Systemd violates social contract

[deltaflyer]

moved to General Discussion as it's not a question

[keithpeter]

Hello All

There has been some technical discussion on debian-devel mailing list. The message below raises an interesting point - what do you actually mean by 'running without systemd'?

https://lists.debian.org/debian-user/20 ... 00751.html

I'm quoting the 5 possible interpretations of 'not using systemd' from that message by Joel Rees

Should we ask Martin whether he would be satisfied if

(1) systemd runs at some pid higher than 1?
(2) systemd is loaded, but doesn't actually run at all?
(3) systemd libraries are loaded, but systemd itself is not?
(4) Some emulation layer provides the functionality and no code from the systemd project gets to touch his disks?
(5) None of the apps he needs ask the OS to do any of the sort of things that systemd uniquely does?

I personally am not going to be satisfied unless #5 is met. That's how deep I perceive the design bugs in systemd to extend.

I think a lot of people are going for alternative (3) in Joel's list. That is how I see the systemd-shim project, and similar work in the OpenBSD world and on the Slackware/Gentoo forums. They want to run full-fat desktop environments and they don't want to e.g. fork KDE and remove dependencies on dbus, logind, systemd which would be a colossal undertaking as I understand it.

Which of the 5 alternatives would you be asking the Debian council to support as a viable alternative? How much development work would the alternative you choose need if supported? Is it possible that the Debian technical committee had little practical alternative as a result of decisions made by upstream projects (logind and dbus)? Is it possible that the encroaching mudball of interdependent subsystems we are seeing is arising as a natural part of the way GNU/Linux is developed (large numbers of independent projects each one responding to changes in all the others in the previous iteration) and that Debian's choices as essentially a packaging/bug-fixing/publishing organisation are limited?

I actually don't have answers. I'm sure a Debian derivative based on alternatives (3) or (5) will pop up quite soon given the strength of feeling around this issue. It might be that an alternative (5) could be a Debian Blend aimed at server folk.

There are a couple of systemd threads here in General Questions where people are trying to find information and make practical suggestions for Jessie.

[Randicus]

Option (6) - Do not infect the system with that shit. Oops. Too late.

[llivv]

3

keeping fingers crossed

thanks for all the work so far

____________________________________

bigonvilles changes to init?
adaption of snapdragon dri to slow?

[keithpeter]

Option (6) - Do not infect the system with that shit. Oops. Too late.

Well, you can use a time machine.

CentOS 6 (support until 2020), possible Wheezy LTS, Slackware while Volkerding thinks he can avoid the manure (he is pragmatic though, as soon as it looks like huge piles of work he will allow the stuff in), Gentoo, and the BSDs. OpenBSD is, amazingly, quite laptop friendly. There is also Ubuntu 14.04 with 5 years support and still on upstart but I think we can all rule that one out

@Illvv: work? Me? Nah! Just a couple of installs trying out edbarx's suggestions and pointers from adenukolnis and mardybear. An alternative (5) desktop is going to be very oldschool indeed I'm afraid by the look of it.

Back On Topic for this thread: politics is the art of the possible. Is a sysvinit 'debian blend' achievable in the time frame for Jessie or Jessie + 1?

[buntunub]

Option (6) - Do not infect the system with that shit. Oops. Too late.

Comments like this will not produce the results we would all like to see achieved - freedom of choice. Can we keep the discussions around Systemd in Jessie+ constructive and focussed please?

In the General section, I replied to your post Chao with some questions just now that I think are pertinant. From what I can gather, the decision to go ahead with Systemd is still open for Jessie. Although a decision was reached for Linux architectures only, it was not established that SysVInit and Upstart will not be supported. Again, so far as I could gather from that overly long bug thread. So, my question is, will SysVInit and/or Upstart be supported in Jessie+, even though they will not be the default INIT system for Linux architectures?

Kiethpeter - Given the above, perhaps the question would be, is it possible to run Jessie using an alternative INIT system before the freeze?.. And to follow up on that, will the freeze happen before alternative INIT system maintainers can get the patches in that would make question #1 possible?

According to what I have read from the bug report on the TCCE decision, the GNOME maintainers would be happy to receive a patch for logind.

And you are right. Systemd should NOT be running as PID 1

[Randicus]

OpenBSD is, amazingly, quite laptop friendly.

I am making this post using OpenBSD. I was a little, and pleasantly, surprised that my fairly new hardware has no difficulty running it. I hope that it can meet my needs (printing, East Asian languages and the like) and that I can figure out everything I need to know to configure the system, because its performance is amazing.

Comments like this will not produce the results we would all like to see achieved - freedom of choice.

Are you really naive enough to believe: 1) Debian developers are making decisions based on what the users want? They obviously are not; and 2) That those developers will read and heed a discussion involving a handful of FDN members? Look at the title of the thread - Systemd violates social contract. Hardly a discussion to find a way to overcome systemd in Debian. There are at least two threads dealing with that.

Can we keep the discussions around Systemd in Jessie+ constructive and focussed please?

Constructive? We do not like systemd. Rant. Cry. It would be nice if ... Nothing posted in this thread will result in anything possitive. The powers that be will not tremble in fear because of this thread. Focused? Focused on what? Debian users' impotence in the face of betrayal? Systemd in Debian is a reality. Users only have one realistic choice; continue using Debian with systemd or use a different system. No amount of whining or serious discussions about theoretical options will change that.

[keithpeter]

We do not like systemd. Rant. Cry. It would be nice if ... Nothing posted in this thread will result in anything possitive...

How many people saw the Squeeze LTS project coming? That happened because the kind of people who run servers saw an advantage in it and were able to muster the resources needed to continue packaging updates on a limited range of software.

