smoke: the inhibitor readback was noctalia's contract, not kuma's

The re-armed sleep-inhibitor check went red on a healthy session —
and the user's own desktop proved it was not the VM: kuma-shell holds
no logind delay inhibitor on real hardware either. That is the shipped
design, recorded in the sleep guard's own prose: a hung shell cannot
stall sleep the way the hung noctalia could, lock-before-suspend is
niri's and best-effort, and the guard is the property's readback. The
check asserted noctalia's contract and is retired as the swap's last
fossil; the notifications probe stays, CI-verified green on the new
pin.

Author: Letdown2491

AGENTS.md

@@ -95,10 +95,14 @@ nostr layer is enabled) — the same road every release ships.
   restart-looping until the unit's start limit ended the waves: a
   probe needing a live shell process was a coin flip wearing a
   deadline. kumaui's idle-without-windows fix (Oct 2026, e90884c)
-  keeps the shell alive holding its names and inhibitor through
-  output loss, and the notification and inhibitor probes are armed
-  again. The lesson stands: a probe that needs a live process asks
-  the runner's timing unless the process is guaranteed alive.
+  keeps the shell alive holding its DBus names through output loss,
+  and the notification probe is armed again. The sleep-inhibitor
+  check stayed retired: kuma-shell holds no logind delay inhibitor
+  by design (a hung shell cannot stall sleep), so that probe
+  asserted noctalia's contract, not this desktop's. Two lessons: a
+  probe that needs a live process asks the runner's timing unless
+  the process is guaranteed alive; and a fix's report that claims
+  more than its diff gets caught by the gate that runs the diff.
 - `egl-headless` wants a host DRM node a runner lacks. The GPU-less
   console combo is `virtio-vga` plus `none`.
 

notes/44.4.0-plan.md

@@ -432,8 +432,12 @@ binaries could not carry it. What the release week found:
   readback, a stale noctalia pgrep — either measured the runner's
   timing or passed vacuously. What remained asked the wiring and the
   journal; whether the shell draws is real hardware's question. The
-  measured facts are in AGENTS.md. Re-armed 2026-10-03: kumaui's
-  idle-without-windows fix (e90884c) keeps the shell alive holding its
-  names and inhibitor through output loss, so the ownership and
-  inhibitor probes came back, and the kumaui pin was bumped to carry
-  the fix.
+  measured facts are in AGENTS.md. Partially re-armed 2026-10-03:
+  kumaui's idle-without-windows fix (e90884c) keeps the shell alive
+  holding its DBus names through output loss, so the notification-
+  ownership probe came back and passed CI the same day. The sleep-
+  inhibitor readback stayed retired, and for a different reason than
+  the wave era: the shipped sleep guard's own prose says kuma-shell
+  holds no logind delay inhibitor by design — a hung shell cannot
+  stall sleep the way the hung noctalia could — so that check asserted
+  noctalia's contract and was the last fossil of the swap.

scripts/smoke.sh

@@ -1449,14 +1449,14 @@ smoke_published() {
             done
             ok "the shell owns org.freedesktop.Notifications"
 
-            # Lock before suspend: `lock_before_suspend = true` is baked,
-            # and the readback is the shell's logind sleep inhibitor.
-            # The same kumaui fix keeps it held through output loss; the
-            # guard below still proves the no-shell half of the property.
-            guest systemd-inhibit --list --no-pager \
-                | awk '$1 == "kuma-shell" && $6 ~ /sleep/ { found = 1 } END { exit !found }' \
-                || bad "the shell holds no sleep inhibitor: this machine suspends without locking"
-            ok "the shell inhibits sleep to lock first"
+            # Lock before suspend: kuma-shell deliberately holds no logind
+            # delay inhibitor — the shipped sleep guard's own prose says
+            # so, so a hung shell cannot stall sleep the way the hung
+            # noctalia could — and lock-before-suspend is niri's
+            # `lock_before_suspend`, best-effort, same as any locker.
+            # A probe here asserts noctalia's contract, not this
+            # desktop's. The guard below is the property's readback: a
+            # session with no shell ends instead of suspending unlocked.
 
             # And the guard for when the shell is not there at all.
             #