If systemd has been adopted, how bad can it be?
|
| Author | Content |
|---|---|
| linuxscreenshot Nov 09, 2014 11:33 AM EDT |
As a long time Linux end user (not a Linux developer) I have three reasonable questions: 1) If most major distributions have adopted systemd, can it really be that bad? 2) These distros wouldn't have blindly adopted systemd. They must see systemd as having more benefits than risks. What are they? 3) If systemd is that bad, can we expect to see a major migration to non-systemd Slackware? |
| gus3 Nov 09, 2014 11:45 AM EDT |
1. If most Fortune 500 companies have adopted Windows, can it really be that bad? 2. These companies wouldn't have blindly adopted Windows. They must see Windows has having more benefits than risks. What are they? 3. If Windows is that bad, can we expect to see a major migration to non-Windows platforms? |
| theBeez Nov 09, 2014 1:09 PM EDT |
Good going, gus3. This highly overused piece of bad reasoning (the crowd must be right) just shows how few real arguments are left to use for the systemd fans. Can 20 million Elvis fans be wrong? Well, compared to Bach, yes. |
| MALsPa Nov 09, 2014 1:11 PM EDT |
I'm another end-user wondering if systemd is really all that bad. Second time recently I've seen similar replies to similar questions. Do those replies really answer the questions, though? And, fair analogies, or not? Can we really compare systemd to Windows, or major Linux distributions to Fortune 500 companies? |
| gus3 Nov 09, 2014 1:12 PM EDT |
Neither Elvis nor Bach did as much damage to the international economy that Microsoft has done. If I have to explain it... |
| linux4567 Nov 09, 2014 1:27 PM EDT |
The post by linuxscreenshot reads like a classic shill post. I wonder who is paying him/her to post propaganda for systemd? I wonder if linuxscreenshot is wearing a red hat... And for those genuinely wondering what's bad about systemd just read the main page of the following web site, all the reasons are listed there: http://boycottsystemd.org/ |
| gus3 Nov 09, 2014 1:33 PM EDT |
I just thought of an experiment that someone should do (could be me, could be anyone). Install Debian or Fedora (or RH, if available), then build and install SysV init. See how it goes. Install Slackware, then build and install systemd. See how it goes. I don't expect a fully running system in either case, but I would like to see which fails later. |
| mrider Nov 09, 2014 1:40 PM EDT |
@linuxscreenshot and @MALsPa - This has been hashed and rehashed on this site and elsewhere so many times that asking that question makes me wonder if you've just awakened from a coma or arrived in a time machine. Suffice to say that the problem isn't so much what SystemD is, nearly so much as the fact that it's positioned such that it's unavoidable. Basically the attitude of the SystemD programmers is "oh, you don't want SystemD - well Tough ****, you're getting it anyway!" At least when I decided that I didn't want any part of KDE 4x, I had the option of using another desktop manager (XFCE in my case). At least when I couldn't get Pulse Audio working, I could use Alsa instead. What if I have some legitimate reason to avoid SystemD? I guess the answer is "go somewhere else because we are no longer interested in you". |
| jdixon Nov 09, 2014 1:44 PM EDT |
> Can 20 million Elvis fans be wrong? Well, compared to Bach, yes. Largely apples and oranges. Elvis was a performer, not a composer. And if you could compare his voice to Bach's, I suspect he'd win the comparison. Of course, the systemd folks want to make the comparison to init an apples to oranges comparison too. They want systemd to be completely in control of everything on your Linux system, with no option for anything else. After all, who needs any other music besides the Monkee's? |
| jdixon Nov 09, 2014 1:50 PM EDT |
> What if I have some legitimate reason to avoid SystemD? I guess the answer is "go somewhere else because we are no longer interested in you". I think Red Hat seriously underestimates the number of people who will do exactly that. Of course, they're not Red Hat customers, so I doubt they care. |
| gus3 Nov 09, 2014 1:53 PM EDT |
Red Hat is concerned with what users will do, only insofar as it affects their profit margins. By the same token, Red Hat is not entirely dismissive about what users will do. |
| linuxscreenshot Nov 09, 2014 2:04 PM EDT |
Can someone please answer my questions instead of assuming I'm somehow conspiring? As an end user, I just want a distro that is stable with an easy to user interface. Should it really matter to 'me' if a distro uses systemd or sysv? I assume that if systemd is as bad as some argue that there would be a mass exodus by developers to Slackware. Maybe behind the scenes there is.. Wouldn't this be a good thing? |
| linuxscreenshot Nov 09, 2014 2:21 PM EDT |
