On Donnerstag, 26. November 2015 00:37:35 CEST, Andreas Hartmetz wrote:=
=0A=0A> First, by copy I mean a temporary copy that is never merged bac=
k into =0A> the main list, it's kept around only to know which processe=
s have =0A> already agreed to have their session saved and submitted th=
e =0A> corresponding data.=0A> ....=0A> The current problem is that som=
e applications die between submitting =0A> their state to ksmserver (Sa=
veYourselfDone message) and ksmserver saving =0A> the list of processes=
 to restart (writing ksmserverrc).=0A=0AOk, that's not a problem.=0A=0A=
> It is AFAICS =0A> safe to assume that an application that has submitt=
ed its session data =0A> is really supposed to be restored.=0A=0AAgreed=
, and if this is only meant as temporary condition we won't restart a s=
ession history either. Given the async nature of IPC the cover-up sound=
s reasonable enough, yes.=0A=0ACheers,=0AThomas