Lennart Poettering wrote:
Generally on Linux only the X session in the fg will get the keypresses. If you switch away with from it, the new one in the bg will get it instead. logind will get active hence only if there's nobody on that specific VT who wants to take the events.
The understanding, who processes keypresses, depends on the point of view. My daemon just connects to an input device, so it receives the keypresses *always* and the active VT doesn't matter.
Well, note that these specific inhibitors are about inhibiting key presses, not actions. i.e. the inhibitor for logind's suspend key (i.e. the handling of the key) is independent of the inhibiting for the suspend (i.e. the action code can ask for).
Inhibiting "shutdown" would go a bit too far. I don't want to prevent shutdowns that have been triggered by, for example, clicking a button in a GUI (KDE or something). And I think there is no way to create a selective inhibitor that only blocks a shutdown triggered via the power key, right?
Is there a way to enumerate the sessions, that could receive the power key, and register an inhibitor for all of them? Maybe even with an event that tells me if new sessions have been created?
The idea behind this was to make it easier for an user to use alternative power key handlers. I hoped to be able to tell systemd to completely keep away from power button events while runtime without having to tell the user that he has to get sure that his init system ignores the power button, first (and then reboot his system, so logind.conf is interpreted, again).
Yours Manuel Reimer _______________________________________________ systemd-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/systemd-devel