Quoting:The post by linuxscreenshot reads like a classical shill post. I wonder who is paying him/her to post propaganda for systemd? Can we relax a bit... I'm just asking a question. |
| linuxscreenshot Nov 09, 2014 2:26 PM EDT |
Quoting:At least when I decided that I didn't want any part of KDE 4x, I had the option of using another desktop manager (XFCE in my case). At least when I couldn't get Pulse Audio working, I could use Alsa instead. Can't we just use a non-systemd distro, if we're bothered by this new direction? |
| Koriel Nov 09, 2014 2:33 PM EDT |
1) The reason it has been adopted is because certain major upstream distro's adopted it which means it becomes far too much work for the downstream distro's to remove it and maintain their distro's (fork) It is nothing to do with good or bad simply workload. The only distro's that don't have this problem currently are mostly downstream distro's based on Slackware ie, Salix, Zenwalk etc as Slackware being the upstream has not adopted it. 2) They didn't blindly adopt it, their is little or no choice its either accept its there or take on the workload of forking some might be able to do this but most won't and thus you are pretty much screwed its pretty much accept it and be damned, I see this as a fait accompli and not how things should be done in the Linux world. 3) Will we see a migration now that is the real question, I seriously doubt it will be en masse but some will certainly, those who learnt their Linux on Slackware like myself might but most wont as again their is little choice. So from these answers you can see it comes down to lack of choice, not good or bad and just incase folks think im biased i'm actually on the wait and see how it pans out fence regarding Systemd as I have no skin in the game its the way its being implemented thats wrong and has zero to do with technical merit. So there you go. |
| mrider Nov 09, 2014 2:39 PM EDT |
Quoting:Can't we just use a non-systemd distro, if we're bothered by this new direction?I assume you mean aside from the fact that Gnome now depends on SystemD? Or that GIMP now depends on SystemD? Or that SystemD is insinuating itself into so many places that it becomes a dependency for seemingly unrelated software? I'm quoting penguinist from this thread -> http://lxer.com/module/forums/t/35620/ Quoting:I think I just ran into a specific case of "systemd cancer". Finally, did you read boycott SystemD? |
| 750 Nov 09, 2014 2:53 PM EDT |
The basic issue that got me going against systemd was the tight coupling between its sub-projects and systemd as init. This when i learned that Gnome's display manager (GDM) was to require logind, a part of systemd, and a part that can't function (as best i can tell) without systemd running as init. End result is that Gnome has set the uncomfortable precedence of producing a desktop environment that requires a specific init. And with the kinds of tentacles systemd is sprouting (networking, firewall, timed tasks), it is only a question of time before more projects develop such a dependency. The musings of conspiracy comes along when one consider that both Gnome and systemd are Red Hat related projects. Also, systemd has sprouted a whole lot of cloud computing "features" lately. And RH is throwing its weight around to get into the cloud computing business. End result is that other distros are likely adopting systemd in part to stay compatible with RH's cloud. |
| linuxscreenshot Nov 09, 2014 3:03 PM EDT |
Quoting:I assume you mean aside from the fact that Gnome now depends on SystemD? Or that GIMP now depends on SystemD? Or that SystemD is insinuating itself into so many places that it becomes a dependency for seemingly unrelated software? Thanks for that clarification and pointing out http://boycottsystemd.org/, which I've now read. I guess switching to Slackware doesn't solve the problem if the issue is extended to software packages.. hmm. Has there been any official responses by the major distribution communities in regard to the systemd opposition? Why did the major distros so quickly adopt something that was so controversial - this doesn't see very Linux-like? |
| gus3 Nov 09, 2014 3:12 PM EDT |
@Koriel, what you do mean by "little to no choice"? For Red Hat, it's a case of "eating your own dog food," but I don't see the same case with Debian or OpenSUSE. There is still a choice, as evidenced by Slackware et al. |
| Koriel Nov 09, 2014 3:22 PM EDT |
@gus3 Agreed in the case of Debian, I can't honestly see why Debian chose the path they are on, they could of easily adopted a wait and see approach and while they were waiting explore other alternatives. Im not denigrating SystemD or anything it deserves a chance at proving itself but not by inflicting it on folks like Pulseaudio was and that turned out be a complete dogs dinner. |
| Steven_Rosenber Nov 09, 2014 4:37 PM EDT |
