Fog Heen: Resigning as a Debian systemd maintainer [LWN.net]

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 17, 2014 14:43 UTC (Mon) by nix (subscriber, #2304) [Link]

All we can do is thank Tollef -- and Russ -- for much good work over the years and let them go on to other things.

Though I am a bit concerned about the volume of amazing talent Debian's tech-ctte has started to haemorrhage :( there are still a few people left, but I'm not sure anything the ctte has done has ever been this acrimonious.

It's time to put on the brakes

Posted Nov 17, 2014 15:06 UTC (Mon) by BrucePerens (guest, #2510) [Link] (23 responses)

Systemd issues are not unique to Debian. The kernel team has had similar problems and they ended up in a widely published whine by the Lennart Pottering.

Let's face it, folks. When a single little project develops personal problems on multiple fronts, it's not the kernel team or Debian's fault. It's not a technical issue. It's the way that Systemd has been pushed upon all of us.

But the project needs to be split up, and it needs to play well with others - including other init systems.

Oh, by the way, my Haswell system doesn't work with Debian Jessie at all, and it's a Systemd issue.

I have had enough.

It's time to put on the brakes

Posted Nov 17, 2014 15:26 UTC (Mon) by Thue (guest, #14277) [Link] (5 responses)

> Let's face it, folks. When a single little project develops personal problems on multiple fronts, it's not the kernel team or Debian's fault. It's not a technical issue. It's the way that Systemd has been pushed upon all of us.

systemd was adopted voluntarily by the Debian TC. From my understanding, there never were a Debian General Resolution to overturn the TC decision about which init should be default because everybody knew systemd would win. Otherwise you can be sure that Ian would have made one. Also remember that all TC memebers, including Ian, preferred systemd over sysvinit (ignoring Ian's acknowledged insincere tactical voting).

Sure, some people disagree strongly with the switch. But when the switch is a result of a majority decision (or at least 50% in the case of the TC), then it is hardly "pushed upon all of us". At least those that voted for it doesn't feel it is pushed upon them.

Also as I understand it, sysvinit still works perfectly well on Debian, and the only reason why it should stop working is if sysvinit supporters do not put in the effort to maintain it.

> Oh, by the way, my Haswell system doesn't work with Debian Jessie at all, and it's a Systemd issue.

When you switch out the guts of an operating system, there will be bugs. Hopefully those bugs will be fixed.

As mentioned, you can still switch back to sysvinit.

It's time to put on the brakes

Posted Nov 17, 2014 15:34 UTC (Mon) by tjc (guest, #137) [Link] (4 responses)

> As mentioned, you can still switch back to sysvinit.

I think a better option would be to allow the user to select an init system from the installer. This would probably greatly reduce the level of acrimony, and reduce the motivation to fork.

It's time to put on the brakes

Posted Nov 18, 2014 13:21 UTC (Tue) by obi (guest, #5784) [Link] (3 responses)

At this point, I'd like to see a fork: at least the anti-systemd crowd would then put in the work to make Gnome and other things work on non-systemd. And just maybe it'd act as a pressure release valve for all the bitterness we've seen for the last few years on the Debian ML.

But if I'm honest, I doubt a fork will happen. It's quite telling that most of the people actively doing the work seem to have no compunctions about working with systemd.

It's time to put on the brakes

Posted Nov 18, 2014 14:11 UTC (Tue) by Zack (guest, #37335) [Link] (1 responses)

Talking of forks and forking is silly in the Debian context. There would be a Debian Blend at most.

The question here is, how willing are systemd proponents (since they are the default init) to accommodate users who do not want to have systemd installed for whatever reason? If the majority is more than willing to acknowledge the need of some for a systemd free OS, there is no problem at all.
If the majority is unwilling to do put some sort of guarantee in place, there might be a need to create a blend which will probably consist of a modified installer, and a handful of packages, but will remain as close as possible to Debian the main project.

