Apple's OS X Launchd Being Ported To FreeBSD

[JX8p]

So, sysv init is now a bad guy for bsd?

FreeBSD never used sysvinit. sysvinit is a clone of the init system used in UNIX System V. FreeBSD inherited bsdinit from its precusors (BSD 4.3 and 4.4-Lite).

With regards to running launchd on other platforms-- the port is currently focusing on isolating all parts of launchd that rely on OS X's Mach microkernel API. With those no longer a prerequisite a port to other BSDs and Linux would be fairly simple.

[garegin]

Good since FreeBSDs current init system sucks monkey balls. It's the most slow thing I have ever used in my life. You totally forget that you have a SATA 3 SSD installed. My Arch Linux installation is able to boot faster than FreeBSDs init is able to set the hostname. I'm not kidding.

It's systemd btw not Sytemd or SystemD or systemD.

I think this plays into the aura of BS that surrounds the FreeBSD community. They constantly claim that their OS is "faster" than Linux, while the benchmarks claim otherwise. Even without systemd, sysv easily beats the pants off bsdinit. A recent interview with Marshall Mckusick shows just ignorant their leaders are. He claimed that "Windows teaches you to tolerate BSODs three time a week, so one time a week is welcome news". This is absolute horses***. I've been using Windows in home and office for a decade, on many machines, and I barely get one BSOD a year. The last time I got a BSOD that was not hardware related was three years ago for a GPU driver update.

[JX8p]

I think this plays into the aura of BS that surrounds the FreeBSD community. They constantly claim that their OS is "faster" than Linux, while the benchmarks claim otherwise. Even without systemd, sysv easily beats the pants off bsdinit. A recent interview with Marshall Mckusick shows just ignorant their leaders are. He claimed that "Windows teaches you to tolerate BSODs three time a week, so one time a week is welcome news". This is absolute horses***. I've been using Windows in home and office for a decade, on many machines, and I barely get one BSOD a year. The last time I got a BSOD that was not hardware related was three years ago for a GPU driver update.

Is sysvinit really faster than bsdinit? I'm pretty sure all init systems I've seen finish in about 2 seconds flat (on a 5400RPM hard drive no less). As for the interview, well, that kind of claim is something I expect from most FOSS prononents, whether Linux, BSD, or something else.

[brosis]

Is sysvinit really faster than bsdinit? I'm pretty sure all init systems I've seen finish in about 2 seconds flat (on a 5400RPM hard drive no less). As for the interview, well, that kind of claim is something I expect from most FOSS prononents, whether Linux, BSD, or something else.

It allows concurrent boot, so yes, its faster. Its also not much different from BSD init, one got links, other text entries. The only case where BSD init might win is slow file open operations.

[garegin]

Is sysvinit really faster than bsdinit? I'm pretty sure all init systems I've seen finish in about 2 seconds flat (on a 5400RPM hard drive no less). As for the interview, well, that kind of claim is something I expect from most FOSS prononents, whether Linux, BSD, or something else.

my hunch is that the slowness has to do with poor hardware support/optimization. on some computers Linux boots unnaturally slow too. Usually those are funky firmware handoff bugs. The slowness I encounter is usually in the early stages of boot which confirms my theory. I've had a vanilla install of pfsense take twice longer to boot than a full install of Fedora with Gnome.
At the end of the day, money talks and BS walks. If FreeBSD was superior to Linux, it's server marketshare wouldn't have been so terrible. And you can't give the "people just haven't heard of it" argument, because every Linux admin I have met has heard of FreeBSD and even played with it. Back in the early 90s FreeBSD actually had a lead over Linux, because the later was extremely immature (early Debian, Slackware), while the former was tried and tested for more than a decade (FreeBSD was just BSD without the AT&T bits)

[brosis]

It was about control, license restrictions and design goals.

The world is better off with LLVM/Clang in it. It's what has gotten GCC to evolve.

Yeah, pretty much. BSD-license has no control, is legally incompetent license (hence most switched to Apache), no restrictions (do what you want, close all you want), BSD init had the only design goal of being KISS for the sake of KISS which leads to poorly integrated system, lack of advanced functionality.

On the other side the only misarchitecture of GNU was monolithic design because they feared emerging of proprietary plugins (which do happen) but on the other side it lead to piles of old hard to manage code. Still, LLVM is exact opposite, there are proprietary extensions available. I guess GNU couldn't care less. This is exact same thing with Gstreamer, with the topmost contributor being codec-selling/licensing/pro-patenting company. This is why you prefer VLC.

Otherwise, GNU has powered BSD since forever and doesn't care what proprietary company those guys are powering next (like Blizzard, Netflix). GNU is also a political movement, that is protecting free software, where BSD is politically an anarchy being a tool for everything.

That said, I hardly imagine why a tool needs launchd. Maybe because they have low resources, and Apple is kind enough to allow them that. Apple still holds all the rights, but they already use LLVM, CUPS, they prefer Apple to GNU, they couldn't care more.

So, where is this shitornado about how launchd is incompatible with legacy and how non-KISS it is; like what pro-BSD trolls created about systemd?

[garegin]

It allows concurrent boot, so yes, its faster. Its also not much different from BSD init, one got links, other text entries. The only case where BSD init might win is slow file open operations.

sysvinit has concurrent boot? since when?

[Ibidem]

sysvinit has concurrent boot? since when?

Since years ago.
It's called startpar.

@JX8p: From "loading init" to login: in 2 seconds? I doubt that.
When people talk about how fast an init system is, it's about how fast it starts all daemons, not just how fast it starts the daemons it comes with.

@brosis:
You clearly haven't searched the BSD forums...I've read a bit of criticism of launchd there, all on the basis of not following unix design principles...

[JX8p]

@JX8p: From "loading init" to login: in 2 seconds? I doubt that.
When people talk about how fast an init system is, it's about how fast it starts all daemons, not just how fast it starts the daemons it comes with.

In both cases the same daemons were being started, near enough. Given that enabling /all/ daemons is a bad idea it's not really an issue of concern.

[enjolras]

On the other side the only misarchitecture of GNU was monolithic design because they feared emerging of proprietary plugins (which do happen) but on the other side it lead to piles of old hard to manage code. Still, LLVM is exact opposite, there are proprietary extensions available. I guess GNU couldn't care less. This is exact same thing with Gstreamer, with the topmost contributor being codec-selling/licensing/pro-patenting company. This is why you prefer VLC.

Are you serious ? Are you such a fanboy you're ready to accept bad design choices in a software just to avoid other reusing your work ? It's crazy. First, the license is supposed to achieve this goal, and second, just because you don't want other to reuse your work as they want (which harmless in my opinion, but ok, let's admit some people don't like that), you'll make your code worse, reducing the quality of the software for people who use it like you expected, and you make it harder for you to maintain ?

It's amazing you can just suggest such a thing...