Earlier this week, KDE developer David Edmundson described in his blog how KDE would be tied to logind and timedated but not systemd itself, at least according to his claim that "The init system is one part of systemd that doesn't affect us at all, and any other could be used.".
Later, in the blog comments, he clarifies that starting with plasma 5.5, in 6 months, they'll drop "legacy" support, according to a decision taken in the plasma sprint.
Even if one can only guess why there is no formal announcement, it seems clear - unless somehow there is a shim or emulator, not only for logind but also for timedated, in 6 months KDE will be unusable unless you are running systemd. And the blog entry makes it clear that the plan is to remove more and more functionality from KDE and use systemd instead.
(Score: 5, Insightful) by Gaaark on Friday February 20 2015, @01:24PM
Since when did a desktop environment start depending on 'boot' systems?
this is an actual question: not a troll, but a curious person.
--- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
(Score: 2) by Gaaark on Friday February 20 2015, @01:27PM
Sorry... just rtfa....
[gives head a shake]
Still seems odd...
--- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
(Score: 4, Insightful) by Nerdfest on Friday February 20 2015, @02:31PM
It's really too bad., I tried out the alpha of the upcoming Xubuntu release and I think it's hands-down the prettiest desktop environment I've ever seen. I'll probably stick with it, but I know others won't because of systemd. People really don't seem to understand about proper component decoupling (again, assuming a shim layer is not being implemented that will convert time and date API calls to the underlying implementations). What the hell is going on these days? Like open protocols, loosely coupled designs are one of the big reasons things can move forward so fast when improving things.
(Score: 3, Funny) by GeminiDomino on Friday February 20 2015, @03:30PM
Has XFCE moved to require the systemd suite too, or is it just the underlying Ubuntu system?
"We've been attacked by the intelligent, educated segment of our culture"
(Score: 5, Informative) by arashi no garou on Friday February 20 2015, @03:49PM
Just Ubuntu; Xfce hasn't had a major release in three years so I doubt it will be adopting anything systemd related any time soon. And the subject of the article and summary suggests that either Slackware will be forced to adopt systemd to keep KDE, or else Pat will be forced to abandon KDE (I believe he still uses it daily). A lot of Slackware users use KDE, so this stands to affect the entire distro, its maintainers, and its users.
This fucking sucks, and I don't even use KDE myself.
(Score: 4, Insightful) by melikamp on Friday February 20 2015, @04:47PM
Slackware does not use the bleeding edge KDE, so it will be years before PV's hand is forced. During these years the cool heads may well prevail and hack KDE back into interoperability. Do not panic, keep calm and carry on, etc. etc. etc.
(Score: 2) by nightsky30 on Friday February 20 2015, @05:36PM
Can we fork KDE and call it FKDE? The "F" can stand for Fork whatever you think is appropriate.
(Score: -1, Troll) by Anonymous Coward on Friday February 20 2015, @06:52PM
Call it "F**k you". Where the two stars stand for "or", of course. ;-)
(Score: 0) by Anonymous Coward on Friday February 20 2015, @11:35PM
Maybe just rename KDE, SDDE (Systemd Desktop Environment) since they decided to allow him to do the forking.
(Score: 4, Interesting) by arashi no garou on Friday February 20 2015, @05:41PM
During these years the cool heads may well prevail and hack KDE back into interoperability.
The same thing was said about systemd not long before it was forced on the vast majority of GNU/Linux users. I'm not giving up on Slackware, I'm just acknowledging that Pat and the team will likely have a huge burden on their hands soon enough, no matter which way it goes (dumping KDE altogether, heavily patching it to work around systemd, or caving and accepting systemd and all that that entails).
And while Slackware releases don't track the bleeding edge, often -current does. I just don't want to see Pat suddenly stop development on the OS as a whole because a hugely important project gets infected with a malaise he's otherwise been able to avoid. He shouldn't have to commit already limited resources to fixing a problem that wasn't broken to start with. Why a DE of all things needs systemd hooks boggles the mind. It makes me glad I use a pretty bare WM in Slackware.
(Score: 5, Insightful) by melikamp on Friday February 20 2015, @07:27PM
Forced by whom? I am neither a systemd supporter nor an LP or KS sympathizer, but the nature of the debate forces me to take their side on some occasions. For what you are saying, the distributions are at fault. The "cabal" cannot twist anyone's hand outside of the Red Hat universe. Debian users in particular are stuck with systemd because of decisions made by their own committee. No one was forcing anyone to adapt systemd, let alone adapt it so soon. As Slackware users, me and you know full well that nothing of practical significance depends on it as of yet.
I agree with some of this, but not all. I agree there is a risk that Slackware will suffer as a result of this shuffle. While I don't use KWin, I understand that losing it is unacceptable for great many users. I also agree that replacing the current init by systemd may be too taxing for the Slackware development team.
But I strongly disagree with "infected" and the rest of the value-laden BS that flies thick in all of these discussions. From the technical standpoint, there is nothing seriously wrong with systemd, meaning that whatever faults it has today can be fixed tomorrow. (For example, binary logs are stupid, but there is nothing preventing admins from keeping text logs as well.) In fact, I see nothing of value in the boilerplate bash code used in Slackware init, so IMHO here's something systemd got right. The "cabal"'s attitude may be rotten at times, which makes fixing bugs harder than it needs to be, but again, this is not a show-stopper as long as the software is free. At the end of the day, users do not care how their machine boots, as long as it does.
So once again I'd like to point the finger at the real offenders: KDE team in this case. They are free to decide whether they want to be portable, and to what extent, and if they limit themselves to Linux+systemd, well, whose fault is that? It's not an "infection", it's just one of the half-a-dozen DEs going non-portable. Sad, but you know what? Fuck it. We'll just keep using our openboxes and what not, while some of us will design and build yet more portable DEs.
(Score: 5, Insightful) by arashi no garou on Friday February 20 2015, @07:51PM
From the technical standpoint, there is nothing seriously wrong with systemd
I said as much in another comment; the biggest problem in my mind with systemd is a philosophical one, not a technical one. It's about power and control, more specifically about a few people taking the Linux community in a direction the community doesn't necessarily want to go. I say it was forced on us because quite simply, it was; when Debian adopted systemd against the will of many of its users and members (several Debian committee members asked for more time to debate the switch and were ignored), much of the GNU/Linux community had no choice but to follow, as Ubuntu and Mint, the two distros with the largest userbase, are Debian based. The whole Debian debacle was when I started really paying attention to systemd and Poettering's motives, and I saw the same tactics used in corporate takeovers (lies, backroom deals, bullying) being used in what is supposed to be the opposite of the corporate world.
Maybe, in the end, systemd will be the technically superior way to do desktop Linux. If so, I'm fine with that. What disgusts me and makes me want to avoid it is all the political bullshit and power plays behind it. That kind of thing doesn't belong in the Free software world, and it most certainly doesn't belong on my systems.
(Score: 2, Interesting) by anti-NAT on Saturday February 21 2015, @05:15AM
The joke of the outrage is that SystemD has been in Fedora since version 15, released in May 2011. As there have been 6 releases since, all including SystemD, clearly the problems the ragers think is going to happen won't. It would have been abandoned by Fedora, or Fedora's users would have abandoned Fedora, if SystemD is as bad as people are making it out to be.
As most people who criticise SystemD repeatedly state falsehoods about it, because they don't actually know what they're talking about (and have no interest in finding out about it), it really says that they're not complaining about SystemD, they're really complaining about change. Technology is always changing, if you can't handle change, go back to pens and pencils.
If the supposed worst happens with SystemD, (a) it can be forked and (b) its API calls are clearly available to be reimplemented.
