| |||||||||||||
Systemd 197 releasedSystemd 197 releasedPosted Jan 9, 2013 2:03 UTC (Wed) by mgb (guest, #3226)Parent article: Systemd 197 released
I've used dozens of operating systems since starting with George 3 and I really do think Linux is the best platform for most applications.
I strongly disapprove of Mr Poettering's ongoing attempts to morph the supremely flexible Linux world into some Microsoft-ish monolithic Lennux thing.
(Log in to post comments)
Systemd 197 released Posted Jan 9, 2013 2:39 UTC (Wed) by dirtyepic (subscriber, #30178) [Link]
I'm guessing you skipped over the long list of names on the bottom of this announcement.
Systemd 197 released Posted Jan 9, 2013 12:38 UTC (Wed) by Zack (guest, #37335) [Link]
Well, one of those names is Robert Millan's.
Assuming he's the same that is involved with Debian GNU/kFreeBSD, there might be a correction of Lennart's dismissal of all that's not to his liking, and systemd might, after all, become a universal init for GNU systems.
(Although, after a cursory search of the systemd-devel list I can't find any patches)
Systemd 197 released Posted Jan 9, 2013 14:19 UTC (Wed) by HelloWorld (guest, #56129) [Link]
A port of systemd to FreeBSD won't happen, the amount of Linux-specific functionality that systemd uses makes that a daunting task. And even if somebody were to do the work, Poettering made it very clear that he wouldn't maintain the resulting ifdef mess.
The best thing for the kFreeBSD people to do would be to create their own non-portable init replacement and use all the FreeBSD goodies such as kqueue, jails, capsicum, devd and whatnot. This project should share the portable subset of systemd's interfaces and add its own extensions for kFreeBSD specific functionality. I can't think of another way to bring systemd's advantages to kFreeBSD while retaining an easily understandable and hackable code base. Yes, there'd be some duplication of functionality, but you can't have everything.
Besides, the main obstacle for wide-spread systemd adoption isn't lack of portability but Ubuntu. They are the only distro with a relevant user base that doesn't offer systemd at least as an option.
Systemd 197 released Posted Jan 9, 2013 16:23 UTC (Wed) by rgmoore (✭ supporter ✭, #75) [Link] I think it's more fair to say that a FreeBSD port won't happen soon. Once systemd has settled down, there's a better chance that necessary changes to enable it on systems other than Linux will happen. FreeBSD may adopt some Linux-like features that will make porting systemd practical (especially once they've proven their worth in Linux) and Lenaert Poettering will probably move on to a new project, so systemd may have a maintainer who's more willing to tolerate some
Systemd 197 released Posted Jan 10, 2013 10:34 UTC (Thu) by HelloWorld (guest, #56129) [Link]
This is not only about the willingness to tolerate ifdefs. I said earlier that porting systemd to kFreeBSD would be a daunting task, and that was probably an understatement. Poettering had this to say about the matter:
https://lwn.net/Articles/524920/ There are real technical issues going on here, to the point that "porting" systemd to kFreeBSD would mean to essentially rewrite it. At least that is the impression that Poettering's comment left me with, and he should know given that he wrote the thing.
Of course, things would be different if FreeBSD were to adopt the essential Linux APIs that systemd needs. But I don't think they will as most of the functionality is already there, just in different form. Linux has devtmpfs/udev, FreeBSD has devfs/devd. Linux has epoll/inotify/fanotify, FreeBSD has kqueue/kevent.
Systemd 197 released Posted Jan 9, 2013 15:49 UTC (Wed) by mbiebl (subscriber, #41876) [Link]
The list of contributors is (auto)generated by going through the git log.
Robert Millan is mentioned, because nss-myhostname was merged into systemd, including its git history, where Robert had provided a kfreebsd build fix [1]. That non-Linux support was dropped later on [2].
Hope that clarifies.
[1] http://cgit.freedesktop.org/systemd/systemd/commit/?id=1d...
Systemd 197 released Posted Jan 9, 2013 5:10 UTC (Wed) by nteon (subscriber, #53899) [Link]
you're a decent troll, but I suggest being more subtle. In time you might be able to pass yourself off better.
Systemd 197 released Posted Jan 9, 2013 6:00 UTC (Wed) by mgb (guest, #3226) [Link]
Do you have anything to say that's on point and not ad hominem?
Linux and the Gnu tools combine to create an extremely flexible platform that is optimal for almost all use cases.
Poettering seems determined to create a monolithic monstrosity more akin to Microsoft than Gnu and Linux.
Not only is this technically abhorrent in and of itself, but the process whereby he takes control of a key feature and then leverages it to force other unnecessary changes is morally abhorrent, and incidentally also in the Microsoft vein.
Systemd 197 released Posted Jan 9, 2013 6:35 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]
Very unsubtle trolling.
> Linux and the Gnu tools combine to create an extremely flexible platform that is optimal for almost all use cases.
Systemd 197 released Posted Jan 9, 2013 11:01 UTC (Wed) by mgb (guest, #3226) [Link]
It's great that non-monolithic Gnu and Linux make it so easy to replace components.
Systemd 197 released Posted Jan 9, 2013 12:43 UTC (Wed) by HelloWorld (guest, #56129) [Link]
> It's great that non-monolithic Gnu and Linux make it so easy to replace components.
It's great iff there is a substantial benefit to be gained by doing so. If there isn't, the ability to replace components only makes things more complex and harder to integrate and test. Poettering understands that, and most 'Unix philosophy' whiners either don't or have very strange ideas about what a substantial benefit is.
Systemd 197 released Posted Jan 9, 2013 6:51 UTC (Wed) by HelloWorld (guest, #56129) [Link]
systemd is not monolithic. Many of its features run in processes other than PID 1 and most of its advanced functionality can be disabled with compile-time configuration options and/or at run-time.
And besides, I don't even care whether it's monolithic or unix-ish or whatever other labels you'll come up with. What I care about is that it is faster, more reliable and more powerful than any of its alternatives. And unlike you, Poettering posted long and detailed rationale for the systemd design in his blog, while you completely fail to give any technical arguments against it. That makes it a lot easier to agree with him than with you.
Systemd 197 released Posted Jan 9, 2013 8:35 UTC (Wed) by tnoo (subscriber, #20427) [Link]
No idea what you are talking about. Personally, I perceive your "monolithic" as "unified". What this has to do with Microsoft escapes me, and in any case you are free to use any boot and logging system you like (several distributions support alternative boot systems).
For me, systemctl is much nicer to use than random init scripts, and journalctl much more convenient than the traditional Unix chaos of randomly formatted logfiles.
So I highly welcome Poetterings work on this, which is much more useful than your comment that Linux is morphed by him into Windows (whatever that means).
Systemd 197 released Posted Jan 9, 2013 9:41 UTC (Wed) by tao (subscriber, #17563) [Link]
So, let me get this straight: systemd adds functionality that is impossible to achieve using sysvinit, still supports using old style init-files *and* is in no way mandatory to use anyway (if you still want to use sysvinit, who's stopping you?!). And that somehow is a threat to the supreme flexibility of Linux? Uhmmmm....
Systemd 197 released Posted Jan 9, 2013 9:49 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]
Well, systemd might be too good so everybody would switch to it. And that's definitely non-Unixy! There must be 100500 different incompatible forks and variations of _everything_!
|
|||||||||||||