There is no major schism on the horizon here. Just a minor one where Debian at large can decide to continue as one whole project, or most-of-the-project-and-then-a-little, possibly at the cost of losing the "Universal OS" moniker.

It's time to put on the brakes

Posted Nov 18, 2014 15:09 UTC (Tue) by obi (guest, #5784) [Link]

I agree on your point that talking about forks (wrt Debian) is a bit silly, and that if one would diverge a bit a blend would be the way to go.

I may be wrong, but you seem to imply that the systemd side should accommodate users who want to avoid systemd; I respectfully disagree - it's up to those users to work with upstreams (like Gnome) and packaging teams to provide patches and implement the things that are needed (alternative logind f.e.). If they did, I doubt we'd even need a blend. IMHO the question is "who should do the work", not "will the work be accepted".

It's time to put on the brakes

Posted Nov 18, 2014 16:01 UTC (Tue) by tjc (guest, #137) [Link]

> At this point, I'd like to see a fork: at least the anti-systemd crowd would then put in the work to make Gnome and other things work on non-systemd.

I'd rather not see a fork. Splitting an already small community does not seem like a good idea, especially over something that can be fixed by a few packages and a modified installer. But if there is a fork, I wouldn't expect things like Gnome to be a problem, since there are alternatives.

It's time to put on the brakes

Posted Nov 17, 2014 15:44 UTC (Mon) by algernon (guest, #11573) [Link] (3 responses)

Is it really appropriate to rant about systemd on a news piece about one of its Debian maintainers resignation from the team? Considering that his resignation was due to the excessive hate... I'm sorry, but your reply is the perfect example of what is wrong, and what is pushing good people away from Debian.

It's time to put on the brakes

Posted Nov 17, 2014 15:55 UTC (Mon) by SEJeff (guest, #51588) [Link] (1 responses)

This is exactly my comment in the other thread about a similar thing regarding a long time contributor leaving Debian. Glad to see I'm not the only one thinking that was kind of uncool, even when it was by the guy who helped coin the open source definition.

It's time to put on the brakes

Posted Nov 17, 2014 16:37 UTC (Mon) by Darkmere (subscriber, #53695) [Link]

I think it's perfectly in character as an example of the Toxic behaviour of people in the debate.

It's also a negative spiral, as they are driving away the people who would fix their bugs, leaving only more bitterness in the wake.

It's time to put on the brakes

Posted Nov 17, 2014 17:04 UTC (Mon) by misc (subscriber, #73730) [Link]

I concur.

That's a bit distasteful and rather wrong.

It's time to put on the brakes

Posted Nov 17, 2014 17:47 UTC (Mon) by mezcalero (subscriber, #45103) [Link] (4 responses)

Wut? systemd doesn't work on Haswell systems? have a link to the bug report? that claim sounds quite adventurous to me...

Also even if that was the case, what does that have to do with whether systemd's components are shipped in one tarball or multiple?

Lennart

It's time to put on the brakes

Posted Nov 17, 2014 22:32 UTC (Mon) by nix (subscriber, #2304) [Link] (3 responses)

I'm reasonably sure Bruce said it was on *his* Haswell system.

It's time to put on the brakes

Posted Nov 17, 2014 23:54 UTC (Mon) by mathstuf (subscriber, #69389) [Link] (2 responses)

I think it's more of "why is Haswell so important a distinction that it needs to be mentioned?". If my black desktop reboots every day and it doesn't have to do with overheating when it's in direct sunlight, but some specific software, why would I add that detail?

It's time to put on the brakes

Posted Nov 21, 2014 9:37 UTC (Fri) by marcH (subscriber, #57642) [Link] (1 responses)

It's not detail for everyone. There are people who don't buy a Dell or a Lenovo or an HP system. Instead they buy an Ivy bridge or a Haswell. From a certain perspective it makes more sense.

