Lennart Poettering
Lennart Poettering
Hmm, one more thing: I think it's correct to just store one of the activation reasons instead of all. But I think we should enforce a clear regime on which...
And I think it should be the oldest event we know about for this activation that we keep (simply because for socket/path units and similar once we see one event...
> > > > A service triggered by a path unit that watches multiple files, they want to know which one was the trigger but then you cannot distinguish the...
> A service triggered by a path unit that watches multiple files, they want to know which one was the trigger But why? Why can't it check the files to...
So I'd rreally make the ActivationEventInfo structure simpler, i.e. no vtable, just add the fields you need directly, even if some of them won't be used in all unit types....
(one of those days, .automount units should acquire this too, and expose this over the bus. i.e. info about process that triggered a mount, and similar) (and socket units, too,...
(and of course activation through bus calls, and of course activation through initial transaction)
thanks lgtm
/cc @msekletar
(i am reviewing just the code btw, i have not enough clue about selinux to comment anything about the ultimate merits of this work. we need someone with a better...