Repository navigation
Skip the shell restart when no Hyprland session is running - #17295
c4software wants to merge 1 commit into
Conversation
[Medium impact] The PR appears safe to merge. SummarySkips the shell restart when the current user has no
No actionable issues found. Tests were inspected, not run. Reviews (1) · Last reviewed commit: "Skip the shell restart when no Hyprland ..." · Reviewed by Greptile |
|
Compared the update/restart callers and full diffs with #13945 and #16599. Both older fixes check Hyprland inside Opus 5.5 High and muse-wonder via pi, thinking high agreed that this caller-only fix should wait behind the older helper fixes, but disagreed on which older PR to use; their independence is not guaranteed. No worker tests ran because this stopped at the competing-fix comparison. This stays open for the maintainer’s choice alongside #16599. |
omarchy updateon a machine with no graphical session sits silently for 60 seconds at its shell restart step, right after printing "All plugins have been reloaded", then reports that the shell did not become ready. Nothing is reloaded and nothing can start: it looks hung, and Ctrl-C there aborts the update. On quattro this step runs mid-update, so that would also skip hooks, mise, AUR packages and boot verification.omarchy-restart-shellspawns the shell throughhyprctl dispatch. With no compositor that call fails at once, but its status is discarded and the 60-second readiness loop runs anyway.The update's service phase now only restarts the shell when the user has a Hyprland process, and says why it skipped otherwise. Other marked services still restart. Checking the process rather than
hyprctlkeeps the restart working over ssh into a machine with a live desktop, wherehyprctlhas no instance signature untilomarchy-restart-shellderives one.Tested on a headless machine (multi-user.target, no compositor): the service phase finishes immediately instead of after 60 seconds. On a desktop session,
Hyprlandis still detected and the restart path is unchanged.test/shell.d/update-restart-phases-test.shgains a headless case, which fails without the fix.Related: #16599 makes
omarchy-restart-shellitself fail fast when Hyprland is unreachable, which would help its direct callers, but the update would still print the reload message first. #16818 rewrites the same lines' messages, so whichever lands second needs a small rebase.