Nobody is forcing other distributions or software to adopt SystemD or its components - they're doing it because they can see benefits in comparison to their existing methods.
Nobody is forcing you to stay using those distributions if you don't like the direction they're heading either. I've changed distributions plenty of times since 1993 - I've used Slackware, Redhat, Debian, Ubunto, ArchLinux and now use Fedora. It's not hard, and interesting to find out the different ways the same problems are solved.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @08:40AM
Fedora won't abandon it, because RH needs it for its security wankery sales pits to the MIC.
Fedora seems to always be on the forefront of adopting something "security" related, and/or pushing stuff that is related to same.
Lately there have been a whole lot of RH people screaming, under the guise of Gnome or Wayland devs, "X11 IS NOT SECURE!!!".
Fedora was also on the forefront of adopting SELinux, and it was a massive pain...
And one would suspect that if one traced the lineage of Polkit etc it would lead back to someone on the RH payroll.
But at the same time the Systemd guys manage to add a DNS "client" that is susceptible to cache poisoning. A issue as old as DNS caching, and has known techniques for avoiding...
In essence Fedora is sticking with it because it is "their" project.
Others on the other hand is likely adopting it, or adopted it, back when it was simply a init replacement. These days however it is so much "more". And has a stated goal of reworking the Linux "distro" (more like making them all march in lock step) from the ground up (cramming everything into /usr etc).
(Score: 2) by urza9814 on Monday February 23 2015, @07:31PM
I've had to add a bunch of systemctl commands to my .xinitrc because I couldn't figure out how the hell to get them to start in a predictable order. Network works fine one day, then I upgrade my system and suddenly the services start up in a different order and the network fails on boot...then I update again and it goes back the way it used to be...so now as soon as the system is up I just immediately reboot all the related services in the correct order since SystemD apparently can't figure that out...
(Score: 4, Interesting) by Anonymous Coward on Friday February 20 2015, @09:28PM
No one was forcing anyone to adapt systemd, let alone adapt it so soon.
That's like saying "I'm not forcing you to stop driving a car, I'm just removing all gas stations in a 100 km radius around where you live. You can still drive if you want to. You can even get gas and refill your car to keep driving. It's your choice." True each application can choose to adopt or not adopt to it, but given how tightly integrated it is with certain applications if you have any dependency on one of those applications you are stuck with the "choice" of either not adopting it and manually working around it (driving 100km each time you want to get a patch), or adopting it (stop driving your car). Nominally you have a choice, but only in a superficial sense.
From the technical standpoint, there is nothing seriously wrong with systemd, meaning that whatever faults it has today can be fixed tomorrow. (For example, binary logs are stupid, but there is nothing preventing admins from keeping text logs as well.)
Unless you happen to disagree with the design philosophy of system (i.e. "having it do everything" with the commensurate loss of control and increased system fragility). That can't be "fixed" as such. And in the meantime, you are still using unproven software which may have security holes, stability issues, maintenance issues, or whatever else.
It's not an "infection", it's just one of the half-a-dozen DEs going non-portable.
Calling KDE "just one of the half-a-dozen DEs" grossly under-represents how big and important it is. And the word "infection" is a perfect term given the network effects of dependencies across software packages.
(Score: 2, Disagree) by melikamp on Friday February 20 2015, @09:43PM
The parent post is a perfect example of the unadulterated value-laden cow dung I was talking about, complete with "design philosophy" nonsense. It fails to address the single point I was stressing: distributions independent of Red Hat are adopting systemd by choice, and userland applications create hard dependencies on systemd not only by choice, but also because they, apparently, intend to be unportable just for lulz. Good riddance. There is no doom coming, systemd is just another init, and not the worst one, and while some upheaval will force some users to seek new pastures, what we observe is business as usual.
(Score: 5, Insightful) by Hairyfeet on Friday February 20 2015, @10:24PM
And I think YOU sir have missed the point of many here, which is that the users have said by a pretty damned large margin "We do NOT want this" and the devs have turned top their own users and said "Fuck you, we do not care what you want"...yes this attitude is very disturbing, and when you see distros that are supposedly "stability over all" like Debian jump on the bandwagon and start throwing up appeals to emotion instead of technical arguments?
Well as I have said in other posts everybody can call me a tinfoil hatter (just as they did everybody who said our calls were monitored before Manning) but the combination of timing, coming so soon after Snowden let loose the TLAs best tricks and many are pushing Linux as a platform for secure communication, mixed with the frankly bizarre posting made by devs defending Poettering....I mean for fucks sake LP in one of his own posts said along the lines of "Can't get systemd working on ARM, fuck it shipping anyway" and devs not only didn't condemn him for treating a critical subsystem like systemd like its a fricking videogame patch and busted his ass for sending it out broken but they rushed to DEFEND such Mickey Mouse behavior? I'm sorry but the top tier Linux guys have NEVER been tolerant of the "fuck it, ship it and we'll patych later" kind of cowboy bullshit, they condemn that kind of shit on the Windows side all the time!
I'm sorry but something is rotten in Denmark, the entire dev community is just too quick and too easy and too busy acting like every day is opposite day for me to believe this is a natural or organic change in the landscape. Whether this is the TLAs that pretty much own Red Hat making sure there is a nice easily exploitable back door so that all these "secure communication distros" are just so much warm fuzzy BS, or Red Hat trying to become the big boss? I have no idea. All I do know is I have been keeping an eye on Linux since around 2003 and I can honestly say I have NEVER seen Linux distro devs behave in such a strange and anti-user fashion before, never.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 1, Troll) by melikamp on Friday February 20 2015, @10:41PM
This has absolutely nothing to do with either systemd or the "cabal" or anything at all in the sense that it's a completely generic problem, and can only be described as "business as usual". This is exactly how developers relate to users during a software development cycle; this is true for the free software, and 3 times as true for the proprietary spyware. Exceptions are rare and precious. And if you or any of the systemd doomsayers were actually interested in solving THIS problem, you wouldn't make it about your personal hatred of the "cabal" and everything they stand for, would you? You would talk about the community culture at large, problems in communication, structural inefficiencies, etc.
Both GNU and Linux are maturing and growing, and it makes sense that each new change in the ecosystem brings a bigger upheaval, that's all. We transitioned from an ecosystem where many users were also contributors to the one where almost all users are pure users. And pure users are NEVER happy with change, and when there are 100 times more of them, they are 100 times more vocal about it.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @08:48AM
Not sure how many are adopting them by choice, or by seeing the 800 pound gorilla throwing its weight around and not wanting to be in the way. Systemd is adopting more and more "container" functionality. CoreOS recently used this to launch a rival to Docker, and recent Systemd releases has gained the ability to import Docker containers. RH event stated that the cloud would be a big focus of theirs moving forward. In essence, if you want to be in the server game going forward you better follow RH's lead.
(Score: 5, Insightful) by Arik on Friday February 20 2015, @09:49PM
You're wrong. The design aim of 'throw everything in PID1' is wrong, wrong, wrong. The problems with systemd are inherent in the design philosophy and cannot be fixed without fundamentally altering the nature of the project.
Your binary logs example is actually a good example - against your thesis. In fact, although you can relatively easily get systemd to produce a second set of logs, in text, and this *appears* at first glance to moot the objection, in fact it is completely unsatisfactory. In case of system crash or failure to boot - the very cases where the logs are most needed and it's most important that they be in text - this setting fails as well.
The distinction you make between 'technical' and 'philosophical' objections is also a mirage. In fact, the 'philosophical' objections are actually technical. We do not oppose overloading the init system with all sorts of other functions because it offends our esthetic sensibilities - we oppose that because we know it is the opposite of good engineering and will inevitably result in bugs and bad behavior.
After you stomp bugs for a decade or two maybe you think to quit leaving out so much food for them.
