Re: Is systemd a benefit or a liability? [Was: btrfs :(]
Russel Coker said,
I think that the problem here is users who are too arrogant.
It's simply not your decision what unpaid developers do with their time. We develop the software and you can use it if you like it.
Being a Debian developer is a volunteer exercise, one of many in todays society. No one holds a gun to our heads (I have been an office bearer in a fire brigade for years as well a working as a volunteer in a number of other enterprises) we do such work because we like to and such work brings other benefits.
I can't think of any way to reasonbly equate "justice" to "non-contributors getting to give orders to people who do the work".
Linux is developed by more people than just Debian Developers, I regularly submit well researched bug fixes to various projects, including the kernels framebuffer code (one of linux's more difficult areas to work in). Very much time is spent on making sure these are well valued contributions. DD's are not the only ones spending time in Linux code................. You may in future wish to get some evidence before being judge, jury and excutioner, and you call me arrogent........!!!! This is getting me no where, I am not surprised systemd is getting plenty of opposition as most of its proponents appear now to be incapable of giving a clear explanation of why they support it. I have better things to do with my development time than try and figure out a project with apparently no clear goals (as so far none has been given). Lindsay
On Tue, 14 Oct 2014, zlinw@mcmedia.com.au wrote:
Linux is developed by more people than just Debian Developers, I regularly submit well researched bug fixes to various projects, including the kernels framebuffer code (one of linux's more difficult areas to work in). Very much time is spent on making sure these are well valued contributions. DD's are not the only ones spending time in Linux code.................
Would you like to have a bunch of people who know nothing about framebuffer code start telling you what you should be doing? What if they start saying that such code should be the subject of a democratic vote among people who have never read it?
This is getting me no where, I am not surprised systemd is getting plenty of opposition as most of its proponents appear now to be incapable of giving a clear explanation of why they support it. I have better things to do with my development time than try and figure out a project with apparently no clear goals (as so far none has been given).
We had lots of long discussions when the issue was still up for debate, you can read them if you wish. Then when no consensus was reached the technical committee made a decision, you can read their analysis of the situation if you wish. This thread on luv-main was never about explanations of systemd. It was always about people who have decided against it without reason who think that they know more about the topic than people who use systemd and have read the code. Changing an init system is not such a big deal. We went from BSD style to SysV without any serious problems. Ubuntu went to Upstart and it worked reasonably well for them even though almost no-one in Debian used it. Solaris uses a very different init system that apparently works well enough for them. Apple has yet another different init system which apparently inspired some of the design decisions for systemd. Examples have been cited of systemd developers being difficult to work with. But this mailing list has given plenty of examples to show that anyone who wants to work on systemd must ignore unqualified people. Submitting framebuffer bug reports doesn't make you qualified to assess the Debian project's decision about init systems. Most people involved in this discussion could have spun up a virtual machine running systemd in less time than they spent arguing. Why not just test out systemd and see if it works for you? -- My Main Blog http://etbe.coker.com.au/ My Documents Blog http://doc.coker.com.au/
On 14/10/2014 1:25 AM, Russell Coker wrote:
Most people involved in this discussion could have spun up a virtual machine running systemd in less time than they spent arguing. Why not just test out systemd and see if it works for you?
There is enough known about systemd today, about it's current over creep and the plans of the systemd developers that I know that I want to avoid it altogether. A fact that with systemd, they consider it a "real operating system", what do we have without systemd? Did you want that video from the Perth LCA? That has some revelations.... it only goes for 9:31 ... of course it is up to you to decide if you want to watch -- I'm not going to command you, which would be pointless anyway. People have posted in DU list that they have had systemd installed due to an updates/upgrade process ... they DID NOT want it. I have absolutely no plan to /spin/ up a VM to decide if systemd is a problem or not because of the knowledge I already have; I do NOT want systemd. If systemd was ONLY an init system and nothing else, then many wouldn't be as concerned; most that seem to WANT or otherwise not care, don't accept the fact that systemd is much more than just an init system right now, let alone where the project is headed. If the main developer didn't screw up Pulse Audio for so many people .... if another developer didn't code so badly that Linus refuses his kernel patches ... if if if if .... I don't need any more ifs, I just want to distance myself and any and all systems that I maintain from systemd as much as possible. A.