Slackware hasn't shipped GNOME for many years, and I expect that most applications can be packaged without a systemd dependency, so if there is a will to stay away from systemd, there is certainly a way. |
| jdixon Nov 09, 2014 4:40 PM EDT |
> As an end user, I just want a distro that is stable with an easy to user interface. To all appearances, systemd is designed to give the latter. The former? Well, let's just say that Poettering's past history in that regard isn't comforting. > Can't we just use a non-systemd distro, if we're bothered by this new direction? Ah, that's the rub, isn't it? The answer seems to be that unless you use Slackware or a derivative, no. And how long Slackware holds out remains an open question. Of course, I'm not sure I'd put it past Patrick to wash his hands of the whole affair and move Slackware to a BSD base. |
| linuxscreenshot Nov 09, 2014 4:45 PM EDT |
Quoting:Slackware hasn't shipped GNOME for many years, and I expect that most applications can be packaged without a systemd dependency, so if there is a will to stay away from systemd, there is certainly a way. I guess these Slackware based distros should see an increase, or may they include systemd? |
| Ridcully Nov 09, 2014 4:48 PM EDT |
Has Torvalds commented ? |
| JaseP Nov 09, 2014 4:59 PM EDT |
I don't think Torvalds really cares one way or another. The thing about Linus is that unless a userspace issue affects him directly (be it his being careful not to introduce user space bugs with the kernel space, or in his own personal usage of Linux) he doesn't seem to comment or be concerned. He's paid handsomely to deal with the kernel, and I've noticed his focus becoming more and more limited. With Hess's departure from Debian, it kind of helps answer the 3 questions above,... Doesn't it?!?! This is part of what I thought would happen,... A major shifting of old-time, core developers,... Their projects orphaned and/or replaced with so-called "improvements"... And,... A major shift toward younger programmers, who aren't as beholden to older standards coming to "prominence" as lead developers (after all Poettering has expressed a deference for The Linux Programming Interface, by Kerrisk, except for the "POSIX parts"). |
| gus3 Nov 09, 2014 5:10 PM EDT |
Torvalds has commented on the poor dev-management process behind systemd, in the form of his rant against Kay Sievers. It's NSFW, so I won't link it here, but it's easy enough to find with a Google search. |
| JaseP Nov 09, 2014 5:48 PM EDT |
@ gus3: Torvalds commented about the Kay Sievers thing, because he (Kay) refused to acknowledge reported bugs which impacted/broke user space. But as far as the systemd thing, in general, he's (Linus) really backed away from any provocative statements. Seeing as though it is totally Linus's style to deal with a problem head on, and provocatively, if need be,... Torvalds isn't too concerned with what's going on in systemd,... I think the one video where he was asked about it, he shrugged, and said something to the tune that he didn't hate systemd. I don't think Torvalds would be too concerned with what happens in the Debian development world, either. So, one core developer leaving the project isn't likely to be much of a blip on his radar... If this causes a major shake-up, with lead kernel developers being impacted,... THEN I think he'd stand up and take notice. |
| mxc Nov 09, 2014 6:39 PM EDT |
To answer your questions as to why you should care as an end user there are two parts. Firstly as an end user systemd or syst v make no difference to you. You weren't aware of sys v and you only aware of systemd because of its affect on the community. I don;t recall any end user ever complaining about start up scripts for example. From a developer point of view systemd makes some things much easier and "fixes" some issues with sys v and the init process. Fixing these issue will eventually result in a better user experience. If this is the case how come people are upset? This brings me to my second point why you should care. Systemd is abandoning the Unix design philosophy. How you feel about this depends on your experience as a developer and system builder or if you someone who is more a power user that prefers ease-of-use over flexibility and the ability to cobble components together to build something different than what the original creators ever envisaged. Yes, sys v has problems and needs to be replaced. Systemd has show the way in how to deal with some of these issues but systemd is like the one ring that is trying to bring all services and components under its control and in the darkness bind them. It is absorbing more and more functionality, making core system components dependent on it. In the end you will be running a systemd distribution not a Linux distribution. Why is it a problem to abandon a tried and tested design pattern? Usually these patterns have evolved for a reason, from years of experience and stood the test of time. When they