If laughter is the best medicine, who are the best doctors?
(Score: 2) by melikamp on Friday February 20 2015, @10:11PM
I am not an OS designer and my personal opinion about the technical merits of systemd is pretty worthless. I can see, however, that major distributions are adopting systemd one after another, motivated in large part by what THEY perceive as technical advantages, and the doom is failing to materialize. I also read the criticism of systemd by people who are qualified to judge it on technical issues, and they seem to agree on the fact that it's not perfect, but basically works. I also knew you have no idea what you are talking about when you mentioned the "design philosophy". Would that be the "UNIX design philosophy", by any chance? "Do one thing" and blah blah blah? Because Linux (just 1 example!) spat all over that, but I don't hear you complaining.
If what you are saying is true, then the bug can be summarized as follows: when boot fails, logs are unavailable, period. What does it have to do with logs being binary or text? I am not saying systemd is bug-free, I am just saying YOUR objections to it make no fucking sense to me.
Even I can tell you don't seem to understand (or may be don't even care to try to understand) how systemd is designed. No, you don't have any technical objections, because if you had one, you'd let us know what it is, and then we would see that like all the other inits, systemd has its pros and cons.
(Score: 5, Insightful) by hash14 on Saturday February 21 2015, @04:36AM
I can see, however, that major distributions are adopting systemd one after another, motivated in large part by what THEY perceive as technical advantages, and the doom is failing to materialize.
The doom will most likely materialize in the same form as Shellshock or Heartbleed or any other of the many other high profile vulnerabilities we just saw last year. But LP would eat his shoe before admitting he's wrong so nothing will get done about it, plus the way they see any attempt to fix their disaster as WONTFIX NOTABUG just shows how adamant they are about taking over the whole Linux ecosystem and tanking it.
But I will be far far away on either Gentoo, BSD, or some other alternative when this disaster strikes, so I can't wait to see the shit hit the fan.
I also read the criticism of systemd by people who are qualified to judge it on technical issues, and they seem to agree on the fact that it's not perfect, but basically works.
Sounds a lot like Windows, which seems to have yet another critical vulnerability every other patch cycle. Why was Debian the most successful project because it was taken over by Red Hat? Because they strove to make it perfect. There is a point to the pain in the labour - because otherwise, you just have a crappy platform that's full of holes, and which nobody understands which is a hacker's dream. Just wait and see - eventually, Red Hat will be reduced to an open source clone of Windows. That's great if all you care about is money and market share, but when it comes to getting shit done, those who want real assurance of stability, security and productivity will be doing something completely different.
(Score: 0) by Anonymous Coward on Saturday February 21 2015, @06:36PM
THIS.
It is the new blossoming openSSL project catastrophe ripe for the inevitable DOSing of the entirety of Linux at some point in the future.
Speaking of openSSL, SystemD might want to add that to the load of crap it handles, or at least contact the openSSL team and let them know SystemD needs to be one of their dependencies, because FUCK KNOWS WHY.
(Score: 4, Informative) by Arik on Saturday February 21 2015, @05:14AM
There are no technical advantages of any consequence, and if you think technical advantages are driving adoption you have not been paying attention. Politics is driving adoption and that is plain as day.
Disagree? Name one.
"If what you are saying is true, then the bug can be summarized as follows: when boot fails, logs are unavailable, period. What does it have to do with logs being binary or text?"
It has everything to do with binary logs vs text logs. When a system crashes with a text file open, the result is human recoverable and readable. When a system crashes with a binary file open, the same is not true.
The right way to do this, assuming there was someone with a compelling need for the binary logs in the first place (which may or may not be true,) would be to do the text log as primary, then mirror it to a binary log.
"I am not saying systemd is bug-free, I am just saying YOUR objections to it make no fucking sense to me."
You may need more experience before you will understand them. The design guarantees bugs, I dont need to wait around for 10 years for the bulk of them to be found to know that they are inevitable (or to anticipate that the cabal will WONTFIX the bulk of them.)
If laughter is the best medicine, who are the best doctors?
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @08:42AM
Could it be that Slack ends up adopting Trinity rather than follow KDE forward?
http://trinitydesktop.org/index.php [trinitydesktop.org]
(Score: 2, Insightful) by Anonymous Coward on Friday February 20 2015, @04:59PM
>suggests that either Slackware will be forced to adopt systemd to keep KDE, or else Pat will be forced to abandon KDE
No need of either; there are enough skilled people in Slackware community to work around any such contrived dependencies.
Just the same as Skype's requiring PulseAudio absolutely had not resulted in us "adopting" the beastly thing. Instead, in mere days, we have developed workarounds, and have no problem since.
(Score: 2) by Nerdfest on Friday February 20 2015, @07:39PM
Sorry, I meant Kubuntu. All use both KDE and Xfce and am too used to typing both.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @12:57PM
One XFCE dev actually forked the consolekit code, and has done some effort towards putting in support for logind dbus calls etc.
https://erickoegel.wordpress.com/2014/10/20/consolekit2/ [wordpress.com]
(Score: 3, Insightful) by frojack on Friday February 20 2015, @09:46PM
assuming a shim layer is not being implemented that will convert time and date API calls
I really can't understand why that approach isn't being taken.
Write it once, and done, no combing through your code to change every instance of an ever increasing number of functions that are handled by multiple different init systems.
This has been the prevailing method of dealing with changes in underlying structures, libraries, compilers, etc since before the systemd tyrants were born, since before linux was born. These shims are tiny, usually argument re-arranging, a jmp and a ret, and very little else. Only occasionally is any significant code or data structures needed.
No, you are mistaken. I've always had this sig.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @08:53AM
How deep does the rabbit hole go? Will we end up implementing systemd to avoid using systemd?
(Score: -1, Flamebait) by Anonymous Coward on Friday February 20 2015, @07:22PM
Did you learn from your experience and turn down your anti-systemd hyperbole level, or are you just going to keep blundering ahead with lies, even after falling on your face?
One thing becomes more and more clear as more software takes these decisions: The people in the "loud pundit" class have different views of systemd than the people actually in the positions to make technical decisions. And members of the "loud pundit" social category value each others pejoratives more highly than the technical points of the people actually in possession of knowledge of the details. What really blows the doors off is that they defend SysV init as some sort of holy cow. Uhm, yeah. If they're not embarrassed the first time they say it, they're probably never going to go into the technical details far enough to even understand the decisions.
(Score: 5, Insightful) by Thexalon on Friday February 20 2015, @01:30PM
When the boot system demanded that the authentication system relied on it. The desktop has to rely on the authentication system to start sessions.
This is my biggest concern with systemd - the clear goal is to make it a dependency of everything other than the kernel. Absolutely nothing in the open source world has been in that position before (not even the Linux kernel, since userspace applications were typically built to work on BSDs and commercial Unixes and Cygwin too).
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 5, Insightful) by digitalaudiorock on Friday February 20 2015, @02:53PM
Exactly...like trying to run Windows without svchost.exe...which is actually a good analogy, because frankly using systemd is very much like running Windows and it's bloated svchost crap...binary Windows-fucking-event-log etc etc etc. I can't believe the open source community hasn't just flat out rejected it because of those phony dependencies alone. They're all clearly malicious...our way or the highway.
BTW...why does everyone suddenly start posting anonymously whenever systemd comes up? Does LP have like mob ties I don't know about? I'm not afraid to say how much I think it sucks...that's for sure.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @10:42PM
Choosing to not have an account when it is not necessary correlates highly with GNU philosophy. Likewise, sockpuppets and friends always post with accounts to appear credible.
