todo: park the open threads, ordered by what it costs to leave them

Everything from 2026-09-25 is committed and the suite is green, so nothing
here is half-applied. What is left is one loose edit outside this repo (the
sound-browser fix, which I could not commit), three restarts and a Pulsar
window reload to collect work already built, two things that need hands in
Ardour, three design questions with no code behind them yet, and the OSC arm
lens.

The 200 ms audition chop is recorded as UNRESOLVED with the one test that
settles it, not as fixed. All three sample-bank surfaces were ruled out: the
Foundry was not serving, La Cale has no index to load, and the sound browser
has no loop or timer on its click path. `paplay` on the same file decides
whether it is the audio path or Pulsar, and that is worth more than another
guess.

TODO.d/2026-09-26-open-threads.md

+# Open threads, parked 2026-09-25 evening
+
+PLN: *"imma fold for now, can we finish or save these as todos? i cant focus
+on this today"*. So: everything still open, in the order it costs to leave.
+Nothing here is urgent. Nothing here is half-applied — every code change from
+2026-09-25 is committed (`66318b6` → `084e692`) and the suite is green.
+
+---
+
+## 1. One loose edit, outside this repo — 10 seconds
+
+`~/.pulsar/packages/tidalcycles/lib/sound-browser.js` is **modified and
+uncommitted**. I could not commit it (outside the working directory).
+
+```
+cd ~/.pulsar/packages/tidalcycles
+git commit -am "sound-browser: stop before starting, and catch the play promise"
+```
+
+What it changes: `play()` used to fire before the outgoing sample was paused
+(two files overlapping while the new one starts — a flam on a one-shot), and
+the `play()` promise was dropped, so a missing file became an unhandled
+rejection, which Pulsar renders as a sticky red box. Neither is the 200 ms
+chop (see 3).
+
+## 2. Two restarts to collect what is already built — 1 minute
+
+```
+systemctl --user restart parvagues-bridge   # the GATE panel appears
+systemctl --user restart gig-log            # drops the dead MIDI subscription
+tools/rig-install.sh                        # links the gig-record black box
+```
+
+**Pulsar window reload** (`Window: Reload`) is what activates the notification
+timeouts and item 1. It restarts the `tidalcycles` package and kills its ghci,
+so never mid-track.
+
+**Do NOT hand-restart `lcxl3-driver`** while no LCXL is on the bus —
+`rig_units.py` records 86 crash-loop attempts from exactly that. It lands on
+the next `gig-up`.
+
+## 3. The 200 ms chop: unresolved, and one test settles it — 30 seconds
+
+Not found. All three "search and play sample banks" surfaces were ruled out:
+the Foundry server was not running (port 8765 is another project's
+`http.server`), La Cale cannot load (no `kits.json` exists anywhere), and the
+tidalcycles sound browser has no loop, no keyboard auto-audition and no timer.
+
+The discriminating test, next time you hear it:
+
+```
+paplay <the exact .wav that chopped>
+```
+
+- **chops from the terminal too** → it is the audio path, not the browser, and
+  that matters much more than the browser. Sampled `pw-top` on 2026-09-25
+  18:0x: the laptop sink carries 2,616,435 cumulative xruns but the counter was
+  **flat** at rest. If it climbs while you reproduce the chop, that is it.
+- **clean from the terminal** → Pulsar-side, and the browser gets instrumented.
+
+Say which window it was and it takes one pass.
+
+## 4. Your hands in Ardour — 2 minutes, whenever
+
+- **The nine faders at −inf.** Not a bug (`project_ardour-faders-are-post-gig`
+  — the save photographs the hands, not the mix). Raise them and Ctrl+S so the
+  session boots the mix you play.
+- **The `Mix` capture track**, optional, your call. Recipe:
+  `tools/check-mix.py --mix` prints it. Step 3 is the one that matters —
+  **disconnect its output from Master**, or it is a feedback loop through the
+  PA, which is what Ardour's own Add Track dialog leaves you with.
+
+## 5. Three design questions, no code until you answer — think, don't type
+
+From `TODO.d/2026-09-25-go-live-wizard.md`:
+
+1. T-2 alert on the **OLED** (cannot be missed, ~15 characters), a **Pulsar
+   notification** (holds a paragraph, one window among your tabs), or both?
+2. Is `[s] skip at my own risk` acceptable in the wizard? The alternative is a
+   gate that only advises, which is what we had on the 24th.
+3. How long is the during-set silence window before "Ardour is not recording"
+   fires — 30 s or 120 s? Needs one rehearsal, not an opinion.
+
+## 6. Still open from the belt-and-braces list
+
+- **Item 7: Ardour's OSC arm lens is broken.** `arm:false roll:false` for all
+  76 recorded minutes, `conflict:true` on every `rec` record. Find out whether
+  the OSC surface is even enabled before anything reads it. Any REC reminder
+  keys on the **files** lens only.
+
+## 7. Low-value findings, recorded so they are not re-discovered
+
+- **26 × `hoogle: not found`** in `~/.pulsar/nohup.out`. Caught by the
+  tidalcycles package, so harmless — but it is a synchronous `execSync` on the
+  renderer thread every time an autocomplete suggestion is highlighted. Either
+  install hoogle or accept the stall.
+- **`test_one_tidal_parser.py::test_no_new_file_parses_tidal_itself` fails**,
+  on `tools/fold-orbits.py`, and failed at HEAD before any of this. Debt to
+  record or to pay off — your call, not a regression.