Le 07.07.2014 23:58, Brian a écrit :
On Mon 07 Jul 2014 at 16:14:25 -0400, Steve Litt wrote:On Mon, 07 Jul 2014 16:03:31 +0200 berenger.morel@neutralite.org wrote:> I forgot that systemd is able to "spy" processes it starts, IIRC, to> avoid tricks like the double fork, which means a better control on > zombie processes. Don't know much about that, though.Dam, dude, doublefork is how my Umenu hierarchical menu works. If thatbreaks, a program I've used since 1998 goes down the toilet.Let me ask you this: If doublefork is banned, how does one keep workingwith the child program when the parent is terminated, without using that stupid nohup command that continually grows its own, huge andinsecure nohup.out files, in whatever its current directory happens tobe? I certainly hope that systemd isn't cancelling a design pattern used since the dawn of time.No banning, no cancelling, just the use of cgroups to keep track of processes. No matter how many times a process forks it can never be orphaned. http://0pointer.de/blog/projects/systemd-for-admins-2.html
Thanks for the clarification, I never made use of this (anti? Seriously, double forking is just fighting the process manager or whatever the name is, right?)pattern in code myself and have not read a lot about it. And if I had to do some daemon softwares, I would try hard to find a portable library (I try as best as I can to write things that can be compiled on all OSes I know and am able to use, so, linux and windows, with hope that they are compatible with BSDs).
Anyway, I think that systemd does not break a lot of things, it has tried to be compatible, so I doubt the compatibility problems are a real argument against it. But it won't avoid me thinking it's not adapted to my philosophy and how I manage my computers. No need to flame about it.