So, no, not this thread, but I imagine that there will be a sysvinit alternative available for alternative (5) when Jessie is released.

[keithpeter]

https://lists.debian.org/debian-user/20 ... 00846.html

and

https://lists.debian.org/debian-user/20 ... 00849.html

Could be a move towards a mailing list specifically to discuss alternative init systems and systemd independence. That is of course usually the first small step in a fork. I'll be watching this one.

[n_hologram]

Since the idea of systemd is still relatively new to me, how would it affect Debian users running, say, XFCE (built from a netinstall)? I've really only heard of it in terms of Gnome3. In addition, I've run across threads from other forums (ie, Gentoo) in which a general consensus was, "if you don't like systemd, don't use it." If/when Debian fully adopts it, would it really affect the whole system so deeply that it would become a matter of "deal with it or get a new distro?"

[dasein]

Since the idea of systemd is still relatively new to me, how would it affect Debian users running, say, XFCE (built from a netinstall)?

One of the many things to loathe about systemd is that it ends up becoming a dependency of just about everything, so avoiding it is difficult, unless you're willing to spend a lot of time/effort cobbling together a low-end faux-DE from pieceparts. (Not saying impossible, saying difficult and definitely time-consuming.)

If/when Debian fully adopts it, would it really affect the whole system so deeply that it would become a matter of "deal with it or get a new distro?"

Pretty much, at least for the vast majority of users. I found myself downloading slackware today, for that very reason. And, purely pedantic point, Debian has already adopted systemd, and shows no signs of recognizing that decision as short-sighted. Once Jessie becomes Stable, using "stock" Debian with any full-featured DE will pretty much mean using systemd. (Yes, Wheezy will still receive security updates, quite possibly for a long time if LTS comes its way; but that merely delays the problem, rather than solving it.)

I found myself downloading slack just today, because I've come to accept the fact that long-term use of Debian means being stuck with systemd, and that's simply not a price I'm prepared to pay. At the end of the day, systemd is nothing more than a bad implementation of a Really Stupid Idea. "Solution in search of a problem" is the nicest thing anyone can say of it.

It's a serious bummer. I truly thought that my distro-hopping days were over.

[golinux]

At the end of the day, systemd is nothing more than a bad implementation of a Really Stupid Idea.

I love your way with words!

[goulo]

Since the idea of systemd is still relatively new to me, how would it affect Debian users running, say, XFCE (built from a netinstall)? I've really only heard of it in terms of Gnome3. In addition, I've run across threads from other forums (ie, Gentoo) in which a general consensus was, "if you don't like systemd, don't use it." If/when Debian fully adopts it, would it really affect the whole system so deeply that it would become a matter of "deal with it or get a new distro?"

FWIW I'm using LXDE as my desktop (with a sid/unstable system), no pulseaudio, and I don't have systemd installed and haven't had any problem being in this state (but I do have a couple libsystemd* packages installed.)

(Hmm, there is an ambiguity in these discussions whether people mean literally "systemd" itself when they talk about having systemd installed, or also libsystemd* libraries (and perhaps even dbus?) That semantic ambiguity is annoying/frustrating/confusing...)

I am concerned that down the road systemd might become "necessary" on my install, in which case I'll leave sadly Debian. I trust neither its motives nor its quality.

[keithpeter]

(Hmm, there is an ambiguity in these discussions whether people mean literally "systemd" itself when they talk about having systemd installed, or also libsystemd* libraries (and perhaps even dbus?) That semantic ambiguity is annoying/frustrating/confusing...)

Hence, up the screen, my decision to quote an excerpt from an email on debian-user mailing list that identifies 5 'states' of systemd-less-ness! I agree that being clear about what you mean is a good idea. I suspect a 'remix' or derivative will emerge based on Jessie providing one or other of the alternatives. Debian at least provides the 'materials' needed.

(Yes, Wheezy will still receive security updates, quite possibly for a long time if LTS comes its way; but that merely delays the problem, rather than solving it.)

@Dasein When your my age, the time horizon you tend to plan over is around RHELs support period

I found myself downloading slack just today, because I've come to accept the fact that long-term use of Debian means being stuck with systemd, and that's simply not a price I'm prepared to pay. At the end of the day, systemd is nothing more than a bad implementation of a Really Stupid Idea. "Solution in search of a problem" is the nicest thing anyone can say of it.

Remember Volkerding is taking an admirably pragmatic view of all this, as is typical of the slackware mind-set. Systemd will be avoided until doing so becomes a lot of work or makes it impossible to provide a decent desktop. See his 'interview' from 2012 on linuxquestions.org. Slackware 14.1 with a few slackbuilds is now on my 'typing box' at home and will stay there for this academic year.

I suspect that when someone unpacks the multiple random walks that got us here (dbus, Gnome, udev, systemd, wayland all dancing around each other iteration by iteration) we'll see how each small individual decision by each of the main upstream projects seemed logical at the time and in the context, but the cumulative effect resulted in an overall structure that noone really wanted.

We need to show support for Debian's provision of diverse 'materials' in the repositories as far as possible.

[dasein]

@Dasein When your my age, the time horizon you tend to plan over is around RHELs support period

And when you're my age, you're wondering if you'll make it through to lunch.

I should start planning migration of my servers to BSD. Speaking only for myself, the introduction of the equivalent of Windows' svchost into Linux dramatically reduces the number of reasons to prefer Linux as an OS. (I dunno what Poettering is smoking that leads him to imagine that Microsoft "got it right," but I want some.)

(On second thought, no, I don't.)