On Tue, Oct 14, 2014 at 01:25:12AM +1100, Russell Coker wrote:
On Tue, 14 Oct 2014, zlinw@mcmedia.com.au wrote:
Linux is developed by more people than just Debian Developers, I regularly submit well researched bug fixes to various projects, including the kernels framebuffer code (one of linux's more difficult areas to work in). Very much time is spent on making sure these are well valued contributions. DD's are not the only ones spending time in Linux code.................
Would you like to have a bunch of people who know nothing about framebuffer code start telling you what you should be doing?
that's not exactly what's happening with complaints about systemd. a lot of highly skilled people with many years of experience have written very detailed technical complaints about systemd and the dangers of software monoculture that are ignored and patronisingly dismissed with "fuck off, we know better than you". At best, trivial benefits ("it boots faster", "you don't need to know about ulimit -c", "you can almost make it do something similar to what you want with some obscure undocumented systemdctl argument") are touted as if they're somehow worth the price of switching from a robust software ecosystem to a software monoculture. i used to call that the Gnome Attitude Problem[1], but it seems to be a common attitude with RedHat projects - which is exactly what systemd is, Redhat's weapon against Ubuntu which they'd lost huge market share to. Which is precisely why debian should have stayed out of it and not taken sides in redhat's commercial war with ubuntu....but what happened was that Debian took RH's side and within days Ubuntu announced their surrender. [1] try submitting a bug report to gnome if you want to experience the joy of being insulted and ridiculed for giving a damn. ditto for systemd.
What if they start saying that such code should be the subject of a democratic vote among people who have never read it?
you can't have it both ways. first you say that systemd won the vote[2] in debian, so that's reason enough for whingers to STFU. now you're saying that votes should play no part. [2] it won the tech committee's vote ONLY because of the chairman's tie-breaking vote. 50/50 is hardly a ringing endorsement, especially for a change as radical and controversial as this.
Changing an init system is not such a big deal.
as has been mentioned many times before, systemd is not just an init system. if it was just an init system, it wouldn't be a problem - it would be just another implementation of a component (i.e. init - sysvinit, upstart, openrc, systemd and others are current options) that can be replaced without too much fuss or bother. by merging in dozens of other functions, however, it makes it nearly impossible to replace systemd (or any part of it) once you've switched to it.
We went from BSD style to SysV without any serious problems.
if by "we" you mean "debian", i don't recall debian ever having a monolithic /etc/rc like freebsd. MCC and SLS and Slackware all had that, and i remember liking the flexibility offered by sysvinit when i switched to debian in 1994.
Ubuntu went to Upstart and it worked reasonably well for them even though almost no-one in Debian used it.
yes, switching init should be no big deal. it isn't, right now. that's because all-but-one of the current init systems aren't designed with lock-in in mind. switching away from systemd will prove to be a huge problem in the future. craig -- craig sanders <cas@taz.net.au>
On 14.10.14 08:53, Craig Sanders wrote:
At best, trivial benefits ("it boots faster", "you don't need to know about ulimit -c", "you can almost make it do something similar to what you want with some obscure undocumented systemdctl argument") are touted as if they're somehow worth the price of switching from a robust software ecosystem to a software monoculture.
Craig, that is the most powerful argument in favour of sysyemd I've been able to find on the net, and reveals why the developers don't claim any real benefits either.
i used to call that the Gnome Attitude Problem[1], but it seems to be a common attitude with RedHat projects - which is exactly what systemd is, Redhat's weapon against Ubuntu which they'd lost huge market share to. Which is precisely why debian should have stayed out of it and not taken sides in redhat's commercial war with ubuntu....but what happened was that Debian took RH's side and within days Ubuntu announced their surrender.
Something like that is what I could sense just from the unix-hostile sequestration of functionality into an obfuscated fuzz-ball. That systemd's motives are steeped in goals of commercial hegemony can now hardly be in dispute. The only question remaining is haw far we will have to go to get away from Systemdix. Erik -- (5) It is always possible to agglutinate multiple separate problems into a single complex interdependent solution. In most cases this is a bad idea. RFC-1925
participants (5)
-
Andrew McGlashan -
Craig Sanders -
Erik Christiansen -
Russell Coker -
zlinw＠mcmedia.com.au