fix: Keep Waybar alive when switching themes from a bar module (#96)

Switching Dark/Light from the Waybar module left the bar gone until the
next login.

DarkLight.sh is wired to a Waybar module on-click, so it runs inside
waybar.service's cgroup. Its cleanup loop kills waybar, systemd tears down
the whole unit, and every process in that cgroup goes with it - DarkLight.sh
included, before it ever reaches Refresh.sh. Nothing is left to start the
replacement bar. Refresh.sh had the same problem on its own: it killed
waybar and relaunched it from a shell that had just been torn down.

Drop waybar from the DarkLight.sh cleanup loop and let Refresh.sh own the
restart, run as a transient systemd-run --user unit so it sits outside the
caller's cgroup and survives the teardown. Fall back to setsid where
systemd-run is unavailable.

Detect systemd management via the unit's LoadState instead of is-enabled.
WaybarStartup.sh starts the unit on demand and leaves it disabled, so
is-enabled reports "disabled" for setups that are in fact systemd-managed.

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>

1 files changed, 5 insertions, 1 deletions

diff --git a/config/hypr/scripts/DarkLight.sh b/config/hypr/scripts/DarkLight.sh
index 913099fc..5bad2af8 100755
--- a/config/hypr/scripts/DarkLight.sh
+++ b/config/hypr/scripts/DarkLight.sh
@@ -354,7 +354,11 @@ ${SCRIPTSDIR}/WallustSwww.sh
 if [ "$no_restart" -eq 0 ]; then
     sleep 2
     # kill process
-    for pid1 in waybar rofi swaync ags swaybg; do
+    # NOTE: waybar is deliberately excluded here. This script is usually launched from a
+    # waybar module on-click, so it lives in waybar.service's cgroup. Killing waybar makes
+    # systemd tear down the whole unit, taking this script with it before Refresh.sh runs.
+    # Refresh.sh restarts waybar detached from that cgroup instead.
+    for pid1 in rofi swaync ags swaybg; do
         killall "$pid1"
     done
     sleep 1