are abandoned its not immediately obvious what the implication are. Often things seem better but as time progress one soon relearns the lessons that lead to the adoption of the approach in the first place. This is my main concern with systemd. It is over-reaching, moving swiftly into uncharted territory and forcing everyone to get on board the same risky enterprise. One only need to compre the cautious approach to the adoption of BTRFS to systemd. BTRFS is even cooler than systemd but its not being rammed down everyone throats. The sound bites from the "technorati" prodcasts is that its not ready for prime time - but some how systemd is? If someone wants to try something out go for it - don't force me to come along for the ride. |
| jdixon Nov 09, 2014 7:09 PM EDT |
> Yes, sys v has problems and needs to be replaced. You don't necessarily have to replace something to fix it's problems. |
| jazz Nov 09, 2014 7:46 PM EDT |
> 3) If systemd is that bad, can we expect to see a major migration to non-systemd Slackware? Debian 7 still has 3.5 years left, Ubuntu 14.04 about 4.5 years, the next Debian will support SysV init for 5 more years, and so on. Non-systemd distros are here to stay, I don't really see a reason to move to Slackware (or Gentoo), or to rush to learn systemd. BTW, I agree with your statement above, systemd is that bad! |
| Ridcully Nov 09, 2014 9:07 PM EDT |
@mxc.......Okay, I hear you loud and clear. You have said something that I also said earlier on another thread, in that as an end user with no real need to "fiddle or program", it doesn't really matter whether my OS uses SysVinit or systemd, as long as it works without any bugs.....BUT...and it is a very big "BUT", I seriously DO care about whether or not the principles under which Unix and Linux operate have been compromised. From everything I am reading, it appears to be the case that they most definitely have been compromised by systemd. Open-ness and the ability for the administrator to employ their own settings seem to have been either reduced or discarded, and the constant unpleasant creep/crawl by systemd to take over the functions of previously separate software modules is most definitely a retrograde move in my books. Let me repeat: To me as a very simple user, it's not going to make any difference - I think. But, there are many others who can see the unfortunate results of what systemd is doing to previously "open" Linux software. I'm going to coin a very clumsy word, but it's the only way to put it: From my perspective, systemd "smells" very like a "Windowsification" of a part of the Linux software suite. |
| mxc Nov 10, 2014 1:56 AM EDT |
@Ridcully - yeah. I agree 100%. Also I do think there is a lot of politics going on here as alluded to by others. One can't help but feel there is a definite marketing campaign to sell systemd to distros and the community. This is odd because usually a project comes out - people start using it and slowly it gets adoption and then get de-facto standard due to wide spread, voluntary use. Take apache, nginx, or something like unity or waylaynd. There is no attempt to force distros to use either one. Just projects competing. Systemd continues to rip out old but flexible components like getty, dhcp, cron etc. Theses things do need improvement but very often features which seem no longer to be needed like a virtual terminal, serial port support etc can be used in unexpected ways later. I suspect most of these systemd people use Apple Mac and probably think that we don't need ethernet port support anymore but need to support some connector on the apple devices. |
| jezuch Nov 10, 2014 3:39 AM EDT |
Quoting:1. If most Fortune 500 companies have adopted Windows, can it really be that bad? That's the wrong analogy. Windows and Fotune 500 companies are more akin to the sysvinit situation: an entrenched, inferior solution everybody hates but had no way to get rid of... until now. Remember, before systemd came along and you started crying about monopolies and monocultures, sysvinit was the monopoly and monocolture. Why not cry about THAT? Unfortunately it's much easier to switch init systems than to switch operating systems across a multinational megacorporation. [I usually try to stay away from this hate-fest but I couldn't resist this time. Oh, and this hate-fest is why Joey resigned. Not because of systemd but because the irrational paranoia.] |
| Ridcully Nov 10, 2014 4:51 AM EDT |
Quoting: Remember, before systemd came along and you started crying about monopolies and monocultures, sysvinit was the monopoly and monocolture (sic). Why not cry about THAT? I may not be an active programmer and administrator in today's world, but even I can see how the above is awfully "shaky". I would be delighted if wiser heads than mine would please respond. |
| jdixon Nov 10, 2014 5:14 AM EDT |
