good test. i was curious on this. i tryed kfreebsd then, but it was very broken for me. but expected more difference for linux.
I don't care for kfreebsd, but as long as it gives a reason to stick with sysvinit, I'm all for it.
Originally Posted by kraftman
init should be init, that's it.
I don't wan't chroot upgrades failing, I don't want to recompile init if early stages need fixing, I don't need snapshotting, and DISK TOOLS DON"T BELONG IN INIT!
I can get 32 MB used with sysvinit and a ton of services, on debian; and chroot stuff works without any magic. I can't tell how a comparable systemd setup would be, but it beats Upstart, and systemd seems to have more of that garbage.
By the way, kfreebsd is not "experimental". It's officially supported.
sysvinit is damn slow and terribly simple crap that should die long ago. It does nothing, but starts services and doesn't give you any power to control them. If you want to stick in 80's you're free to do so. That's the problem kfreebsd is officially supported in Debian and I hope it will be dropped soon. It's the same joke as sysvinit.
Originally Posted by Ibidem
Interesting conversation that was going on here. I for one obviously only saw it now.
Personally, I like kFreeBSD, it opens up a few choices for the Linux world. At the same time I wasn't aware that it was keeping Debian from going with systemd ect, but I must say, after playing with Debian/kFreeBSD for a while, I perrsonally am falling in love and will probably switch a few of my pcs at home over.
To add to that, I like seeing these benchmarks, they are informative to me at least, who doesn't have time to do them myself.
There is a bad interaction between the FreeBSD kernel and GCC 4.4 and later. It forced the Debian developers to compile the FreeBSD kernel with -O1, which gives Debian GNU/kFreeBSD a handicap.
We had the same issue in Gentoo FreeBSD. We partially solved it by adopting Debian's patch and modifying it to support -O2 with Clang. Here is the Gentoo bug: