Well, since the Debian init system committee appears to be deadlocked between the worst two out of all possible options I decided to make this poll to see what the Debian user base is actually thinking.

Canonical's Upstart is licensed under the GPL v2, so it is not a "non-Gnu way".

Really, I thought they had their own CCA agreement, it was actually one of the main reasons debian didn't want it.

They have the CLA, but that doesn't change the license of the project. Debian could simply fork it or just maintain their patches out of tree.

Yes.

It really is an interesting situation. As a recovering Ubuntu user I can see the problem quite well.

I am not sure that upstart is actually the best solution although it works fine. I am not an expert at all on booting a system.

Ubuntu is based on Debian. Ubuntu has to work with upstream no matter how they feel about it. So you have a large number of Debian devs, about a third I think, that are Ubuntu devs also.

Obviously they kind of like a system they are used to just like anyone else.

Canonical has worked for years, very hard and dilegently, to alienate the rest of the Linux community. Done a fine job of it. There for there is automatic resistence to using anything that has a hint of Canonical/Ubuntu "coruption" in it.

While Canonical and Ubuntu deserve to be thought of by the Linux community as bad members of said community this should not completely blind the community to things that actually may be a good idea no matter where that idea comes from.

The kernel project does use code from MicroSoft. Surely we can consider code from Canonical.

The problem with porting systemd to other platforms like hurd is a real problem. This is a serious matter and needs serious thought.

Myself I would love to rant about not using upstart because it comes from Canonical. I hold a grudge. This is not, however, about Canonical or Ubuntu. It is about what is best for Debian. I think we need to keep that foremost in our thoughts on this issue.

I certainly hope that people that actually know enough to make this call are doing just that.

Well, while this sample certainly isn't large enough to be representative it appears that less than 15% of people actually WANT systemd.

I don't think you have a decent sample because most people know too little about thie issue to have anything but some silly social media type opinion on it.

Kind of like the "ubuntu non gnu way" in reference to upstart. Most people are aware that I have some fairly strong opinions on the general character and worth of Canonical and Ubuntu. These are not possitive opinions. The trend seems to indicate they are getting more negative all the time. However, there is nothing, as far as the software code goes, wrong with upstart.

This has been implied outside the discussion among Debian folks that have to actually make this decision. It has never come up with people that know anything about upstart.

It does, however, show some of the human problems with making this call. A huge number of Debian devs are also Ubuntu devs. Many of them undoubtedly support upstart. Some others probably oppose it simply because it comes from Canonical.

This is much more than a debate on which one of these systems is best. They all do the job. Which one is best for Debian to use, with the future of Debian in mind is what the people that need to make this choice have to deal with. This is complex and difficult.

Should, really, have been decided back before Wheezy was released. Was put off because of just this sort of hostility towards some of the possible solutions. This leads me to think that the best bet is probably to just punt the thing down the road. Do something awkward for Jessie, like the choice of options being left to the user (think of the amount of space needed in the ISO just for this crap) to try and sort out (many of those users having no idea what they want to do). That would give some indication what users were thinking.

The important people in this is the dev community for Debian. These people have worked together for years with one goal in mind. Debian Stable. When this whole thing finally blows over they still need to do that.

In that light, while it is certainly fine for us to discuss this, some of the things I have seen are just inflamatory, stupid and, at best, simply silly. Is this really something we want to encourage?

We have Linux users on this forum that don't understand that Linux Mint and Ubuntu are not the same thing. Are these people qualified to make this choice for Debian. I doubt it. Am I qualified to make this choice? That one I know and the answer is no.

I do know what an init system does. Have made very minor tweaks to it. Understand the differences, advantages and disadvantages of the options? No. Haven't got a clue.

I have used upstart. Works fine. I have used systemd as put out by Arch. Works fine.

Great. The folks at Canonical and the folks at Arch have decided which system is best for them. This has little bearing on what is good for Debian. People with a vote in this within the Debian system deserve a little bit of room to think. They really don't need a bunch of people with minimal understanding of the ramifications of this decision on an organization that is older than some of the users of its product. Or people like me that are considerably older than Debian but didn't even know it existed 16 years ago.

Yes we have a right to speak our minds. And have opinions. We don't need to be screaming campaign type slogans about the different choices.

I am sure there is plenty of that behind the scenes anyway. Devs are people too. Let's just hope they can make a choice that works in the long run without hurting their ability to get along with each other.

3 members found this post helpful.

My viewpoint is simple. Why replace sysVinit that has worked perfectly well for thirty or forty years? From what very little I know about the issue, to me it appears to be a case of change for the sake of change.

1 members found this post helpful.