> Remember, before systemd came along and you started crying about monopolies and monocultures, sysvinit was the monopoly and monocolture. Why not cry about THAT? Because sysvinit was flexible enough to allow things like the Slackware init scripts. It started your system and left the rest up to you, in keeping with the Unix philosophy. |
| jezuch Nov 10, 2014 5:30 AM EDT |
I realize that my tone might have a little been too aggressive and I'm sorry about that (and about a typo or two that slipped by ;) ). But I really don't get this vicious opposition, and it's sad to see ostensibly rational people behave like obvious trolls spreading myths and not bothering to learn about the thing they criticize. The thing is that despite what is claimed ("Red Hat wants to destroy Debian!" and other silly conspiracy theories) what Lennart et al. want to do is create a modern init system that Linux deserves. I'm actually astonished that it took so long for anyone to do this. It's not about taking toys away from graybeards, or about the "Unix Way" (this is a straw man anyway), or anything like that; it's about being in the 21st century, having all those marvelous features that Linux offers and actually using them. That they also include other things than dependency management is because it makes sense for these things to be integrated together. That it's not "portable" is a natural consequence of this one basic assumption that it's all about Linux. That it doesn't support some weird use case one person is using is not a matter of attitude of its creators - it's a matter of them not knowing about this weird use case (and of this use case being weird). It's new so it can't possibly satisfy everybody (yet). But it's open source and the solution is to work with the upstream - OR to create something better (or with a better maintainer if you relly don't like Lennart). That's all there is to it. IMVHO. |
| jezuch Nov 10, 2014 5:35 AM EDT |
Quoting:Because sysvinit was flexible enough to allow things like the Slackware init scripts. It started your system and left the rest up to you, in keeping with the Unix philosophy. So, sysvinit-using distros have to build around that, the result being an ad hoc, informally-specified, bug-ridden, slow implementation of half of systemd ;) (With apologies to https://en.wikipedia.org/wiki/Greenspun %27s_tenth_rule) |
| Ridcully Nov 10, 2014 6:38 AM EDT |
Quoting:what Lennart et al. want to do is create a modern init system that Linux deserves. It seems to me, jezuch, that all they had to do was update sysVinit......and keep its consistency with the open principles of Unix and Linux. systemd appears to do the exact opposite......and continues its crawl to absorb other Linux modules that work quite well on their own. But what would I know ? |
| jdixon Nov 10, 2014 9:11 AM EDT |
> ... what Lennart et al. want to do is create a modern init system that Linux deserves. Uhm, no. What they want to do is build a modern central control structure for Linux, not just an init system. > ...the result being an ad hoc, informally-specified, bug-ridden, slow implementation of half of systemd Slow? My Ubuntu 8.04 system booted in about 30 seconds. Bug-ridden? Pot, meet kettle. And quite a bit less than half at this point. |
| Bob_Robertson Nov 10, 2014 9:52 AM EDT |
I don't think anyone is nay-saying the _init_ function of systemd. Everything else systemd is trying to do, yes. I sneer at taking all those small, WELL KNOWN, thoroughly thrashed-upon subsystems and subsuming them into a huge, bloated, annoying package, to be a serious regression. A mistake. Wrong. I consider the lack of plain text log files to be a deal-killer all by itself. |
| CFWhitman Nov 10, 2014 12:50 PM EDT |
@jezuch SysV init was not a monopoly or a monoculture. Slackware uses BSD init and always has. Gentoo uses a modified init called OpenRC. Ubuntu uses Upstart. KahelOS was using DEMONS, which is a modification of SysV init. Pardus uses Mudur. GoboLinux uses BootScripts. The embedded system OpenWRT uses procd. Also available for Linux are eINIT, Initng, runit, s6, Epoch, and for embedded systems, busybox-init. That's probably not all of them. This is also evidence that initialization isn't an area where no development efforts have been going on. The objections to systemd are mostly not about its merits as an init system. They are mostly about its extension to other areas which has lead to odd dependencies on systemd that other init systems have always managed to avoid. It's great that systemd takes advantage of Linux features to create a more robust initialization. It's not great that it attempts to lock out all competing initialization schemes by extending into places it doesn't belong. You may think it makes sense to integrate all these systems together, but 40 years of Unix history seem to indicate otherwise. |
| gus3 Nov 10, 2014 1:52 PM EDT |
