With initial workspace tracking disabled, windows naturally open on the active workspace. Remove the explicit Hyprland workspace dispatch and let shell actions, shell restarts, and presentation terminals launch directly.
omarchy-restart-shell now kills by config dir via quickshell kill and
relaunches with --no-duplicate, polling kill until every instance is
gone since kill returns without waiting and -n silently exits if one
remains. The locked-session refusal reads Hyprland monitor state alone,
which covers a hung shell holding the lock; the redundant shell IPC
probe, availability guards, and omarchy-shell path indirection are gone.
Shell IPC calls now time out (2s default, OMARCHY_SHELL_IPC_TIMEOUT to
override) so probing an unresponsive shell fails fast instead of
hanging the caller.
omarchy-hyprland-launch checks dispatch output for "ok" rather than the
exit code, which is 0 even on Lua errors, so a failed dispatch falls
through to the bash -lc fallback instead of silently launching nothing.
The Quattro upgrade cutover delegates to omarchy-restart-shell instead
of carrying its own copy of the lock check, pkill, and readiness loop.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The /proc/*/environ scraping, wayland/X11 socket probing, and LANG/DBUS
guessing arrived with the Omarchy 4 live upgrade cutover, whose upgrade
script no longer exists. Every remaining caller runs inside the
graphical session with full env. Keep a five-line guard deriving
HYPRLAND_INSTANCE_SIGNATURE from the newest instance runtime dir so
restarting over ssh still works.
Launching now goes straight to the Lua dispatcher: on quattro, classic
hyprctl dispatch arguments are evaluated as Lua and fail with exit 0,
so the old exec fallback silently did nothing. Check for "ok" output
instead of trusting the exit code, with setsid as the real fallback.
180 lines to 94.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
quickshell kill currently trips Quickshell #539 (notifyInternal2 event
loop crash via IpcKillCommand::exec), which leaves the Wayland surface
lingering long enough that the new instance starts on top of the dying
one — a duplicate bar. SIGKILL drops the client connection and the
compositor cleans up surfaces immediately; the shell has no persistent
state to flush. Sub-20ms restart, no duplicates.
The bin names should reflect the thing they manage. `refresh bar` is
misleading because the command resets the entire shell.json (which holds
plugins beyond just the bar), and `restart quickshell` names the runtime
rather than the product. Both now end in `-shell` to match the rest of
omarchy-shell.
Updates every caller in autostart, the menu, voxtype helpers, and the
shell README. No behavior changes.
David Heinemeier Hansson