SysVInit still works if you take the time to learn and use it. Obviously Lennart and Kay aren't as brilliant as they claim if they can't write scripts. Fedora and Red Hat developers are no better. Their scripting skills are some of worst possible.

Systemd is nothing but an overhyped, overglorified mess that is only pushed down our throats by Red Hat so they can redefine what they feel is GNU/Linux into RH/Linux. It's already destroyed many sound and working projects and forcibly deprecated many packages that never needed fixing, just small updates.

At least Slackware, LFS, Gentoo, and several others have people who are responsible leaders. I would hope Debian does as well. The UNIX ecosystem can not survive on Linux alone. Solaris, illumos, BSD, and many others in the vast ocean of UNIX have all contributed to Linux in various ways. To ignore that is to spell doom for Linux.

OpenRC would be the best choice if stand-alone sysvinit isn't cutting it. At least OpenRC is multi-OS friendly to the UNIX ecosystem, and does the same stuff systemd claims to do.

1 members found this post helpful.

I like the simple and uncomplicated way things have been done inall Unixes except for Slowaris until now,Pottering's lies non-withstanding it is a fact that the current init system and philosophy has worked just fine for over four decades and on all sorts of business-critical servers, thank you very much.
Furthermore we are all entitled to our opinions as users, and of the ability to state it in whichever way we see fit, moreover as advanced users (read who actually know what the difference between an init script and a cron job are ) I believe that it is our right to stand up for what we like.
I generally do ot care for Ubuntu, I do not believe it has actually contributed much to the FOSS community, other than a few lazy and not very competent or interested desktop users, nor do I care much for the Upstart "innovations", if I want a watchdog to keep my daemons running it would be trivial to use something like nagios, or even to start a wrapper daemon written in the likes of Perl, Python or Ruby.
There is also the important issue of the CLA, which I do not think has been properly addressed, furthermore Ubuntu is straying more and more from what I consider common Linux with stuff like Mir and Upstart(Not a genuinely bad thing, since I do believe that freedom of choice is critical where FOSS is concerned, however IMO Ubuntu isn't all that good at providing actual choice to users), frankly I do not care much for their way of doing things, nor do I think that forking upstart is the best option here, since it will likely mean that there would be two hostile development camps.
Neither do I see a major need for socket activation, a few extra cpu cycles and an mb or two of memory at most vs. a slower start or no start at all for stuff I might need urgently, like connecting over ssh to a system experiencing major issues for example, and that can't start mthe ssh daemon due to those issues is no tradeoff at all in my opinion.
However I would still choose upstart as the lesser evil, as it is not trying to sink its claws in other OS functionality and to be a paradigim shift that basically makes using alternatives much harder, nor does it force so many useless dependencies on you.
If I said I do not want cgroups, or that I have anything against them I would be lying, but really being forced to adopt them just because of init is dictatorial, I do not care for D-Bus either.
systemd is a disgusting octopus, trying to force you to adopt not just its way of booting your system but also forcing you to ditch standard cron, use a replacement for syslog, udev, rewrite your applications to use some D-Bus and more *kit style garbage and I do not care for it, or for Pottering.
This is NOT the UNIX/FOSS/GNU way.

This!

I totally agree with your sentiment, Red Hat is trying to be the Microsoft of the Unix/Linux world IMO.

LIVE FREE OR DIE!

Even then we've had init systems like s6 and RunIt which work well also.

Sorry but Lennart Poettering reminds me too much of Jim Jones to trust him. Too big an ego, and not enough people to stand up and stop him. At least I won't be drinking the poison kool-aid.

If push comes to shove, I'll go back to Windows and use Cygwin.

I have no idea who Jim Jones is, but he has always reminded me of another class A manipulative megalomaniac and psychopath, Robert Moses.
he is a German, but was born in South America, right?
I wonder if he has any relation to a few germans that ran away there after this little military conflict which they kicked off blew up in their face

From a GNU/Hurd pov, I highly disagree with that.
It might be harder for kFreeBSD (the cgroup part), but not for mach + hurd.

Do read:
https://teythoon.cryptobitch.de/post...-init-systems/

An official statement is being prepared.
Debian/Hurd porters position in the default init system debate (still in draft):
http://thread.gmane.org/gmane.linux....rts.hurd/15235

CGroups = Jails

It's not so much that kFreeBSD or any other BSD has to support CGroups, it's systemd needs to support Jails and other non-Linux only software and system features.

I don't know where you got this from, but jails are a completely different thing. You can use cgroups and containers to achieve something similar to jails, but you don't have to use it that way, mostly it is used it to limit resources for specific processes and make it easier to reliably shut down their child processes, without separating them from the original system.

1 members found this post helpful.