Switching update channel from the menu installs a new kernel but never offers to reboot
Checked against 4.0.4-1 and quattro. /usr/bin/omarchy-channel-set line 99 is omarchy-update -y. /usr/bin/omarchy-update sets OMARCHY_UPDATE_UNATTENDED=1 for -y. On 4.0.4-1, confirm_reboot in /usr/bin/omarchy-update-restart is a bare gum confirm that does not check the variable. On quattro it prints $1 Run omarchy-system-reboot when ready. when the variable is 1, which matches the fallback lines in #13416 (reporter on omarchy-dev edge). omarchy-channel-set always uses the newly installed package's omarchy-update (it runs pacman -S first), so a switch to edge already gets the suppressing version, which is consistent with the record. omarchy-state set reboot-required is at lines 83 (into dev) and 95 (leaving dev) on 4.0.4-1. omarchy-state writes to ~/.local/state/omarchy, and omarchy-system-reboot runs omarchy-state clear re*-required, so the verify step's 'marker gone' holds after rebooting that way. The kernel check matches upstream's logic, and pgrep -x Hyprland matches the running process here. omarchy-channel-current exists and prints the channel. #8986 is the upstream issue that introduced the unattended check, so it is relevant context.
Symptom
Switching channel from the Omarchy menu (Update > Channel, or omarchy-channel-set edge / stable / rc) finishes without the usual "Reboot now?" choice, even though the switch replaced the kernel or Hyprland. The last lines are plain text instead of a prompt:
Linux kernel has been updated. Reboot? Run omarchy-system-reboot when ready.
Hyprland has been updated. Reboot? Run omarchy-system-reboot when ready.
The user closes the terminal, keeps working on a kernel whose modules are gone, and things like USB devices, VPN modules or a new monitor stop working until the next reboot.
Cause
omarchy-channel-set ends by handing off to omarchy-update -y (line 99 of /usr/bin/omarchy-channel-set on 4.0.4-1). -y sets OMARCHY_UPDATE_UNATTENDED=1. On builds from the quattro branch (the edge and dev channels at the time of upstream #13416), confirm_reboot() in omarchy-update-restart checks exactly that variable and prints the fallback text instead of calling gum confirm. The switch is launched into a real floating terminal with a person watching, so the variable suppresses the one question that mattered.
On the stable 4.0.4-1 package confirm_reboot() does not check the variable, so the reboot prompt still appears after a channel switch there. The fault therefore shows when switching to or between edge builds, or once a stable release picks up the quattro change.
Fix
After any channel switch, check by hand whether a reboot is needed.
uname -r # the running kernel
ls /usr/lib/modules/ # the installed kernels
ls ~/.local/state/omarchy/reboot-required 2>/dev/null && echo 'reboot marker set'
readlink /proc/$(pgrep -x Hyprland)/exe # ends in (deleted) if Hyprland was replaced
If the running kernel has no directory under /usr/lib/modules/, the marker exists, or Hyprland shows (deleted), save your work and reboot:
omarchy-system-reboot
Switching to or from dev always sets the marker (omarchy-state set reboot-required in omarchy-channel-set), so always reboot after those two.
Verify
After the reboot, uname -r matches a directory in /usr/lib/modules/, ~/.local/state/omarchy/reboot-required is gone, and omarchy-channel-current reports the channel you chose.