A possible plan to fork Debian to avoid systemd

16 respostas [Última entrada]
salparadise
Desconectado
Joined: 09/08/2013

http://debianfork.org/

This is very welcome news.

quantumgravity
Desconectado
Joined: 04/22/2013

Sorry for the offtopic, but i love the design of the page, especially the fonts.
Does anybody know which fonttype this is?

onpon4
Desconectado
Joined: 05/30/2012

It looks like it uses fonts from Google called "Oswald" and "Noticia Text".

onpon4
Desconectado
Joined: 05/30/2012
quantumgravity
Desconectado
Joined: 04/22/2013

Wow, thank you!

onpon4
Desconectado
Joined: 05/30/2012

Honestly, I think this is silly. If you love SysV-init so much, there's nothing stopping you from using it.

I found this really funny, by the way:

http://forkfedora.org/

For those who don't get it: Fedora has been using systemd for a long time now (about 3 years).

onpon4
Desconectado
Joined: 05/30/2012

I'd also like to note that there's no guarantee that such a Debian fork would have a standard for including only libre software in its main repo like Debian does. Depending on these people's attitudes about that, it could be that such a fork may have no proprietary software repos at all, or it could be that such a fork may merge the main and non-free repos. Let's not go endorsing the effort until what happens with that is clear.

salparadise
Desconectado
Joined: 09/08/2013

I think it's more about sending a message that systemd is not welcomed by all and that if a distro as old and as serious as Debian intends to fork over it, that perhaps the resistance to systemd, far from being a small group of malcontents, is actually a seriously considered objection.

It is beyond belief to me that so many people are happy to have more or less every distro now under the control of Red Hat, for that is what has happened. And Red Hat is a Corporation and Corporations don't dance for people, they dance for profit.

quantumgravity
Desconectado
Joined: 04/22/2013

Debian is probably the most frequently forked distro out there. 99℅ of the forks died sooner or later, and therefore i don't think this page is worth mentioning so far.

jbar
Desconectado
Joined: 01/22/2011

I'm not skilled enough to have an own opinion on init systems, but reading https://wiki.debian.org/Debate/initsystem/systemd I have the impression that systemd is simpler and easier to mantain and debug than the previous init system.

Magic Banana

I am a member!

I am a translator!

Desconectado
Joined: 07/24/2010

It is. http://forkfedora.org gives an excellent example. The so-called "veteran admins" of http://debianfork.org have certainly spent all their working life hacking the kind of ugly script on the left (and their dependences spaghetti-style, with some hazardous 'sleep', etc.). They do not want to learn systemd's short descriptive files (that systemd organize in a nice dependence tree). They do not care that systemd is faster, safer (using cgroups), uses a uniform format for log, etc. That is called "conservatism". It is plain stupid.

In contrast all members of the Debian technical committee but Ian Jackson (who was leading Debian 15 years ago... and has not done much since then) prefer systemd, Upstart and OpenRC to sysvinit. And (again setting Ian Jackson apart), each of them prefers systemd or Upstart (the order varies but twice more members preferred systemd, hence the final decision) to OpenRC, which is a sysvinit++. Here is a reference: http://mrpogson.com/2014/02/09/systemd-init-system-in-debian-jessie-democratic-decision-or-uncivil-war/

salparadise
Desconectado
Joined: 09/08/2013

They do not want to learn systemd's short descriptive files (that systemd organize in a nice dependence tree). They do not care that systemd is faster, safer (using cgroups), uses a uniform format for log, etc. That is called "conservatism". It is plain stupid.

With respect you are attributing thoughts and feelings for which you have no evidence.
The objection is philosophical. Try as you and others might, to derail the argument, or to try to misrepresent it as immaturity, laziness, fear or ignorance, the point remains - some of us object on philosophical grounds and I've yet to hear anyone who likes systemd respond to this point honestly. Because systemd does violate the long standing and previously widely praised and accepted philosophy and there is simply isn't any argument capable of dealing with that point. So every argument is turned into something else, so that point can be ignored. Because there is no answer other than "you're right, systemd does violate that philosophy".

onpon4
Desconectado
Joined: 05/30/2012

The only "philosophical" objection I've heard is the one that systemd is a "monolithic" program that doesn't stick to the "Unix" philosophy of "doing one job well". Problem: systemd can't be described as "monolithic". systemd is a huge collection of many different programs that happen to share a development team and repository. You might as well call GNU "monolithic" with that logic.