(Score: 2) by digitalaudiorock on Saturday February 21 2015, @01:32AM
Wow...way to contribute to the discussion. I'll let other "sock puppets" weigh in on your bit of wisdom there.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @08:56AM
Given the kdbus project, it may well be that the kernel comes to depend on systemd to do the talking to the userland...
(Score: 1) by Anonymous Coward on Friday February 20 2015, @01:34PM
According to the blog, it doesn't. It just depends on something that happens to depend on the init system. You could always develop an alternative with the same badly defined and frequently changing interface, right? Something like wine trying to translate calls to Windows internals...
(Score: 2, Insightful) by khallow on Friday February 20 2015, @03:55PM
(Score: 4, Informative) by Gravis on Friday February 20 2015, @02:40PM
Since when did a desktop environment start depending on 'boot' systems?
actually, it doesn't.
The init system is one part of systemd that doesn't affect us at all, and any other could be used.
the problem we have here is that logind has been integrated into systemd.
(Score: 2, Interesting) by Anonymous Coward on Friday February 20 2015, @04:36PM
It's a good thing. The polarization of the 'community' on the issue has resulted in a need for users of the free/Free ecosystem to wakeup and take responsibility for the tools they need. Generally the slow moves toward the current situation had been ignored because if you didn't need it, you didn't use it. Now the coup has be sprung people have shown who they are and what colours they fly in such a way as they can be clearly identified.
As a generally if enough people view something as damage to the network, it will be worked around. It turns out that the people working around this current damage are the same people who worked around proprietary software restrictions the first time. Compared to that this is are walk in the park.
No doubt, as the proposed target audience turns out not to exist, and the people who are able move somewhere else/create something new, the carpetbaggers with come running to try and takeover the new commons as theirs crumbles.
(Score: 5, Insightful) by Marand on Friday February 20 2015, @02:55PM
The gist of it is this:
"systemd" is a bunch of different things, ranging from an init system and binary (ick) logger to programs that manage login sessions, date and time, etc. It's a blanket moniker for a group of programs that Poettering and co. have decided will be the standard components for Linux systems. With that mindset, they're not worrying about those parts depending on each other, because it's supposed to be a standard kit that you use all together.
(brief tangent follows)
It's sort of like how FreeBSD separates the base system, which updates all together at once, and is separate from the other packages. Except that where people approve of FreeBSD's doing the same thing, it's not being received well among Linux users for various reasons, ranging from the absurd (conspiracy theories, misinformed beliefs, "it's different so it sucks") to the reasonable.
Part of that is because of the personalities leading the project, because of a reputation for bad designs and even worse implementations that then take others years to fix into something usable. It also doesn't help that Poettering and Sievers act like arrogant, know-it-all shits, actively interfering with improvement of their software. It's not a design flaw, it's a feature. We don't care if it corrupts logs, breaks randomly, or prevents booting. You just hate handicapped people or you'd be agreeing with me. NOTABUG WONTFIX GTFO. If someone else, more receptive to changing the design of the components, and not responsible for abominations like pulseaudio, were spearheading the whole thing a lot of the contention probably wouldn't exist.
Another problem is that it's not the way Linux has traditionally worked, so you have a group of people basically pushing for high levels of interdependency at a low level, where it's always been a very mix-and-match affair. In a relatively short period of time, this small group has basically declared itself the final authority on the Linux userland and started ripping up all the fixtures to put in its own, without any consideration of the others that happen to live there.
Anyway, back to KDE. The parts of it that KDE is apparently going to require are just another bullshit dbus, hal, etc. type of thing. No init system requirement. Those parts are created with an expectation of certain things, which creates a dependency on the systemd init, but it can be circumvented, at least for now. Debian has a small package called "systemd-shim" that gives the other systemd pieces (logind, etc.) what they need to operate (just cgroups support I think?) while allowing you to install any init system (and thus logger) you want.
As long as the shim's being maintained, Linux users can still avoid those parts of systemd, which is important IMO because a lot of the systemd complaints are about the init and logging system, specifically. (That's where most of my gripes are)
(Score: 2) by arashi no garou on Friday February 20 2015, @06:21PM
I agree with pretty much everything you said. I feel the problems with systemd are more philosophical than technical (though there are definitely technical issues too), and while I'm no conspiracy theorist, anyone can recognize that it's a power play by Poettering and Sievers driving the rapid growth of the project. I'm sure they honestly want "a better Linux OS", but they are going about it in a dishonorable way.
But all that aside, what happens when the shim no longer exists? A shim is meant as a temporary fix, not a permanent solution. There are ultimately two paths for any GNU/Linux ecosystem project: Accept systemd or reject it. A shim is just there to allow one to take longer to decide.
(Score: -1, Flamebait) by Anonymous Coward on Friday February 20 2015, @07:53PM
And by "dishonorable," we mean, "they were hired to make the technical decisions, and they made decisions I wouldn't have made. And since I don't have their resume, and can't compete for the decision-making job, waaaaaaaaaa. waaaaaaaa. waaaaaaa."
Your neckbeard is crying all over the floor, jeeze.
(Score: 2) by arashi no garou on Friday February 20 2015, @08:47PM
Wait, you're saying that Poettering and Sievers are employees of Debian? What crack are you smoking?
And nice try with the "neckbeard" comment; your ad hominem is crying all over the floor.
(Score: 2) by Grishnakh on Friday February 20 2015, @08:14PM
How are they being "dishonorable"? Their job is to create this system (they work for Red Hat, remember, who obviously wants this system). Other distros are adopting it, rightly or wrongly. Red Hat does not have the power to force other distros to adopt anything (or else we'd all be using RPMs right now), so obviously these distros are doing it willingly. How is that dishonorable?
I get the complaints about PulseAudio etc. However, the distros all adopted that willingly too. And for good reason: despite its many problems, it was better than the alternatives.
Basically what this all seems to me is: Lennart and company see a big problem with Linux, they cobble together a solution to attempt to fix it, and everyone else in their armchair bitches and complains that this solution isn't good enough. However, no one actually bothers to make their own solution; instead they seem to want to just keep using the broken stuff they've been using for 15 years and never change it. Ok, I get there's complaints about the technical quality (bugginess) of Lennart's stuff, about the way his cabal treats people, etc., but again, what alternative do we have? No one else wants to step and and do something similar. At best, we get other solutions which have far worse technical problems, like upstart, or we get solutions which simply don't go nearly far enough (OpenRC) to address the problems. I think we saw this with PulseAudio too, with competitors like ARtS.
As for shims, that's totally not true about them being temporary. I don't know where you got that crazy idea. Look at Nvidia's proprietary drivers for instance: they use an open-source shim to interface with the kernel. That's not a temporary solution; it's been that way almost forever, and isn't going to change unless the kernel devs suddenly decide to have a stable ABI for proprietary drivers (not going to happen). Or, if you forget about software, look at window shims: you use these to make your window square and level when you're installing it into your wall. Those shims are never removed; they always stay in between the window and the framing.
(Score: 3) by arashi no garou on Friday February 20 2015, @08:44PM
By "dishonorable" I mean the backroom tactics I've read about that led to the Debian decision. systemd may be a great piece of software, but if so it should compete on its merits, not on the whims of someone not even a part of the Debian group.
(Score: 3, Interesting) by Grishnakh on Friday February 20 2015, @09:26PM
What "backroom tactics"? I must have missed this. I thought the Debian decision was made by a vote between the people in charge of the project, not some outsiders. From what I remember, there was a lot of complaining by an Ian Jackson who seemed to be holding everything up because he didn't like it, but that's just one guy, and then when everyone else voted against him he got mad and quit the group.