It's time to put on the brakes

Posted Nov 21, 2014 12:20 UTC (Fri) by mathstuf (subscriber, #69389) [Link]

I build my home machine from parts, so that's what I do too. I guess I never considered calling it by the processor name but more as "my main machine" since it has gone through a few hardware upgrades over time. *shrug* Communication is hard and the Internet has made it much harder to get those details you get face-to-face which are already pretty hard sometimes.

It's time to put on the brakes

Posted Nov 17, 2014 18:07 UTC (Mon) by rodgerd (guest, #58896) [Link]

Funny that a guy who made such a huge deal about what he considered death threats from another developer is now dismissing similar threats.

And by funny I mean groteque.

It's time to put on the brakes

Posted Nov 17, 2014 18:38 UTC (Mon) by lambda (subscriber, #40735) [Link] (1 responses)

I'm running Jessie with systemd on a Haswell system just fine. I'm curious what you're doing differently, or if you have a bug reference.

I've been running Jessie/systemd for months now, on a couple of different systems, and the only systemd related issue I've encountered was a race in which cups left dangling symlink that had previously been valid, which caused systemd to crash; however, I was able to gracefully kill off important processes and sync, then reboot, everything worked fine afterwards, and the bug was fixed not too long afterwards. And this is the kind of issue you expect when running a testing distro that's in the middle of transitioning init systems and using the much more advanced capabilities of the new one.

So far, systemd has helped me debug some other issues, as "systemctl status" and the status output for individual services is much more informative than sysvinit, and journald has helped capture early boot message that were lost by dmesg because I had enabled some verbose DRM debugging messages without increasing the ring buffer size.

systemd is probably not my ideal init system. It is a bit more complex and monolithic than I would prefer. But it's so much better than sysvinit and upstart. I haven't tried other new or alternate init systems, but systemd is the only one that has real buy-in from a large number of other upstream projects, has full stable distros running on it, and actually provides the facilities to do proper multi-user login management (at least, as far as I know; please correct me if I'm wrong).

It's time to put on the brakes

Posted Nov 17, 2014 23:52 UTC (Mon) by CopperWing (guest, #82856) [Link]

Maybe Bruce is just using systemd-logind with systemd-shim, uselessd or whatever Debian is developing by itself to replace systemd as PID 1.

But without a proper bug report of his it's difficult to tell...

It's time to put on the brakes

Posted Nov 17, 2014 19:06 UTC (Mon) by pizza (subscriber, #46) [Link]

> Oh, by the way, my Haswell system doesn't work with Debian Jessie at all, and it's a Systemd issue.

Bruce, I'm quite disappointed in you -- because you, of all people should know what is or isn't a useful bug report:

How to report bugs efficiently:

http://www.chiark.greenend.org.uk/~sgtatham/bugs.html

And From ESR's "Smart questions FAQ":

http://www.catb.org/~esr/faqs/smart-questions.html#beprecise
http://www.catb.org/~esr/faqs/smart-questions.html#idp540...

It's time to put on the brakes

Posted Nov 17, 2014 22:55 UTC (Mon) by ovitters (guest, #27950) [Link]

The person didn't resign due to systemd. LWN quoted why. It your haste to blame systemd you seem to have overlooked that.

hold on a minute

Posted Nov 18, 2014 0:08 UTC (Tue) by HelloWorld (guest, #56129) [Link]

Tollef Fog Heen has been working (for all I know) in his free time in order to improve Debian and make it more useful for its users. And now that he resigns due to constant flaming from trolls who never did anything useful for society themselves, you can't think of a better thing to do than to put the boot in and blame the victims? I'm sorry, this kind of behaviour is just revolting and obnoxious, and certainly not worthy of a former DPL.

It's time to put on the brakes

Posted Nov 18, 2014 7:29 UTC (Tue) by edomaur (subscriber, #14520) [Link]

What ?

What kind of system are you running ? I am on a Haswell box under Manjaro (arch derivative, systemd) and everything is fine (except the printer but that has always been a problem in Linux and others)

It's time to put on the brakes

Posted Nov 18, 2014 13:20 UTC (Tue) by Doogie (guest, #59626) [Link]

systemd works fine on my haswell Fedora box. Perhaps you need to use a better distribution.

