` gained an optional
+  `LAUNCH_WRAPPER`, empty for every other app.
+- `gig-up.sh` sweeps dead session references **before** Ardour opens, gated on
+  the mirror being mounted. Idempotent; never blocks the gig.
+
+### Infra
+- `rig-doctor.py` was asking the wrong interpreter about the wrong binding: it
+  probed `sys.executable`, so a pyenv shim answered for an interpreter no unit
+  uses, and its table credited midiviz to PyQt5 — the binding midiviz does not
+  import. It cried wolf on `mido` and `PyQt5` while both worked, and could not
+  see the one dep that was genuinely gone.
+
+### Doctrine
+- **A green check is not a working rig.** Every fault tonight passed its own
+  check. The doctor asked the wrong interpreter; `reconcile()` watched the one
+  leg that survives a replug; a `linked` unit looked identical to a broken one.
+- **Any-subscription is not a health signal.** A replug leaves debris: aconnect
+  kept printing the board wired to clients that no longer existed. Health means
+  *our* client, on both ends.
+- **One audio server.** jackd2 and pipewire-jack both installed, with ldconfig
+  preferring jackd2's libjack, is a trap that presents as three unrelated faults.
+- **Never throw a launched app's output away.** `>/dev/null 2>&1` in `launch_bin`
+  is why a 13-minute modal was invisible.

armada/tasks/completed-archive.md

@@ -7,6 +7,70 @@ snapshot in `board-archive.md`. Newest first.
 
 ---
 
+## n/a — rose_rouge on a new laptop: five green checks, five real faults
+- **Description:** First real test of the portable-rig work on a fresh XPS24, mid
+  gig-setup. Asked to play `rose_rouge.tidal` with the full viz stack (LCXL3, Pulsar,
+  HUD, midiviz) and to explain an unresponsive tray icon. Five independent faults stood
+  in the way, and none was visible where you would look first.
+- **Done:** rose_rouge played. Commits `8a69bbf` (tray left-click), `f8593e0` (rig-doctor
+  interpreter + PySide6 row), `4128a03` (fresh-jig plan + a stale command caught),
+  `a887029` (LCXL3 replug resilience), `3d4c798` (gig-up session sweep + cheatsheet
+  gotchas), `5d050dd` (Ardour via pw-jack). PySide6 6.11.2 installed; session cleared to
+  0 sources / 0 regions with all 15 routes intact; `font-scale` 150% -> 100%. Plan in
+  `docs/2026-09-06-ardour-fresh-jig.md`; setup gotchas in `docs/GIG-CHEATSHEET.md`.
+- **Learnings:**
+
+  *A green check is not a working rig — every fault tonight passed its own check.*
+  `rig-doctor` probed `sys.executable`, so PLN's pyenv shim answered for an interpreter
+  no unit runs (all five use `/usr/bin/python3`): it reported `mido` and `PyQt5` MISSING
+  while both worked, and still could not see the dep that was actually gone. A check
+  that cries wolf is a check nobody reads the day it is right.
+
+  *A stale dependency table hides a crash loop indefinitely.* The doctor credited midiviz
+  to PyQt5 — the binding midiviz does not import (it has been PySide6 since `30a704e`).
+  So it asked about the wrong binding, found it, and reported green while midiviz
+  crash-looped at `RestartSec=5` for a whole session. Two GUIs, two bindings, two rows.
+
+  *Ask about the thing that breaks, not the thing that survives.* `reconcile()` watched
+  the virtual-port -> SuperCollider leg, which is virtual-to-virtual and survives a
+  replug untouched. The board-side legs, which do not, were unwatched — so it reported
+  healthy through the exact event that killed the LEDs.
+
+  *Names cannot detect a replug.* The kernel destroys the sequencer client and rebuilds
+  it, dropping subscriptions, and rtmidi is never told. The board came back as client
+  **24 — the same number** — so `find_ports()` answered and every string matched while
+  nothing was delivered. Only the subscription knows.
+
+  *Any-subscription is not a health signal; it can be satisfied by ghosts.* The first
+  detector asked "does the DAW port have a subscriber?" and returned True on the very
+  state it existed to catch: a replug left debris, and aconnect kept printing the board
+  wired to clients 132/133 that no longer existed. Caught only by testing it against
+  the live fault. Health = OUR client (match `pid=` vs `os.getpid()`), on BOTH ends.
+
+  *One audio server, and the trap has three faces.* jackd2 and pipewire-jack both
+  installed, with ldconfig resolving `libjack.so.0` to jackd2's and no ld.so.conf.d
+  preference, means: the ALSA backend reserves BOTH cards and drops the laptop to "Dummy
+  Output" (no Tidal AND no Zoom); the JACK backend starts a *second* server on
+  `hw:NVidia,3` and sends audio out an unplugged HDMI port; and qjackctl shows a graph
+  that is not the one the rig is in. `pw-jack` is the whole fix, and it is the same
+  mechanism the SC unit already used (`LD_LIBRARY_PATH` -> pipewire jack dir).
+
+  *Never discard a launched app's output.* `launch_bin`'s `>/dev/null 2>&1` is precisely
+  why Ardour blocking on a modal for 13 minutes with 0.08s of CPU was undiagnosable.
+
+  *A refusal can be correct and its documented remedy still wrong.* `043` recorded the
+  modal fix as `--apply` alone; since then eleven takes' audio moved to the mirror, so
+  the tool now sees 270 recoverable sources and rightly declines. The doc was right when
+  written and wrong by that night — drift a plan should catch instead of a gig.
+
+  *`--allow-archived` is safe only for a caller that verified the copy itself.* The
+  refusal it overrides is right for an accident and wrong for deliberate archival, and
+  only the mirror's presence distinguishes them — hence gig-up's mount gate.
+- **Deps.** None. Unblocks the visuals last mile (blocked on xps22, unreachable:
+  resolves to `192.168.1.3`, "No route to host").
+
+---
+
 ## #85 — Pack-level sample grouping (the grouping above the folder)
 - **Description:** The Dirt-Samples FOLDER is a loose convention, not the true grouping —
   `rample*` is one bank sliced A0..S57, `808bd/cy/hc/mc` is one drum machine per voice,

