I think it's too early to tell since it hasn't really propagated to the "server" distributions yet. The real test for systemd will probably come with RHEL 7 and/or the next Debian release (if it gets integrated).

The big problem I see is that systemd is trying to be the black monolith of init systems. Gnome has (or had) a dependency on it, so even if Slackware stays away from it, there's a risk that some other upstream projects might not build without it. systemd also has some weird dependencies of its own, such as including a web server: http://thread.gmane.org/gmane.linux....a.devel/169082
No, that's because Ubuntu has their own init system called upstart, which seems incapable of shutting down my computer half the time.

1 members found this post helpful.

Both RHEL (7) and SUSE Linux Enterprise (12) plan to adopt it so at least there's two competing companies who think it's worth it.

It looks likely that both Gnome and KDE will use systemd --user for starting and managing the user session in the future. Gnome's Mutter, KDE's KWin and Wayland's Weston compositors will require systemd-logind for handling the device access among other things with Wayland. Gnome currently uses some systemd APIs (systemd-{hostnamed,timedated,localed...}) to configure the system, these can be implemented without systemd though. Gnome aims to keep the dependency on essential features optional so the starting Gnome session without systemd will likely be supported for long time, same goes for KDE.


systemd doesn't depend on a web server. It's a optional and separate daemon for the systemd-journald logger that can be disabled on both compile and runtime.

It's fine. I don't have to adopt them. They use RPM, that alone is reason enough to stay away.

Fine, they can build whatever 0pointer-based desktop operating system they want. Nobody has to and almost nobody will use it, just as before.

The whole discussion is completely irrelevant to Slackware and off-topic here and there's nothing changed since the last one.

Why I'm even answering someone with a fake account and a single post?

Do you have documentation about the process of installing systemd on Slackware? I am generally interested in init systems and would like to give that a try.

On the original topic: I don't like some things in systemd either, but that is no reason to start an inflammatory thread, especially with the obvious FUD that comes up all the time in similar threads. I don't know what you mean with obscure, but systemd is not badly documented and since at least two enterprise distros will use it in their next release I highly doubt that it is as unstable as you claim.

Well, Lennart isn't getting any love here due to how fickle he is with projects and how he feels he has to stick his nose in everything, stir the pot, then walk away.

His track record more than speaks volumes about how many people feel about him.

If and when Slackware will ever get systemd is a mystery known only but to Patrick, and I'm fairly certain hes been testing it and keeping a close eye on it's progress, but hasn't committed to it for one reason or another, probably due to his target audience being minimalists and power users as well as system administrators who like the Keep it Stupidly Simple approach to Slackware and it's BSD scripting.

LFS won't be getting it either. They tried to import it with 7.3, but it really made a mess. A book to built it exists but you have to build and rebuild so much against it it's ridiculous.

SOURCE?! X.Org has said nothing about Wayland and Weston both requiring systemd. FUD! It's only required AFAIK if you launch Weston as a user without a dedicated weston-group.

Systemd still has a few issues with parallel loading with some services still not being handled properly. OpenRC and RunIt both had similar issues with parallel loading and made it entirely optional. This is why the linear service loading design is still highly favored by system administrators of mission critical servers and devices. Less, headaches, and less worry.

GNOME already poisoned themselves, hence why DropLine and MATE exist. If KDE wants to poison themselves also, then let them. Most stuff still only requires the DBus anyway, so why systemd would be absolutely required is just asinine. We can all get along with alternative DEs like Xfce, MATE, Trinity, LXDE, and others.

Only a small fraction of systemd is not well documented, the rest has fairly decent documentation but logind and ConsoleKit, which have the worst documentation to date.

And this comment is exactly the reason why we should and need to have a discussion about it now.
From Pat's comment, I conclude that Pat has already made up his mind. And once we have reached the point where Pat is making the switch, it will be too late to have a discussion.

The first thing that annoys me is that systemd is using MS Windows terms, i.e. "services". On *nix we have daemons, not services.

I am not in a position to make more informed comments, cause I have not used it or read about it.

3 members found this post helpful.

What leads you to the conclusion, that a Red Hat style OS has anything to do with Unix?

It's the new "Linux standard base". If you want Unix, you can move to OS X. Solaris may qualify as Unix, too. Or the BSDs. That's why they are "a thing of the past", ya know?

Well, "Red Hat style" covers a lot of ground, but Linux does in fact have a rich and valuable Unix heritage - it did not develop out of a vacuum! It is "Unix-like" by definition and by design, and for many the core value proposition of Linux is its Unix foundations.

We ignore or discard that Unix heritage at great loss.

1 members found this post helpful.

gee - Thanks very much for making me feel old.

But - here's something that I've been wanting to say for quite some time now:

When reading a developer's project page, I usually understand "the itch they're trying to scratch". But with Wayland and systemd I am completely at a loss what it is they're trying to achieve.

For example from Wayland's wikipedia page:
On my system with X Windows, my windows are not being torn apart, are not lagging behind, etc. "that we'll never see redrawing" - I think it's in the nature of application windows that they're being redrawn, because their contents change. To be honest - I am not even sure what they mean with these terms in the context of a display manager. So - yes - maybe I am getting too old for this.

