On the sickness of our community

Posted Oct 16, 2014 13:10 UTC (Thu) by ksandstr (guest, #60862)
In reply to: On the sickness of our community by niner
Parent article: On the sickness of our community

This entire "hey, stop blaming the victim!" trope is poison to a rational discussion. It permits anyone to pre-emptively don the cloak of victimhood and then be excused from anything and everything. Analysis is rendered taboo in its cradle and cynical rules-gamesmanship rules the day in the name of "decency" and "politeness".

We don't call victim-blaming when someone jumps in front of a speeding car and is then said to have been in a position to prevent himself from being run over, and therefore partially responsible. Even if that someone jumped onto a zebra crossing, and was therefore formally in the right. Cars don't stop on a dime just as there's no way to anger broad swathes of the Linux community without getting flamed to a crisp again and again.

It's the same thing with systemd as it was with Pulseaudio. Lennart uses every shenanigan in the book to push his software through dependency creep, disregarding any technical critique (e.g. the brittleness of the binary logging mechanism, the monolithic IPC architecture and its tendency for becoming wedged, etc.) and especially not giving two fucks that said creep breaks everything it touches outside of the Brave New systemd World. There are good arguments being made that systemd is Red Hat's hostile takeover of the Linux user-space and they are going systematically unreported.

Those who read Lennart's G+ post will have noticed certain other tropes, such as complaining about the "boycott systemd" campaign as though he were entitled to a total absence of criticism. Further there's the populist reference to "white men in their 30s and 40s" in the pejorative sense -- racism and sexism if I ever saw it -- and a pre-emptive refusal of further discussion in the very same article, like a seagull making its mark on a beach. Needless to say, this is not a recipe for a healthy bipartisan discussion; and from the content of this LWN article, Mr. Corbet isn't interested in having one either.

This kind of unquestioning pro-Lennart publicity will only fan the flames further. Mark my words: the Linux community does not take well to having technology dictated to it.

to post comments

On the sickness of our community

Posted Oct 16, 2014 16:01 UTC (Thu) by anselm (subscriber, #2796) [Link] (8 responses)

Cars don't stop on a dime just as there's no way to anger broad swathes of the Linux community without getting flamed to a crisp again and again.

s/broad swathes of/a tiny but vocal minority within/

There. Fixed it for you.

There are good arguments being made that systemd is Red Hat's hostile takeover of the Linux user-space and they are going systematically unreported.

Consider Debian. Debian is not particularly known for adopting sub-standard solutions and, in the absence of any form of leverage, is certainly not susceptible to a “hostile takeover” from Red Hat. Yet still the distribution has decided to go with systemd as their default low-level plumbing on Linux. The technical discussion that led to that decision is publically available for anyone to look at, and while there are still flame wars on some Debian mailing lists, so far there don't seem to be enough Debian developers who are sufficiently unhappy with the pro-systemd decision to start a GR to try and overturn it.

It is worth reiterating that by now the systemd development community includes people from a variety of distributions beside Red Hat, and there is no evidence that systemd and its future directions are especially dominated by Red Hat. If you make claims to the contrary then feel free to produce the allegedly “systematically unreported“ support for them.

On the sickness of our community

Posted Oct 16, 2014 16:25 UTC (Thu) by ksandstr (guest, #60862) [Link] (7 responses)

First, I don't see where you derive your argument that the anti-systemd side is a ``vocal minority'' from, nor do I see the relevance. Free Software is neither a popularity contest nor a shouting match: if it were, systemd would've won by sheer number of fanboys alone, and it hasn't as evidenced by holdouts such as Gentoo and Debian.

Secondly, there's no need for leverage over Debian when leverage exists with regard to the upstream packages which Debian distributes and on which all Linux distributions depend. These are components such as udev, udisks, upower, dbus, xorg, policykit, consolekit, the list goes on. As Theodore Y. Ts'o said, ``we have commit privs and you don't''. You'll note that versions of each that bring with them dependencies on Lennartware have been uploaded into unstable, and a reduction of support for systems that don't run either systemd proper in its most recent version, or systemd-shim (which is perpetually behind the curve).

For example the testing package of xfce4-power-manager has been unable to suspend or hibernate systems since June 2014; and cryptsetup's boot script has had significant trouble with boot-time password entry due to systemd influence in console input handling. These things worked before systemd came along, and now they do not. How much more damning could it get?

Moreover, systemd got into Debian during a time when it was marketed as ``just an init system''. You'll agree that it has expanded into taking over the roles of syslog, dhcpcd, pm-utils, network-manager, and that many other functions are still in the pipeline. This is not what Debian voted for in their GR.

Furthermore, the parties whose views are going systematically unreported are of course those that're outside systemd development. (sheesh.) Why would an opponent lend credence to a project s/he opposes, thus furthering its goals of having systemd in every Linux installation and VM instance everywhere? (As evidenced by Lennart's juvenile decrying of Gentoo as ``haters''.)