backlog.md

@@ -2038,3 +2038,53 @@ url: "/home/pln/Work/Art/GLITCHWAVE/outputs/bg_live4.gif"
   url: "/home/pln/Work/Art/GLITCHWAVE/outputs/digital_flow.gif"
   url: "/home/pln/Work/Art/GLITCHWAVE/outputs/sunrise_flow.gif"
   url: "/home/pln/Work/Art/GLITCHWAVE/outputs/glitch_ocean_compressed.gif"
+
+# Rig — opened 2026-09-06 (gig night on the new XPS24)
+
+Floated during the session that got rose_rouge playing, not done. Full context:
+`armada/tasks/completed-archive.md` (top entry) + `docs/2026-09-06-ardour-fresh-jig.md`.
+
+## Blocking the visuals last mile
+- **Every scene asset is missing on this box.** `rose_bloom.gif` is nowhere under
+  `~/Work` or `~/Pictures`, and `~/Work/Art/GLITCHWAVE/` does not exist at all — so
+  **16 dangling `-- url:` refs**: 11 `.tidal` files point into `~/Downloads` (transient,
+  which is how rose_bloom vanished) and 5 into `GLITCHWAVE/outputs/`. PLN: "lets move it
+  to one repo ahah and ensure gifs are decently compressed". Plan: one home
+  (`visuals/scenes/`), compress before committing (gifs in git are permanent weight —
+  check whether the HUD accepts webp/mp4 first), rewrite all 16 refs repo-relative so a
+  fresh clone gets visuals like it gets code. **Blocked: needs the files off xps22**
+  (`192.168.1.3`, "No route to host" — PLN restarting it) or the Freebox.
+
+## Sound / mix
+- **`Tidal 12` fader is at `-inf dB` — SILENT.** Caught by gig-up's readiness gate.
+  rose_rouge's d12 is "LE JAZZ C'EST DU PIANO", so that orbit reaches nothing. Decide
+  mistake-vs-intent, then `tools/fader-baseline.py --restore` or `--capture` (Ardour
+  closed). Same shape as decision 1 in `armada/tasks/043`.
+- **`/usr/local/sbin/perf-audio` not deployed**, so gig-up cannot set performance mode
+  and the box played at `gear: None, mode: silent` with the CPU capped at powersave.
+  Needs the root-owned perf.sh + sudoers rule (README "Deploy mode-switching").
+- **Block size 1024 (~21 ms)** — high for livecoding. Worth tuning the PipeWire quantum.
+- **No ld.so.conf.d preference for PipeWire's libjack**, so `libjack.so.0` resolves to
+  jackd2's. `pw-jack` works around it per-app; a system-wide preference would also make
+  qjackctl show the graph the rig is actually in, instead of an empty jackd2 one. Needs
+  sudo, and would leave jackd2 effectively unused — which is the point.
+- **The headset mic auto-connects into `SuperCollider:in_1/2`** on every SC boot
+  (`SC_JACK_DEFAULT_INPUTS=system`). `check-audio-graph.sh` prunes it, but it recurs.
+- **Re-run `check-audio-graph.sh` at the venue with the UMC plugged in** — it cannot
+  prove the last monitor hop until the interface exists (2 warnings tonight, both that).
+
+## The jig, second half
+- **Only the reference-clearing half of the fresh-jig plan is built.** The stem sweep
+  (copy -> verify -> clear, idempotent, at launch as well as close) is designed but not
+  implemented. Three open questions still need PLN: per-gig sessions vs one stable jig;
+  where swept stems land (`take-master.py`/`take-lens.py` should own it, not a new tool);
+  whether the local 53G ever comes back or the mirror stays the archive of record.
+
+## Housekeeping
+- **`rig-doctor` does not check LV2 plugins** — the same gap class as the PySide6 miss.
+  The missing `lsp-plugins-lv2` / `zynaddsubfx-lv2` only surfaced via Ardour's own dialog.
+- **The rig now carries two Qt bindings** (PyQt5 for perf-tray, PySide6 for midiviz).
+  Documented in the doctor rather than unified; a PyQt5 port of midiviz is ~30 refs but
+  would move it off the Qt6 it was written and tested against.
+- Untracked: `start_minimal.scd`, `sandbox/debug.tidal`. Also `BootTidal.hs.broken` and
+  `BootTidal.visuals.broken.hs` are still sitting in the repo root.