On the other hand - there are glaring bugs in programs that we use daily, which are not being addressed.

https://www.libreoffice.org/bugzilla...g.cgi?id=67787

Inventing exciting new stuff is always more popular than fixing bugs.

BTW: I'm not using LibreOffice, I stick with OpenOffice and an age-old StarOffice 5.2 (I run the Windows version of it, because the Linux version doesn't work anymore due to glibc breakage).

BSD, Solaris, Illumos, and HP-UX are not really systems of the past. Just becasue they aren't as mainstream in the UNIX-world as Linux is, doesn't make them any less relevent.

Just becasue they don't support the same software and hardware as Linux does, doesn't make them any less important. When you get into servers and networking you see systems like FreeBSD taking over and where high capacity Datacenters are concerned you'll start seeing Solaris, HP-UX, and such.

Yes, using the Windows terminologies is a poor excuse of a way to describe Linux daemons, but systemd is effectively duplicating on main service found in Windows by trying to be the main all-purpose daemon. Systemd, is trying to effectively duplicate svchost.exe, which has ballooned the requirements of Windows since Vista to staggering levels by trying to take in more and more responsibility.

And since when do any distributions truly conform to the LSB compliancy? The Linux Standard Base, and all are a loosely knit collaborations in designs of what a Linux system should be, but isn't required to be. No Linux distribution has to conform to any set level of standards other than providing the basic GNU core system and tools and not violating the GNU licenses by including non-GNU compatible in license software in a primary build for distribution.

The Red Hat model is yet again one of many loosely knit specifications for a Linux distribution. There's also the Debian model, the Slackware model, the LFS model, and various other models of Linux standards.

As far as what Patrick has publicly stated, he has only stated that he is looking into systemd, but hasn't made a choice of wether or not he will include it, but will, through speculation only, ONLY include systemd when and if the code has become viable enough to be of general usage, stable enough to be a true long term solution, and wants to make sure systemd is put into Slackware correctly. The if, and when, we get it, are simply just that, if and when, and so far that if and when are just that, if and when.

1 members found this post helpful.

Aside from the technical specs, choosing and using a distribution is based a lot on psychology. Trust is to be earned. There's a nice chapter by security expert Bruce Schneier about this, in one of his latest books. I've come to appreciate Slackware's soft-spoken no-bullshit approach, conveying a general sense of reliability. Without even going into the technical details, I know now I can trust the Slackware team for taking technical decisions that make sense, and build my stuff on that base.

Still aside from the technical specs, I'd think one of the main problems with Systemd is Lennart Poettering's loud affective cretinism and blatant immaturity. Whenever I read one of his blog posts or Google plus discussion threads, the only thing that happens is that my blood pressure goes up by a few notches.

From a merely human point of view, the Slackware-plus-Systemd combination feels like including a lame college standup comedian in a great Shakespeare drama. Which may explain the foul eggs and the rotten tomatoes in discussions about Systemd.

3 members found this post helpful.

@ReaperX7
Sorry - but that is not what he wrote. He wrote:
If he is looking into it, he would've written:
http://www.englishleap.com/common-mistakes/when-if

I've found that claim to be meaningless. On a stock system, sure systemd is a little faster, but once you background various daemons at boot, and you use things like preload or ureadahead. There really is no difference to traditional init systems.

In fact, for me; When i last used Arch Linux on my desktop i had a pretty basic system with preload and some backgrounded things, the boot time (with xorg automatically launching) took 13 seconds from bios to X. After a fresh install with systemd on Arch, it took around 20 seconds to boot.

I would use slackware for comparison, but i've never wanted to manually install systemd onto anything at all

If systemd can only vouch for faster boot times, then it is literally useless. In my experience, it can't even do that.

1 members found this post helpful.

Olav Vitters, Gnome release manager

Martin Gräßlin, the maintainer of KDE's window manager, KWin

You can use Weston without systemd-logind but then you are going to have trouble with session switching and running it without root. David Herrmann made four insightful blog posts about the subject.

The actual developers, distribution maintainers and users at large scale seems to like "his" projects (it has had contributions from 145 people during the past year alone) and therefore they get adopted so quickly.

The Lennart hate club seems to be going strong on LinuxQuestions.org but that doesn't really reflect reality. You could look for LWN.net or reddit.com/r/linux for different point of view.

It seems to work extremely well in demanding environments too. For example Pantheon runs more than 554 000 systemd units in over 150 000 containers. Sure, there's always something that needs a little work but when those quirks have been worked over we have better system than ever.

KDE will most likely carry the backwards compatibility for a long, long time in to the future. It's just that you miss on new features like faster startup. The startkde script is horribly slow but it hasn't changed in years so maintanance probably won't be a problem. You can probably hack around the logind depenedency on Wayland systems or just use X11 in the future too.

I fail to see the problem with systemd-logind documentation. The commands. the configuration files, the APIs and the pam_module has been documented and new developers, like David Herrmann could easily pick up its developement.

It seems that more and more distributions are moving to systemd (Mageia, Arch Linux, Fedora, openSUSE, RHEL, SEL, Sabayon, Nux, Chakra, Frugware, Tanglu...) and even more are considering it, most importantly Debian so I don't see where this comes from.

What a welcomming community...

You should read up on systemd because it hasn't never done that. The attitude towards it booting fast is that it's a side effect of doing things right.

2 members found this post helpful.