Shared publicly  - 
 
Good news to hear that +Arch Linux (official) is switching to +systemd for good! The ArchLinux folks have already done a fantastic job at upstream collaboration, and I am looking forward to more of that!

So, I guess it's official now that Fedora, RHEL (hence CentOS, Oracle too), OpenSUSE, Mageia, Mandriva, Tizen, ArchLinux have switched for good or are in the process of switching. Debian, Gentoo have it in their package repositories but not as default. Slackware of course, is, well, Slackware. Ubuntu is a bit weird, on one hand they don't include it, but then again Canonical tries to sell it as part of their IVI solution.

That looks pretty good, doesn't it?
78
13
Jan Alexander Steffens (heftig)'s profile photoBartłomiej Piotrowski's profile photoLaurent Carlier's profile photoChandan Rajendra's profile photo
20 comments
 
+Andreas Proschofsky they had a booth at the GENIVI meeting, and they tried to sell Ubuntu as GENIVI implementation. Now, GENIVI requires systemd, hence they had to add it too. We had a very weird discussion about that with their booth people, where we asked them something like "So, do you support systemd officially?" and they said something like "It's community maintained!"; "So, your GENIVI implementation is community maintained?" -- "No, we support that."; "But GENIVI implementations includes systemd" ... didn't lead to anything...
 
Gentoo couldn't just set systemd by default because of absent openrc networking compatibility. At gentoo, networking configurations scripts integrated to openrc, so just using raw systemd on non-notebook configurations with NM is not an option.
 
"It is written by Lennart Poettering. I have no idea how this ever became a “genuine” argument against systemd, but almost any discussion involving systemd ends up being about Lennart also writing pulseaudio. Maybe we need an updated version of Godwin’s Law for this situation… Any time an argument reaches this point, I know all things technical are out the door and I stop following."

That part was funny, now we have two ways to end a technical debate, nazis and "it was written by Lennart" O:-)
 
+Pierce Andjelkovic And honestly, I think they should. People should think they know best when they make decisions in their fields. I do believe this too for myself in my field, it's only healthy and right for them to do as well. Not being sure about your decisions usually means you didn't do your homework I guess. I mean, I am sure that people disagree with many of their decisions, but I am sure they have their reasons.
 
+Kai Mast Yes, and it sucks balls... Although I like the Aptd author and I think the Aptd code is great, doing the compat-layer only produces lots of problems for us and it blocked PackageKit adoption in the first place. With our recent changes on PackageKit, the compat-layer will be completely broken, and there's nobody at Ubuntu fixing it at time. (Every compat-layer will always be behind the original product in features and spec support)
For systemd, I don't think creating a compat-layer is even possible.
I'm tracking https://blueprints.launchpad.net/ubuntu/+spec/foundations-q-session-management for systemd support in Ubuntu, as separating systemd-logind from systemd is not possible. (And I so much want to use systemd in Ubuntu/Debian ^^) If they really want to add systemd features to upstart, they're developing in a private and closed branch, as there has been only bugfixing activity in upstream's main branch.
+Lennart Poettering Exactly!
 
The only thing i can wail about systemd is, that i have to make an alias sctl to systemctl because it takes too much letters to type in bash until tab-completition kicks in. And "systemctl restart smb.service" is too much to type - i'm lazy! :) Other than that, i'm already owned... ;)
 
+Lutz Griesbach +Tim JP, it might delight you to learn that very recent versions of systemd (F18 and beyond) understand a couple of shortcuts when specifying service names. Example: "systemctl status /dev/sda" is now equivalent to "systemctl status dev-sda.device", "systemctl status /home" is now equivalent to "systemctl status home.mount" and finally, "systemctl status httpd" is now equivalent to "systemctl status httpd.service". If the latter was the biggest missing feature for you, I guess you should rejoice now... ;-)
 
I (non-seriously) vote for 'svc' just too poke Solaris. ;-)
 
For what it's worth to those waiting for it there is active development going on now by a contributor to add proper completion support in C to get very efficient and accurate completion candidates. It's gonna rock!
 
I'm of a different opinion, I think that surety and conviction are often a bad sign, that the person doesn't understand enough to perceive problems or learn from them. I find that the more knowledgable the person the more they realize they don't know everything and are up front about caveats in their reasoning.

That said, systemd looks like very much like the right way to go about things and the reasoning behind its workings seems solid to me, the general criticisms have not been so well reasoned on my opinion.

I'm glad that it is finding success and is on its way to becoming the standard. 
 
So, you've conquered audio, system startup; what's next? How about GUId?
 
+Auke Kok is working on user sessions for systemd.  It definitely makes sense to unify the handling of different type of sessions.
Add a comment...