(Score: 3, Informative) by tangomargarine on Friday February 20 2015, @11:03PM
"Everyone else voted against him" is a lie. The 8-person vote was 4 in favor of systemd as their first choice and 4 in favor of the other alternatives. And IIRC the only reason systemd won that vote was because the guy who "broke ties" (whatever you call it when you have a plurality but a majority is required) was in the systemd voting block.
Go look up the original Slashdot/Soylent articles on the vote. I swear Debian made their voting system as hard to understand as possible.
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
(Score: 1, Troll) by Grishnakh on Friday February 27 2015, @08:25PM
It's not their fault you're too dumb to understand Condorcet voting, or any kind of voting system where you're allowed to vote for your preference for candidates.
(Score: 3, Informative) by arashi no garou on Saturday February 21 2015, @01:52AM
It's my understanding that there was a split vote, some of the voters wanted more time to think about it, but some heavy pushing by LP and KS privately caused those requests to be ignored. The vote was rushed and the swing voter said what amounted to "fuck it, roll with it" and it was done.
That's what I read back when it was going on, though I've heard from some (admittedly heavy systemd supporting) sources that it didn't go down like that, that it was a simple majority vote without drama and the perceived drama was manufactured by the anti-systemd camp. I find that hard to believe though, given the amount of controversy on both sides.
Either way, I choose not to run systemd because of that very controversy; I don't need drama with my OS. Hell, my Raspberry Pi's OS has some systemd parts in it, it hasn't crashed and burned and sacrificed goats like some anti-systemd folks swear it will. But I'm looking forward to getting my new RasPi 2 next week that should be able to run Slackware without issues.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @09:14AM
Poettering puts code for Logind into Gnome, while also declaring (he had gotten maintainer status) consolekit as dead to the point of shutting down the relevant mailing list.
Then he turns around and claim that nobody wanted consolekit since nobody was maintaining the supporting bits in Gnome.
Err dude you already declared consolekit dead and logind as the path forward...
The guy twists words like a snake coils, and then turns around and yells at Torvalds for making the Linux community a toxic work environment.
There is a video out there of one guy doing a presentation about what he sees as absurd layers of complexity in present day Linux. One of the projects called out happens to be one of Poettering's, and he is in the audience. Poettering starts heckling the presenter, to the point of asking of the presenter has something against disabled people (because some of the complexity could be used to help them in some way). In the end the presenter calls it quits, and Poettering jumps on stage and grabs the mic (beer bottle in hand no less)...
(Score: 2) by choose another one on Friday February 20 2015, @08:38PM
Shim is a misnomer. One of the aims of modular development (aka the Unix way) is to enable the use of alternative implementations of some components - and that is exactly what these "shims" are, alternative implementations of systemd components without the coupling to the rest of systemd.
Key thing of course is that someone has to maintain the shims, but only a small fraction of the effort that goes into complaining about systemd would be required for that...
(Score: 2) by arashi no garou on Friday February 20 2015, @08:52PM
Key thing of course is that someone has to maintain the shims
That is exactly my point. What happens when the shim maintainer walks away because either she accepts systemd or moves on to, say, BSD? A shim is a band-aid on a bleeding gash.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @06:53PM
I saw that video and felt the presenter was making some excellent points. Poetterng's behavior was surprising, childish and inexcusable. Systemd is being developed by a spoiled brat and while there are many other good reasons to avoid it, that fact alone is enough to turn many away.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @09:01AM
Well the BSDs are monolithic projects from the outset.
Linux on the other hand has been more like a bag of Lego bricks, each independent but interlocking via defined interfaces.
(Score: 5, Insightful) by moondrake on Friday February 20 2015, @02:58PM
At times as these I regret not having the "overrated" mod anymore. You made a fairly silly comment, which is aligned with the general opinion, and gets modded up as insightful...
You realized this later (see sibling), but I cannot, in good conscience, mod you Troll or Flamebait (as I think you simply did not check the FA yet), so the comment stands...
For the record: I dislike systemd. But I also see that some parts of it (specifically logind actually) are very useful and _better than what we had before_, as they provide a single backend for doing things that all DEs need. The only downside is that they rely on other parts of systemd that I am not too happy with.
(Score: 3, Insightful) by DeathMonkey on Friday February 20 2015, @06:49PM
At times as these I regret not having the "overrated" mod anymore. You made a fairly silly comment, which is aligned with the general opinion, and gets modded up as insightful...
Meanwhile, the abusive down-modding that removing "overrated" was supposed to fix continues unabated with "redundant" or "troll."
(Score: 2) by arashi no garou on Friday February 20 2015, @07:04PM
Don't forget "Offtopic"; I've seen a few posts modded as such that were topical. I can only imagine it was someone who meant to hit that "Disagree" button and suddenly slipped...
(Score: 2) by hemocyanin on Friday February 20 2015, @10:05PM
Downmodders like to see the score go down. Disagree is a fail in that regard.
(Score: 2) by arashi no garou on Saturday February 21 2015, @01:55AM
That's the joke...
(Score: 2) by moondrake on Saturday February 21 2015, @09:21AM
Well, it does not make much sense to mod comments disagree when they contain a logical fallacy. Or when they ask questions that have been answered in the FA.
I would be happy if we had a 0 mod to point out such cases.
(Score: 3, Insightful) by khallow on Friday February 20 2015, @07:25PM
You made a fairly silly comment, which is aligned with the general opinion, and gets modded up as insightful...
He asked a question not made a "fairly silly comment". Perhaps you could just answer the question.
(Score: 2) by moondrake on Saturday February 21 2015, @09:01AM
The question that he put in his comment was not a very good one because the FA addresses precisely this thing. The parts that KDE depends on is not the boot system, and this is clearly explained if you just read the article. (That you highlight this question again is simply beyond silly, and not insightful at all dear mods...).
(Score: 1) by khallow on Saturday February 21 2015, @05:57PM
The parts that KDE depends on is not the boot system,
Wasn't hard, was it?
(Score: 2) by moondrake on Saturday February 21 2015, @07:19PM
No, and Gaaark realized this long before you posted, which I referred to in my comment. I really do not see your point.
(Score: 2, Informative) by khallow on Saturday February 21 2015, @08:22PM
No, and Gaaark realized this long before you posted, which I referred to in my comment. I really do not see your point.
Let's look at this as a timeline. A poster, Gaaark asks a reasonable question. Why does KDE depend on systemd? Let me remind you that not everyone is born with perfect knowledge of what systemd and its dependencies are.
Three minutes later he has an answer to his reasonable question and he posts it. Drama averted right? No, for you then post your condescending and useless reply a full hour and a half later.
So to summarize, your original reply adds nothing to the discussion aside from raising other peoples' blood pressure.
(Score: 2) by moondrake on Saturday February 21 2015, @08:58PM
I disagree. He did not read the article at the time of that first post. The answer to his question was in the FA. I pointed out that he should not have been modded up and had no way to correct it in a sensible way.
That comment was offtopic perhaps, but not totally without merit imho. I do not think calling his comment somewhat silly is condescending. I think it is a quite nice description for posting without reading what the story is about. My somewhat offtopic point was about the moderation and since I could not downmod, I could at least point out the post was not relevant for a sound discussion of the article. Moreover I raised the question of whether we do not need some mod system for this type of situation.
You on the other hand, did not read or understand the article, and posted a moronic reply (this is condescending yes, but by now I think you deserve it). Instead of silently ignoring my reply and not make matters worse for both of us, you resort to trolling (i.e. first you call me out to answer the question, then later argue my post was irrelevant to begin with). Discussion is over.
(Score: 1) by khallow on Saturday February 21 2015, @09:10PM
I suggest the following two step approach:
1) Apologize.
2) STFU, unless you have something useful to say.
Now, having said that, I think it's time to take my own advice.
