Krzysztof kuhy Rudnicki fcd3f7ed2f perf(idle-off): replace controller-watch fork storm with a single systemd-inhibit ... turn_off_auto_idle_screen_shutdown.sh --watch-controller forked 4 xset + 1 xdotool per joystick event (blocking on a dd read, debounced by sleep 0.3) — ~21 procs/s while a controller was connected during gaming. The xset poking was redundant: startup already disables DPMS/blanking. Replace reset_idle_activity / watch_js_device / the polling watcher with one long-lived `systemd-inhibit --what=idle:sleep` held only while a /dev/input/js* device is present, re-evaluated on udev input add/remove events (event-driven; 30s presence-poll fallback). EXIT+INT/TERM traps release the inhibitor on every termination path. Verified live: subtree stays {systemd-inhibit, udevadm} with zero dd/xset/xdotool; exactly one inhibitor held; clean release on SIGTERM, no orphans. Takes that loop from ~21 forks/s to 0. Behavior change: keeps the session awake while a controller is connected (not only during active input). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>		2026-06-04 18:14:07 +02:00
..
cleanup-2026-05	chore: spring-clean repo root (move docs, relocate batch3 script, drop stale outputs)	2026-05-14 20:01:09 +02:00
superpowers	perf(idle-off): replace controller-watch fork storm with a single systemd-inhibit	2026-06-04 18:14:07 +02:00