jdixon wrote:a modern central control structure for LinuxYeah, we know how easily "central control" turns into a single point of failure. |
| jdixon Nov 10, 2014 2:34 PM EDT |
> Yeah, we know how easily "central control" turns into a single point of failure. Well, we do. I'm not so sure either Poettering or current Red Hat management does. I can see Whitehurst now, surveying the company he's taken over and going "Is this any way to run an airline?" |
| linuxscreenshot Nov 10, 2014 4:46 PM EDT |
Interesting read about boot times on DistroWatch. Two weeks ago I made some statements with regards to systemd and SysV init. Specifically, I tested how quickly an installation of Debian Testing "Jessie" would boot using systemd and then compared that boot time against Debian booting using SysV init scripts. I also timed Arch Linux booting using systemd for comparison with Debian booting with systemd. In my tests I found, in environments as identical as I could make them, it did not matter which init technology was used, Debian would boot to a graphical login screen in 30 seconds running either SysV or systemd. Arch Linux, running systemd, booted to a graphical login screen in 40 seconds. As I quoted in my previous post, both Debian's team and Fedora's claim systemd is faster at booting a Linux distribution than SysV or Upstart. The idea that systemd is faster than other init technologies has become so ingrained in open source culture that it is routinely considered common knowledge. Following my limited testing of systemd and SysV, someone e-mailed me to ask if I would duplicate the tests, this time booting to a command line environment rather than a graphical login screen. I was happy to do so. Once again, I ran Debian Testing with systemd and SysV init systems. Using either init system the boot times varied by a second in either direction from one boot to the next. When the times were averaged out, both init systems brought Debian to a text console prompt in 17 seconds. For comparison, I booted Arch Linux with systemd three times. To get to a text console login prompt on Arch took an average of 22 seconds. |
| albinard Nov 10, 2014 4:57 PM EDT |
Does anyone on this or the other systemd thread actually KNOW anything about it, or are you all just complaining about change? |
| CFWhitman Nov 10, 2014 5:17 PM EDT |
Really? At this late date we get the old claim that people criticizing systemd are complaining about change? That was nonsense from the very beginning. Things change in Linux all the time. Anyone who uses Linux is not very likely to be the type that just blindly complains about change. It really sounds like you haven't read the thread at all. A good percentage of systemd's critics think it's a fine addition as an init system. That's not the problem with systemd. To sum it up, systemd smells like lock-in. A lot of Linux users have come to Linux because they don't like lock-in tactics. That's why systemd gets so much criticism. |
| albinard Nov 10, 2014 5:41 PM EDT |
I seem to recall that there was a lot of griping long ago about packages - hey, you should have to compile everything from source, it wasn't proper to have things so easy. I rest my case. Permanently. |
| Bob_Robertson Nov 10, 2014 5:53 PM EDT |
> At this late date we get the old claim that people criticizing systemd are complaining about change? So it seems. |
| frankiej Nov 10, 2014 6:21 PM EDT |
Quoting:Systemd continues to rip out old but flexible components like getty, dhcp, cron etc. While I cannot comment on the others, I have been reading about systemd timers and cron and from what I can tell, it isn't a forced replacement, both can run in parallel. systemd timers provide a wider array of options for scheduling for those that need them, or ties into other systemd features, but cron isn't pushed out of the picture by any means for anyone that wants to continue using cron. While I personally don't have a need for said features, I can picture use cases where they would be beneficial. |
| mrider Nov 10, 2014 6:41 PM EDT |
Quoting:I seem to recall that there was a lot of griping long ago about packages - hey, you should have to compile everything from source, it wasn't proper to have things so easy. And those people have the choice to use Gentoo. So what's your point? |
| CFWhitman Nov 10, 2014 7:01 PM EDT |