(Score: 2) by HiThere on Friday February 20 2015, @08:59PM
I think your use of the word "only" unjustifiably trivializes the problem. Otherwise I pretty much agree.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 1, Funny) by Anonymous Coward on Friday February 20 2015, @01:25PM
We can now hear the Knell of Death Echoing.
(Score: 3, Funny) by Anonymous Coward on Friday February 20 2015, @10:03PM
It's echoing because it's run by PulseAudio, playing through multiple audio systems in other rooms after a slight delay, and then crashing.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @11:37PM
What, systemd is replacing "echo" now?!!!
(Score: 2) by Subsentient on Saturday February 21 2015, @01:31AM
Don't act so surprised.
"It is no measure of health to be well adjusted to a profoundly sick society." -Jiddu Krishnamurti
(Score: 5, Informative) by Anonymous Coward on Friday February 20 2015, @01:41PM
This not-unexpected development is one more reason to be glad that a fork of KDE3 is still alive to this day.
http://www.trinitydesktop.org/ [trinitydesktop.org]
(Score: 4, Informative) by Thexalon on Friday February 20 2015, @02:46PM
Related warning: MATE 1.12 is slated to add systemd-logind [mate-desktop.org]. I get the distinct impression that the way things are going the only non-systemd desktop left will be twm.
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 3, Insightful) by arashi no garou on Friday February 20 2015, @06:32PM
I think pretty much all of the bare WMs are safe. It's the resource-heavy DEs, the ones that have been trying to emulate Windows and OS X, that are falling victim to the systemd tentacle approach.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @09:29AM
It boils down to having a "common" way to delegate minimal and temporary system changing rights, so that the user can hook their laptop up to a new wlan, or mount that thumb drive with the photos of dear aunt Tillie, via a GUI.
But more and more it seems like what is driving it all is a desire to placate the MIC.
Multi-seat for instance seems to be about being able to airlift one computer and a bunch of screens and keyboards into the field, yet make sure that only the spooks can read how quickly the war is going to the dogs (the "lesson" from Vietnam has not been to avoid dirty wars, but to control the information flow from the warzone).
Containerization is being pushed by systemd and Gnome as a security thing. And Fedora, the place where systemd etc is germinating, was a early adopter of SELinux.
A bunch of other stuff seems to be treating users like sheep in need of protection from themselves, or chasing theoretical corner cases.
Perfect is very much the enemy of good in the end.
(Score: 3, Insightful) by engblom on Friday February 20 2015, @03:41PM
Sadly no major distro is using this one as default desktop. Slackware avoids systemd so lets hope Slackware will adopt this one in a soon future.
(Score: 4, Informative) by Gravis on Friday February 20 2015, @02:37PM
the article is here [davidedmundson.co.uk].
(Score: 0) by Anonymous Coward on Friday February 20 2015, @02:51PM
Now even though we are suppose to be getting good news it's not tied to systemd, you sure can feel the pressure it exerts on everything around it.
"One Ring to rule them all, One Ring to find them, One Ring to bring them all and in the darkness bind them."
Its back to the C64 for me.
(Score: 1, Informative) by Anonymous Coward on Friday February 20 2015, @03:52PM
The original version. [soylentnews.org] Of course Tolkien could not understand it back then, so he changed it to the ring version. ;-)
(Score: 3, Insightful) by The Archon V2.0 on Friday February 20 2015, @04:30PM
> Its back to the C64 for me.
Yeah, the going gets tough and you cowards run right back to a Microsoft OS.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @08:27PM
Ha! Well played... I loved (LOVED!) my Commodore gear and had nearly forgotten that it was, in fact, running a Microsoft OS. Still, my c64 was utterly awesome and now I'm off to eBay for a trip down memory lane... :-)
(Score: 2) by The Archon V2.0 on Friday February 20 2015, @09:07PM
Yeah, the 64 was awesome. Only reason I knew it was running Microsoft BASIC was because I saw the copyright for BASIC 7.0 on a C128 start screen many years later, thanks to emulation.
(Score: 2) by janrinok on Friday February 20 2015, @08:37PM
Er - the C64 being referred to is, I believe, the Commodore 64 [oldcomputers.net] from Commodore Int'l Ltd. How you got modded as 'Insightful' for mentioning Microsoft is a mystery to me.
Unless I've just missed a 'whooshing' sound - but then you would have to explain that to me in more detail which means it wasn't as funny as perhaps you intended?
(Score: 2) by janrinok on Friday February 20 2015, @08:39PM
(Score: 2) by The Archon V2.0 on Friday February 20 2015, @09:09PM
It's cool. There's no major sign BASIC 2.0 is Microsoft, it's something I think most of us learned long after the fact.
But then, who used BASIC after the learning stage for anything but simple programs and the LOAD command?
(Score: 0) by Anonymous Coward on Friday February 20 2015, @09:30PM
janrinok here:
My search of Wikipedia suggests that the OS was called GEOS (which is what I remember in the UK C64) and was written by Berkeley Softworks (later GeoWorks). In fact, I can find no mention of an MS OS for the C64. The basic I remember as GeoBasic, which also ties in with Wikipedia.
Myself, I was a Nascom 1 and 2 user, and subsequently a Galaxy user.
(Score: 2) by Arik on Friday February 20 2015, @09:39PM
GEOS was an option you could add later. The actual built in OS on the 64 was KERNAL, and IIRC it was developed in-house and went back to the PET days at least. The *shell* was Commodore BASIC, also developed in-house back at least to the PET, but I believe it was a licensed fork from MS BASIC at some point.
If laughter is the best medicine, who are the best doctors?
(Score: 2) by The Archon V2.0 on Friday February 20 2015, @09:40PM
The window-icon-mouse-pointer OS in the C64 was GEOS.
This part, that comes with every C64 and starts when the computer does (if you don't have a cartridge plugged in), is a Microsoft OS: http://www.zweigrafiker.com/c64/assets/images/c64.gif [zweigrafiker.com]
As evidenced by: http://www.commodore.ca/products/128/c128_basic_7_screen_shot.gif [commodore.ca]
And: http://www.pagetable.com/?p=43 [pagetable.com]
(Score: 4, Funny) by Anonymous Coward on Friday February 20 2015, @03:01PM
Well, there's slack, slack, systemd, slack and slack. That's not got much systemd in it.
(Score: 1, Funny) by Anonymous Coward on Friday February 20 2015, @03:44PM
It's not got much anything else in it either, which is why it's not a drop-in replacement for the people using something like Ubuntu.
(Score: 5, Insightful) by Foobar Bazbot on Friday February 20 2015, @04:47PM
... [slackware]'s not a drop-in replacement for the people using something like Ubuntu.
Of course not, slackware's a whole Linux distribution. All you need to replace the people using something like Ubuntu is a very small shell script.
(Score: 3, Insightful) by arashi no garou on Friday February 20 2015, @07:12PM
Trollololol.
While Ubuntu's Software Center locks up and greys out your screen when it tries to install the 600MiB of deps for a 500KiB app you wanted, I'll be using installpkg to install just exactly what I need and no more. It's called being lean, and it's not a bad concept. ;-)
(Score: 3, Funny) by DECbot on Friday February 20 2015, @09:23PM
It's called being lean, and it's not a bad concept.
I'm big boned you insensitive clod!
--Signed: systmed.deb
cats~$ sudo chown -R us /home/base
(Score: 2) by Ryuugami on Friday February 20 2015, @10:38PM
--Signed: systmed.deb
Oh god, it's self aware!
