docs(tasks): park the card-profile learnings — wireplumber 0.4 lua rules, DMIC…

docs(tasks): park the card-profile learnings — wireplumber 0.4 lua rules, DMIC guard, ardour sink follow

tasks/launcher-consolidation.md

@@ -25,3 +25,24 @@ Queued 2026-09-20, after gig prep. Full picture established that day (see
    README, rig_units comments.
 4. Verify with: `tools/gig-up.sh --converge` from a cold state, RIG UP from the
    Bridge, desktop icon — all three must paint the board and end in GO.
+
+## Parked 2026-09-20 evening (card-profile incident)
+
+- **`~/.config/wireplumber/wireplumber.conf.d/50-alsa-config.conf` is inert.**
+  This box runs WirePlumber **0.4.17** (Lua era): `monitor.alsa.rules` SPA-JSON
+  fragments are silently ignored. The period-size/headroom tuning in it has
+  never applied. 0.4-correct home: `~/.config/wireplumber/main.lua.d/50-alsa-config.lua`
+  (copy the system file and edit — user file replaces wholesale; a partial
+  `alsa_monitor.rules = {...}` in a numbered fragment CLOBBERS the built-in
+  rules table by index and dropped the card to profile `off`, proven
+  2026-09-20, reverted).
+- **SOF DMIC guard not restored.** The old `51-disable-sof-dmic.conf` was a
+  0.5 fragment — it never worked on 0.4, and the DMIC node (`pcm4c`, the PCM
+  that hung capture in #36) has been open all evening without harm. Restore as
+  a rule appended INSIDE a full copy of `50-alsa-config.lua`, matched by node
+  name `alsa_input.*HiFi__hw_sofsoundwire_4__source`. Do it with no session
+  running — every wireplumber restart cost Ardour its JACK backend today.
+- **Ardour does not follow the default sink (#41).** After any restart it
+  auto-links Master to the first sink (Headphones), not the default (Speaker).
+  A qjackctl disconnect of the headphones leg or an ardour port-restore is
+  still a manual step. Consider a link-reconciler in tidal-ardour-autoroute.