On Sun, 28 Sep 2014 18:10:52 -0500 green <greenfreedo...@gmail.com> wrote:
> Jonathan Dowland wrote at 2014-09-28 13:05 -0500: > > The more and more I read people objecting to the modularity of > > systemd, the more I am reminded of the Tanenbaum/Torvalds debate re > > microkernels. > > It does seem to be related, though not so much due to systemd being > modular as monolithic. systemd introduces a large amount of "core > code" at PID 1 that needs to be stable and reliable. > > Microkernels, as I understand, aim to support a highly modular system > *design* but are themselves minimal (Minix 3 has about 4000 lines of > executable kernel code). This "core code" can be more easily audited > and maintained. Servers, eg. device drivers, are supervised and can > not bring down the system (in the context of the kernel). (See > <http://www.minix3.org/other/reliability.html>.) > > So yes, perhaps one major reason some people dislike systemd (too much > "core code") is the same reason some people like the microkernel > design. I don't think anyone had a problem with the concepts around microkernel. The problem is that nobody in the Free Software community could write a decent microkernel. GNU had been trying forever. So the argument for the Linux kernel was: It's here now, it works well, and if we had a microkernel that worked as well, we'd use that. Of course, if all the pieces of the microkernel, or the Linux kernel (at that time) added up to 500K lines, and required huge changes for applications that ran on it, and broke everything that came before, people would have taken to the streets. SteveT -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140928230445.74a2d...@mydesq2.domain.cxm