Magic Banana

I am a member!

I am a translator!

Desconectado
Joined: 07/24/2010

He is not the only one (see, e.g., http://0pointer.de/blog/projects/the-biggest-myths.html by systemd's leader) but Michael Stapelberg has "honestly responded" to those concerns a long time ago: https://people.debian.org/~stapelberg/2013/06/09/systemd-bloat.html

He even listed the tens of very small (dozens of kB for the largest) binaries that the systemd project ships. How they "undermine the basic design principles of 'do one thing and do it well'" (as written on http://debianfork.org/) remains a mystery to me.

As I was pointing out, the big discussion was all about systemd vs. Upstart (by Canonical). Systemd won and Canonical decided to stop Upstart's development and make Ubuntu switch to systemd. Trisquel being based on Ubuntu, it is currently using Upstart and you seem to ignore it. It will switch to systemd when Ubuntu will as well (in 2016 I guess).

The discussion has never been about sysvinit vs. the more modern init systems because every member of Debian's technical committee but Ian Jackson (see http://www.chiark.greenend.org.uk/~ijackson/ to understand that he has mainly been inactive for the last 15 years) agrees: sysvinit is inferior. It is a nightmare to administrate (a bunch of interdependent Shell scripts, spaghetti-style) and to debug (every daemon writes its log in a different format), it does not exploit today's hardware (no parallelism whatsoever), it is insecure (a binary can escape the supervision of its parent by double-forking), etc.

Ian Jackson now proposes a general resolution (https://www.debian.org/vote/2014/vote_003#text) whose fundamental sentence is:

In general, software may not require one specific init system to be pid 1.

So, basically, he proposes that all software would align on the lowest common denominator, sysvinit, and not exploit any of the features of systemd. Call that what you want. I call it "conservatism" and consider it plain stupid.

And because he proposes a "general resolution" (a pretty rare event apart from the election of a new leader), all Debian developers (most of them knowing very little about the init system) could fight each other on it. How good is it for the Debian project?

onpon4
Desconectado
Joined: 05/30/2012
mprodrigues
Desconectado
Joined: 05/03/2014

I don't think the fork will solve the problem, anyway a balanced position on why proponents of systemd will never get along with opponents of systemd http://uselessd.darknedgy.net/ProSystemdAntiSystemd/ and here is a list of common fallacies that proponents of systemd using http://judecnelson.blogspot.com.br/2014/09/systemd-biggest-fallacies.html including an explanation of why the joke in fedorafork is not valid.

salparadise
Desconectado
Joined: 09/08/2013

The future of GNU/Linux development is now owned by a Corporation whose biggest customer is the US military and if that doesn't make alarm bells ring, there really is very little hope.

The whole systemd thing is based on a lie - that prior to its appearance, Linux was all but unusable. We hear phrases like "spaghetti like mess of files that was sysVinit". I've been using Linux since 2002 and the first time I heard anyone dissing sysVinit was when I started reading up on systemd over the last few months. All I see is the exact same sort of spin and manipulation that's employed to swing public opinion behind whatever new and stupid policy has been thought up in the corridors of power. It's straight out of 1984. A few systemd developers start saying "we hate sysVinit, we've always hated sysVinit" and lots of people repeat it, parrot like and suddenly it's become 'the new truth'. Suddenly loads of users see sysVinit, a system they used for ages without thought or complaint, as archaic, poorly written and in need of replacement. It's called "manufacturing consent", where you learn to want to something new, that you didn't know you needed and thus didn't really need, in order to suit the agenda of someone else.
I guess that mostly what is shows is that the world of Linux is considerably more compromised than previously thought and that the majority of users have been carefully programmed to not care about such things (the hostility shown towards anyone promoting libre-software is indicative of this) and so the future is not as bright as it once was. And that's the other side of this - we live in a world where the sort of underhanded Corporate tactics used to gain control of markets, to stifle competitors, etc, is so par for the course that it no longer raises shouts of anger or outrage (this is a most dangerous position to be in). And now it's being applied to us.