On Tue, Oct 29, 2013 at 08:37:02AM +0000, Thorsten Glaser wrote: > Olav Vitters <olav <at> vitters.nl> writes: > > > Most of systemd is not in pid1. This was explained by a blog references > > But (by the time of the jessie freeze, at least) it will need systemd to > be pid1 to work. Same thing, really, just picking words.
No, not at all. If you say that everything is in PID1, this implicates that there is a big potential for bugs. Bugs in PID1 is bad, you don't want an unreliable init system. Aside from that, having so much code in PID1 means the memory footprint is pretty big, likely security issues to be big as well. If someone suggests that whole of systemd is in PID1, I assume that *that* is the problem. Not anything else. What is meant instead is that systemd provides loads of "building blocks". This is in pretty much any systemd talk that Lennart has given (except initially). That this is meant is not obvious to me, nor do I understand it. The PID 1 argument has lead to explanations such as: http://people.debian.org/~stapelberg/docs/systemd-dependencies.html A whole list just to explain that things are *not* in PID 1. This was based on feedback regarding everything being in PID1. Everything in PID1 I can discuss, because it is not true. But the "building blocks" is something entirely different. -- Regards, Olav -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20131029085523.ga18...@bkor.dhs.org