If a shit storm's on the horizon, it's good to know far enough ahead you can at least bring along an umbrella. - D.Weber
(Score: 0) by Anonymous Coward on Friday February 20 2015, @11:31PM
*WHOOSH*
(Score: 2) by Magic Oddball on Saturday February 21 2015, @06:02AM
PCLinuxOS [pclinuxos.com] is a good drop-in replacement for regular users (I ditched 'buntu 5 years ago during the "we want window buttons on the left and we don't care what users want/need so STFU" debacle), as I thankfully learned in December. It was a huge relief to be able to trust my computer to stay up & fully-functional for more than a day or two at a time, which was the best I could manage under any of the systemd distros (including before I realized the newly-installed systemd was the reason my laptop had turned so unstable).
I looked into Slackware, but like Gentoo, it's a distro for people that want to spend their time delving into the nitty-gritty levels of Linux, not for those of us that think Linux makes a wonderfully flexible, fast, user-controllable OS that we can use while focusing on entirely unrelated tasks. :-)
(Score: 3, Funny) by nightsky30 on Friday February 20 2015, @05:55PM
I DON'T LIKE SYSTEMD!!!
Queue the vikings...
(Score: 5, Insightful) by engblom on Friday February 20 2015, @03:09PM
While my comments here might sound negative, I still like Linux.
Everything around GNU and Linux is about reinventing the wheel. GNU is an effort to reinvent UNIX in an own way.
I have been a Linux user since the end of the 90's, so I was not experiencing the beginning of the journey. However, since I begun with Linux we have seen huge amount of reinventing the wheel.
We had the painful OSS -> Alsa reinventing. *BSD stayed with OSS and improved on it and still today *BSD is using OSS. Meanwhile Linux had several years of buggy sound support. It was apparently not good anymore with the normal UNIX way: device files.
We saw a lot of tools trying to solve the audio problem: esd, artsd, jackd. Slowly Alsa became somehow stable. Then it was time to reinvent again the API with pulseaudio. Even today there are problems with pulseaudio. For example, if you press "mute" on keyboards, it will mute both Alsa and pulseaudio. However only Alsa is unmuted upon a second press. You have to dig in pavucontrol to enabling the sound. The fine grain control over the mixer was lost with pulseaudio.
On the graphical side we also saw big changes. There was a fork of XFree86. I fully agree Xfree86 was developed slowly but x.org just messed up everything. Graphics on my computers had been stable until the fork. With the fork you could not be sure anymore you had working graphics. Still today you see problems with many graphic adapters. Soon we might again have a working X as xorg is on its way to stabilize. Just in time for reinventing again with Wayland at any moment.
The desktops naturally had to be messed up too. We had the battle between KDE and Gnome because some people wanted to reinvent the wheel there too. KDE was far superior to GNOME in performance in every way (okay, KDE2 was quite messy, I confess). Between GNOME and KDE, KDE used less RAM in every test. It had features like cached thumbnails so you instantly could use a huge picture folder. GNOME was buggy and slow. Visiting a picture folder was truly painful. KDE3 was super great. It had everything you could expect from a desktop. It was time again to reinvent the wheel.
KDE4 came out and we saw the beginning of the slow death of KDE. KDE is not dead yet, but will it ever catch up what it could have done? After the terrible launch of KDE4 the much of the users escaped over to GNOME, which still had less features than KDE3, but had become somehow useable. Were we going to see a good desktop now? Oh, no GNOME3 is getting released. GNOME3 was a terrible reinvention of the wheel which caused a cascade of reinvention. Now we suddenly have Mate, Unity and other reinventions.
The init system had been working OK for too long time also. That could not be allowed so it was time to reinvent the wheel and get systemd into every distro. Now some computers take really long time to boot because of some bugs in systemd. Nothing is happening while systemd waits for something to timeout. The ridiculous part is that systemd was supposed to speed things up. I have seen this problem in plenty of systemd installations.
I am sorry, but we will not see the year of Linux Desktop any soon. People are too eager to reinvent to wheel rather than improving on what you have.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @03:21PM
To defend linux:
Well, a lot of work in coding is refactoring. Every few years, we want a new significant feature that was not envisaged few years ago, so we have to refactor our code base to make support of that new feature possible. This is just the way software is. If you don't refactor, then you end up bolting on new feature and new feature and the application architecture creaks and then collapses.
To use an analogy, if you keep on adding floors to a building, maybe it falls down. You can stick some RSJs in to hold things up, and maybe add another floor or two, but in the end you have to rebuild the walls to support the new structures. This is the way it is with software development as well.
(Score: 5, Informative) by engblom on Friday February 20 2015, @03:29PM
Refactoring is one thing and rewriting for the sake of rewriting is another. For example, OSS has seen several refactoring in the BSD world. Nowadays you can play sound from many sources without using esd/artsd/whatever. That was not possible in the beginning. Alsa was a complete rewrite.
Systemd is not a refactoring. It is a total rewrite of the init system.
With KDE4, really big parts were rewritten.
The only thing I mentioned which is pretty much refactoring is xorg.
(Score: 5, Informative) by Marand on Friday February 20 2015, @03:51PM
For example, OSS has seen several refactoring in the BSD world. Nowadays you can play sound from many sources without using esd/artsd/whatever. That was not possible in the beginning. Alsa was a complete rewrite.
You keep mentioning OSS vs ALSA, but you have yet to mention one of the motivations for a rewrite: the OSS license was changed from free to proprietary, which effectively left it abandoned for Linux and BSDs. With that in mind, it's not really surprising that developers on the GPL side would decide to start over and make the new project GPL to prevent the same thing happening again. Especially when faced with putting considerable effort into fixing the outdated OSS that they got stuck with because of the abrupt license change.
It's easy to forget that now, since it's largely glossed over these days, especially with most Linux users showing up well after the spat occurred. Plus OSS was eventually re-licensed again back to an open license (actually multiple...). However, at this point, switching back to OSS would be a massive undertaking, and would require trusting the owner to not pull the rug out from under everybody a second time.
(Score: 5, Informative) by Anonymous Coward on Friday February 20 2015, @04:01PM
Actually, Alsa was not the only licensing-driven change. x.org was triggered by an XFree86 license change. Gnome was started because of licensing issues with the original KDE (or rather, the underlying Qt framework), which later got resolved.
However there's no licensing reason for pulseaudio, Gnome 3, KDE 4, systemd and Wayland.
(Score: 3, Insightful) by Marand on Friday February 20 2015, @04:35PM
I was actually covering some of that in another response to his original "reinventing the wheel" comment , but it turned into a very long post, so it took a bit to get it typed out and try to check for obvious errors. I forgot to mention Xorg vs XFree86 though, thanks for bringing that one up. A lot of the changes have been licensing related. Or Poettering NIH-related (pulse, systemd).
(Score: 4, Informative) by engblom on Friday February 20 2015, @04:36PM
If OSS changed from free to proprietary license you should fork what you have rather than rewriting from scratch. They wanted to reinvent the wheel rather than forking which proves my point.
(Score: 1, Informative) by Anonymous Coward on Friday February 20 2015, @04:57PM
It's been almost 20 years already, but I thought the reason for moving to ALSA was to move away from Hannu Savolainen(?)'s commercial-open source split Open Sound System to something that was owned by the community.
(Score: 2, Informative) by Anonymous Coward on Friday February 20 2015, @07:17PM
OSS was maintained elsewhere. The author always produced two versions of it, one GPL and one not. The non-GPL version included drivers that were based on NDA info. The GPL version included everything else. Everybody was OK with this.
The problem was that the non-GPL OSS version was also available for UnixWare, OpenServer, Solaris, and FreeBSD. This required a kernel abstraction layer, which is verboten in the Linux world. Linux kernel coding standards prohibit any sort of compatibility layer because such layers make life difficult for kernel developers.