/kidding

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 17, 2014 15:22 UTC (Mon) by Zizzle (guest, #67739) [Link] (8 responses)

Anti-systemd people win the war through attrition - enough death threats, name calling, and nastiness will ensure no one wants to work on it.

A nice bunch to have in control of your distro.

There used to be a time when if you didn't like a piece of free software, you created something better.

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 17, 2014 18:44 UTC (Mon) by rahvin (guest, #16953) [Link] (7 responses)

But the goal of these people isn't to create, it's to destroy and through that destruction force other volunteers to do work.

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 17, 2014 20:08 UTC (Mon) by tjc (guest, #137) [Link] (2 responses)

I suspect that membership in the group "these people" is negligible in this case.

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 18, 2014 0:12 UTC (Tue) by rgmoore (✭ supporter ✭, #75) [Link] (1 responses)

It only takes a handful of bullies to make the environment really miserable if they put some real effort into it. This seems to be one of the big recent lessons of the internet; a relatively small number of trolls are enough to drive people off the internet if they're willing to be sufficiently awful.

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 18, 2014 8:05 UTC (Tue) by jezuch (subscriber, #52988) [Link]

> It only takes a handful of bullies to make the environment really miserable if they put some real effort into it. This seems to be one of the big recent lessons of the internet

A recent case study from Peter Watts:

http://www.rifters.com/crawl/?p=5370

[And the person mentioned there is/was just *one* person. The sad thing is that even just one person behaving like this accretes a group of abusive minions who look up to their "leader".]

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 20, 2014 4:59 UTC (Thu) by xxiao (guest, #9631) [Link] (3 responses)

It's different nowadays, most OSS developer are paid full-time employees, less people are going to create new software just because he does not like the existing one. Things change fast.

Without Redhat it's impossible for systemd to become so troublesome so fast. If Lennort did systemd in his spare time, this would be quite different and less destructive. He is controversial enough, now with Redhat's backup, it becomes fatal.

I don't care if systemd is good or bad technically, what I want is to have a choice, free software is about choice, if you don't honor that, please develop something else.

I see systemd as a tumor from Redhat that splits the linux community. Redhat's largest competitor is Debian-derivatives, systemd serves well for disrupting Debian.

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 20, 2014 12:05 UTC (Thu) by jwakely (subscriber, #60262) [Link]

This is just dumb conspiracy theory crap. Red Hat's largest competitors are proprietary vendors, not other GNU/Linux distros. Red Hat is not big enough to develop GNU/Linux on its own, so disrupting or harming other distros would also harm Red Hat, especially a major distro like Debian that does loads of work for the ecosystem and the community. Additionally, many RH employees were Debian users before they got hired and continue to be Debian users now, so would be very unhappy if they were being used to harm it.

(But then I'm a full-time RH employee, so I'm probably being paid to say that and monitored by my controller to make sure I don't admit we're trying to destroy Debian, or whatever silly idea you want to believe.)

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 23, 2014 22:51 UTC (Sun) by csamuel (✭ supporter ✭, #2624) [Link] (1 responses)

Please cite the sources and numbers backing up your assertion that "most OSS developer are paid full-time employees".

All the best,
Chris

Fog Heen: Resigning as a Debian systemd maintainer

Posted Nov 24, 2014 18:48 UTC (Mon) by mathstuf (subscriber, #69389) [Link]

Well, there are a few things which need tweezed apart even in the assertion here. Do we mean developers by body count? Code output (not that measuring this isn't already fraught with various issues)? How do you split personal contributions from on-the-job contributions? Does *any* full-time employment put them into the category, or just ones which have some FOSS duties, or does it have to be 100% of the time is for FOSS? For some combination, things are, to me, leaning towards a definite yes (any full-time paired with either body count or code output), but others aren't so clear-cut.

So while numbers would be nice, let's get first make sure we know what we're asking for :) .