albinard wrote:I seem to recall that there was a lot of griping long ago about packages - hey, you should have to compile everything from source, it wasn't proper to have things so easy. So you say something completely irrelevant, and then, of course, you rest your case. You contributed zero the discussion. You made absolutely no counterpoints to the criticisms leveled at systemd. You didn't even address any of them made here or elsewhere either to refute them or to explain why they were acceptable trade offs. The funny part is, though I've never heard of a large number of users objecting to binary packages, and it's a completely apples to oranges comparison to the discussions about systemd, it doesn't matter because users who want to compile everything still can even now. The real question is whether or not people who want to run certain software packages will still be able to do so without running systemd as their init system. The thing is, I'm developing a wait and see attitude toward systemd. I think that it's got some qualities worth keeping. I just would hope that it would not end up the sprawling mess it seems to be moving toward and users would still have the opportunity to use the myriad other init systems that exist without giving up a bunch of desktop software to do it. The proponents of systemd make me look like I'm entirely against it, when that's not really the case. I just have my reservations. Proponents of systemd seem to think we should all just accept everything about it with no reservations. They want us to believe there are no drawbacks. That can't possibly be the case, and that attitude tends to make people wary. |
| 750 Nov 10, 2014 8:53 PM EDT |
Is it just me or has the line break in here gotten borked? |
| BernardSwiss Nov 10, 2014 10:48 PM EDT |
It's not you. I first noticed it about the time of Ridcully's Nov 09, 2014 post. What's causing it is anybody's guess. |
| Ridcully Nov 11, 2014 1:39 AM EDT |
It ain't me......well, not deliberately anyway......I wouldn't know where to start. |
| kikinovak Nov 11, 2014 8:19 AM EDT |
"The objections to systemd are mostly not about its merits as an init system. They are mostly about its extension to other areas which has lead to odd dependencies on systemd that other init systems have always managed to avoid. It's great that systemd takes advantage of Linux features to create a more robust initialization. It's not great that it attempts to lock out all competing initialization schemes by extending into places it doesn't belong. You may think it makes sense to integrate all these systems together, but 40 years of Unix history seem to indicate otherwise." This statement by CFWhitman above must be the most spot-on comment I've ever read in the long systemd debate. +1 on that. |
| jdixon Nov 11, 2014 9:30 AM EDT |
> ...hey, you should have to compile everything from source, it wasn't proper to have things so easy. You say that as if it's not a reasonable option. Gentoo and (to a lesser extent) Slackware would disagree with you. > I rest my case. Permanently. Promises, promises. |
| nmset Nov 11, 2014 9:51 AM EDT |
>extension to other areas This does not prevent one from using dhcp, cron, syslog-ng, NetworkManager, pm-utils... Also, if other libraries like GNOME, GIMP and what else link to systemd, to query said functionality/feature/property's status, we can't blame systemd for that, it's the choice of those library maintainers. The one thing I really don't like about systemd is the binary logging. Next is compelling reboot if systemd is updated, although I don't know if sysvinit update meant a mandatory reboot. On the other hand, sysadmins would appreciate that on any distro they deal with, there's a common denominator, every where they'll find systemctl for example, instead of distro specific utilities to manage services. |
| Bob_Robertson Nov 11, 2014 10:26 AM EDT |
@750, Page width problems like this are caused by over-length URLs and other inclusions that statically set the text window width, which then gets over-written by the right-hand margin. It happens all the time, when people include long URLs, graphics, and other things that cannot be wrapped. |
| vainrveenr Nov 11, 2014 11:26 AM EDT |
Quoting:"The objections to systemd are mostly not about its merits as an init system. They are mostly about its extension to other areas which has lead to odd dependencies on systemd that other init systems have always managed to avoid. It's great that systemd takes advantage of Linux features to create a more robust initialization. It's not great that it attempts to lock out all competing initialization schemes by extending into places it doesn't belong. You may think it makes sense to integrate all these systems together, but 40 years of Unix history seem to indicate otherwise." Also, do refer to similar "spot-on" comments within the long LXer thread on systemd, 'Systemd is like a cancer metast[asi]zing all over Linux', found at http://lxer.com/module/forums/t/35620/. |
| 750 Nov 11, 2014 12:33 PM EDT |
@jdixon compiling from source is also what allows Gobolinux to exist (tho it does have the option of pre-compiled packages as well). |
| gus3 Nov 11, 2014 12:39 PM EDT |
Edmund Burke wrote:Those who do not know history are doomed to repeat it. Henry Spencer wrote:Those who don't understand Unix are condemned to reinvent it, poorly. |
| jdixon Nov 11, 2014 12:41 PM EDT |