On the sickness of our community

Posted Oct 16, 2014 16:45 UTC (Thu) by pizza (subscriber, #46) [Link] (1 responses)

> For example the testing package of xfce4-power-manager has been unable to suspend or hibernate systems since June 2014; and cryptsetup's boot script has had significant trouble with boot-time password entry due to systemd influence in console input handling. These things worked before systemd came along, and now they do not. How much more damning could it get?

If that's the best you have, that's hardly "damning".

So you've identified two bugs/regressions in the *testing* packages; assuming they have been reported, I'd presume that would get fixed as part of Debian's standard release freeze cycle. Debian has blocked releases for much less.

Meanwhile, if you're unhappy with the quality or quantity of other people's work, you're free to contribute.

On the sickness of our community

Posted Oct 16, 2014 17:14 UTC (Thu) by ksandstr (guest, #60862) [Link]

Well pooh-pooh to you as well, sir, and good day.

On the sickness of our community

Posted Oct 16, 2014 17:41 UTC (Thu) by anselm (subscriber, #2796) [Link] (2 responses)

If there are really so many people who don't like systemd (to avoid more loaded terms), then how is it that all the major Linux distributions have managed to adopt systemd without there being either a massive exodus to distributions like Slackware or else an obvious and popular initiative to produce systemd-free forks of Fedora, openSUSE, or Debian? All of these distributions are freely available, and the “we have commit privs and you don't“ argument doesn't count when everyone is free to make a forked repository where they have commit privileges.

This suggests that people are quick to bitch and moan but not so quick when the time comes to act. If not having systemd on one's system is not important enough to one to actually do the legwork (which is really straightforward compared to many other free-software projects since the required bits and pieces already exist; it's not as if one would have to write sysvinit and a zillion init scripts from scratch) that puts into perspective their complaints of how they're getting screwed over by those people who are in fact prepared to spend their time working on stuff.

There are issues with systemd but there are also people within the systemd community and the various distributions who are committed to getting these issues fixed. In the long run, systemd can only get better at delivering what most Linux users will find useful. It is up to you whether you want to avail yourself of this or whether you prefer to pursue another approach, but please stop complaining that it is all a huge plot to prevent you from having other people work on your behalf, for free, in order to produce the Linux that you want.

On the sickness of our community

Posted Oct 16, 2014 21:46 UTC (Thu) by flussence (guest, #85566) [Link] (1 responses)

> If there are really so many people who don't like systemd (to avoid more loaded terms), then how is it that all the major Linux distributions have managed to adopt systemd without there being either a massive exodus to distributions like Slackware or else an obvious and popular initiative to produce systemd-free forks of Fedora, openSUSE, or Debian?

Where's your facts and figures to support that assertion? I see new recruits on the Gentoo forums every week who straight up cite systemd as their reason for switching. comp.sysutils.supervision.general is quite alive as of late too.

On the sickness of our community

Posted Oct 16, 2014 23:05 UTC (Thu) by anselm (subscriber, #2796) [Link]

Well, there are no visible projects aiming for systemd-free versions of the mainstream distributions. For example, for all the heated debate going on around systemd in Debian, no Debian developers have so far resigned in disgust after the decision for systemd. Similarly, it seems to be business as usual with all the other mainstream distributions.

Maybe some people are looking at Gentoo more closely now. If so, more power to them. However the idea that “large swathes” of the Linux community are really opposed to systemd to a point where they seriously consider switching distributions just to avoid it is probably wishful thinking on the part of those who don't like systemd.

FWIW, I teach Linux system administration for a living and thus get to meet rather a lot of Linux sysadmins of various backgrounds in my professional life. I have yet to run into one who didn't think systemd was a good idea and a considerable improvement on the status quo. For many people it comes as a bit of a shock at first but then it grows on them, and the more they find out about it the more they like it.

On the sickness of our community

Posted Oct 16, 2014 20:18 UTC (Thu) by mathstuf (subscriber, #69389) [Link] (1 responses)

> cryptsetup's boot script has had significant trouble with boot-time password entry due to systemd influence in console input handling

Works flawlessly for me in Fedora, so it's not impossible. Maybe there's a bug with Debian or the script?

> Moreover, systemd got into Debian during a time when it was marketed as ``just an init system''. You'll agree that it has expanded into taking over the roles of syslog, dhcpcd, pm-utils, network-manager, and that many other functions are still in the pipeline.

Of those you list, only dhcp (just the client side though IIRC, not dhcp*d*) and networking weren't in systemd already and they had (AFAIR) been announced by that time. My understanding is that for *simple* network solutions, systemd will suffice. If you need bridging, VPN, or anything complicated, use NetworkManager (or the old ifcfg scripts…which still work because I still use them).

On the sickness of our community

Posted Oct 20, 2014 16:40 UTC (Mon) by nye (subscriber, #51576) [Link]

>Works flawlessly for me in Fedora, so it's not impossible. Maybe there's a bug with Debian or the script?

Debian's version has some extra functionality (keyscripts) that's not supported. It could be reasonably argued that the way it's currently implemented is pretty ugly, although the flipside is that it could be considered simple and not over-engineered.

Last I checked (several weeks ago), it looked like it was going to be hard to find consensus on how to handle it, with discussions on it having petered out. Possibly it's just a case of the discussion having become less visible though.