> ...compiling from source is also what allows Gobolinux to exist There are a number of source based distributions and distributions which allow or encourage compiling from source. I only mentioned the two I knew the most about. A complete list of them would be useful for those looking for a systemd exit strategy though. |
| tuppp Nov 11, 2014 6:06 PM EDT |
nmset wrote:This does not prevent one from using dhcp, cron, syslog-ng, NetworkManager, pm-utils... Perhaps, but one is still required to use systemd. nmset wrote:Also, if other libraries like GNOME, GIMP and what else link to systemd, to query said functionality/feature/property's status, we can't blame systemd for that, it's the choice of those library maintainers. We most certainly CAN blame systemd! -- It is the systemd devs who push for such unnecessary upstream dependencies! Here's Lennart himself in bed with the Gnome devs, cooing for Gnome dependency on systemd. nmset wrote:The one thing I really don't like about systemd is the binary logging. Oh, there are a lot more things not to like about systemd! nmset wrote:On the other hand, sysadmins would appreciate that on any distro they deal with, there's a common denominator, every where they'll find systemctl for example, instead of distro specific utilities to manage services. On the contrary, it is the sysadmins who most fervently oppose systemd on Debian. Only systemd supporters want such forced "universality." |
| gus3 Nov 11, 2014 6:24 PM EDT |
@tuppp, Federico's reply to Lennart's posting has a telling insult near the end. It pretty well sums up devs' attitudes towards the consumers of their code (users & admins). |
| Ridcully Nov 11, 2014 6:45 PM EDT |
I've been watching this and a previous thread. There seems to be very, very widespread condemnation of systemd....apart from the RedHat coterie of course. So, my next questions are: What is everybody going to do about it ? Can anything be done about it ? The code appears to fly in the face of accepted Unix/Linux practice and Joey's departure seems to spell out his concerns. Short of waving placards in the street at RedHat, what can be done about this unhealthy bit of code ? |
| gus3 Nov 11, 2014 7:29 PM EDT |
@Ridcully: You can have my Slackware when you pry it from my cold, dead fingers. Or when PatV switches to systemd. |
| Ridcully Nov 11, 2014 7:58 PM EDT |
@gus3.....unfair......foul.......send 'im off ref......Would you KINDLY stop making me laugh when I am trying to be serious. LOL . Anyway, well said gus3......although I have never used Slackware......wouldn't know where to start. Looks like I'd better keep this installation of openSUSE 11.4 running for as long as I can. |
| jdixon Nov 11, 2014 9:39 PM EDT |
> What is everybody going to do about it ? Can anything be done about it ? At the moment it looks to me like about all you can do is use a source based distro, use Slackware and let Patrick now that switching to systemd isn't what you want, or switch to one of the BSD's. If Debian does fork, that's also an option. I guess you could also just keep using your current non-systemd based distribution until it's no longer safe to do so, of course. |
| jdixon Nov 11, 2014 9:42 PM EDT |
> ...although I have never used Slackware... Come on in, the water's fine. :) |
| frankiej Nov 11, 2014 10:03 PM EDT |
Quoting: The code appears to fly in the face of accepted Unix/Linux practice and Joey's departure seems to spell out his concerns. I finally read the original article, Joey's farewell post, and checked out his blog. No where in there could I detect any dislike of systemd itself. Quite the contrary, he seemed to have a positive comment about the timers aspect per his blog post. Maybe I am missing something here, but perhaps it is just the heated discussions as jezuch suggested in an earlier post. |
| CFWhitman Nov 12, 2014 9:48 AM EDT |
If you'd like to try out Slackware, but are afraid that manual dependency management will be a pain, then you might try Salix or one of the other Slackware derivatives. The only thing I really miss when using a Slackware derivative is the selection of pre-packaged software found in a Debian based distribution thanks to extensive repositories and the commonness of downloadable deb files. However, if you actually want to compile software, Slackware has about the most predictable and standard environment in which to do so. I believe that Joey Hess quit Debian because he felt that the decision to go to systemd had already been made and the way the organization is set up is allowing that decision to be undermined and put to another vote. In other words, if anything, Mr. Hess is for systemd and doesn't like that the decision may not be completely settled at this point. How much his being for or against systemd is affecting his perception of shortcomings in the organization is up for debate, but according to him he's quitting because of Debian politics and not because systemd will or will not be used. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!