Alan Cox (at Red Hat at the time) ripped out the kernel compatibility layer. In doing so, he forked OSS. He soon lost interest in maintaining OSS, as you'd expect given his reason for forking was not a desire to write audio drivers. Linux OSS was thus badly maintained, missing out on all sorts of new features that mainline OSS (as seen in FreeBSD and elsewhere) was getting.
The ALSA people came alone proposing a solution. They would provide the features that Linux had been missing out on, and they would take care of audio. Linux accepted... then came close to ripping ALSA back out again when he saw what a mess it was. The very worst of it got patched up. It's still garbage, with a Windows-style library interface instead of a documented device file interface.
(Score: 1) by WillR on Friday February 20 2015, @09:36PM
It's still garbage, with a Windows-style library interface instead of a documented device file interface.
Serious question from someone who hasn't used *BSD on a machine with sound hardware lately: how does modern OSS (or anything else that does sound output The Unix Way(tm) by writing to a documented device file) handle multiple apps trying to play sounds at the same time? In the 1990s on Linux OSS/free it was either "you can't do that" or "use ESounD and esddsp", neither of which was entirely satisfactory...
(Score: 5, Informative) by Marand on Friday February 20 2015, @04:30PM
While there are some good points here, you're glossing over or completely ignoring a lot of relevant background to some of these things. This isn't a rebuttal of your "people reinvent the wheel" point; think of it as a clarification of some of the reasoning and histry behind parts of it. In the same order you mentioned them:
1. OSS to ALSA was inspired by the OSS owner switching to a proprietary license and leaving Linux and BSDs with outdated crap. It's not surprising that the GPL guys would decide to start over with something that's GPL instead of potentially helping someone by working on their old codebase and possibly getting the improvements 'borrowed' again.
2. esd, arts, jackd, and pulseaudio are an abstraction layer and not in direct competition with ALSA and OSS. You can have audio without pulse, but pulse without ALSA will be dead air. Pulseaudio complaints are warranted, considering a lot of the things pulse was intended to do were already possible in ALSA directly, like software mixing, and it took far too long to become something remotely usable. esd and arts predated pulseaudio and had a purpose that wasn't already filled at the time. Poettering has a thing for NIH, though, so it should surprise no one that years later he decided to write his own esd replacement.
Even with a mature ALSA, though, JACK (which also predated pulseaudio) has a purpose, because it presents advanced audio routing features at low latency for professional audio applications. Things like live rewind, retroactively choosing to record audio that played 30 seconds ago, etc. These are niche things that make sense as a separate stack on top of ALSA.
3. KDE vs. GNOME wasn't wheel reinvention, it was another licensing spat. KDE existed first, at least as a concept, but at the time Qt wasn't GPL, so the GNU folks went "okay we're going to do that too, but GPL", and started a rush to get a competing GNU product created. By the time Qt was freely licensed, GNOME was already in full swing and there was no way its devs would just drop the project and go to KDE.
4. KDE's "terrible launch" was the result of them deciding to call the pre-release 4.0, and a bunch of distros ignoring warnings not to package it yet. Debian didn't switch to KDE4 until 4.3 or so and it was smooth transition to a pretty stable desktop by then. Blame your distro for making a bad decision, not KDE for creating a pre-release.
Distros haven't made the same mistake this time, though, and are being more conservative with KDE5. At least they learned.
5. Which init system? BSD-style init, SysV style init, or OpenRC? Which variant of SysV, since the distros have all tweaked them to their own liking? We've always had a variety of init systems, and even when distros started with a similar base, they diverged based on their own needs. It just wasn't news because each one did its own thing quietly and well, without half the userland parts being tied to it.
Bitching about systemd is great and all, but don't pretend init was standard prior to systemd init being shoved down our throats by Poettering and pals. It just seemed like it if you never switched distros or looked too closely at the parts under the hood.
---
Finally, it's also not fair to argue that this is the albatross around Linux's neck, without also mentioning that Microsoft and Apple have done their share of wheel reinvention and "screw it let's make everyone deal with massive system changes" as well, sometimes in ways that required complete rewrites of software or purchasing of new hardware. Times change and the software changes with it, sometimes requiring rewrites.
Windows 9x to XP was a major architecture change, as was XP to Vista, which was the source of many driver problems and incompatibilities and software breakage. They've also done things like switch from MFC to pushing .NET, and that point where they doubled down on Silverlight, just to later pretend it doesn't exist.
Apple did similar switching from OS 9 to OS X, Carbon to Cocoa, abandoning powerpc completely for 32bit x86, then again from x86 to 64-bit. Possibly other things, but I'm not as familiar with their history.
(Score: 2) by Arik on Friday February 20 2015, @06:38PM
No, I'll blame the distros for their decisions, and still have plenty left over for the KDE team calling an alpha '4.0' as well as for abandoning a system that was nearing maturity and starting over for no compelling reason and with no thought whatsoever for the users they screwed over in the process.
"Bitching about systemd is great and all, but don't pretend init was standard prior to systemd init being shoved down our throats by Poettering and pals. It just seemed like it if you never switched distros or looked too closely at the parts under the hood."
Sounds like you completely missed the point on systemd. Yes, we had (and have) many different init systems - THAT IS GOOD. They all play nicely except for systemd, and THAT is the biggest problem with systemd - not that it is Yet Another Init System but that it is the one that is by design hostile to all the things that make *nix systems worth using, and refuses to play nicely with the ecosystem. It's a cancer.
Linux would be fine if someone would just be a dear and kill Redhat.
If laughter is the best medicine, who are the best doctors?
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @09:48AM
Yep. At this point in time talking about systemd as a "mere" init is a distraction. It has pretty much reimplemented anything not coreutils.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @08:02PM
And by "shoved down our throats" we mean, "chosen by the people who make the distros we pretend to love when not calling them names."
Chosen on the technical merits by the person in charge of the decision doesn't exactly scream "shoved" or "down our throats." It actually seems to be the correct way for the decisions to be made.
Just choose a different distro. Some people are pointing out that their complaints to systemd are philosophical, not technical. I couldn't agree more! There are still people pretending to have technical complaints, but they can't manage to find any that aren't already on the "myth list." They also can't manage to make any philosophical point that doesn't include pejoratives and argumentum ad hominem. To be honest I haven't even heard a "philosophical" complaint about systemd that isn't either based on hatred of individual humans involved, or else busted myths. A philosophy of hate is still a philosophy, but it doesn't seem very likely to persuade educated professionals working technical jobs for distros.
(Score: 3, Interesting) by arashi no garou on Friday February 20 2015, @09:10PM
No hatred for anyone on my part, I just don't care for politics in my software. Back when Pat Volkerding cut Gnome out of Slackware, he did so for very technically valid reasons. Even so, Slackware lost a lot of users over it, and endured a bit of hate. I stopped using Slackware for a while not because Gnome was cut (I was using Fluxbox at the time), not because I suddenly hated Pat, but because I got sick of all the fighting and bullshit surrounding it. I figured it was the right time to explore other distros for mainstream use, and I'm glad I did. I discovered Arch Linux (which, again, I dropped later on when there was a big political row in the community), and explored tons of distros I wouldn't have bothered with if I had stayed for the drama. I eventually settled back into Slackware once everyone stopped caring about the Gnome debacle (probably largely due to how much Gnome 3 sucked and how difficult it was to compile and package - Pat's original argument, finally validated by the masses).
Now we have Debian making the decision to adopt systemd, a project fraught with controversy and political shenanigans. Again, people are leaving Debian in droves (both users and devs) over something political. Sound familiar? It does to me, and I am personally better off without it. But I won't tell other people "systemd is bad m'kay" because it may not be bad for them. I've only ever given reasons it's bad for me.
Chosen on the technical merits by the person in charge of the decision doesn't exactly scream "shoved" or "down our throats."
That's just it: It was chosen by "the person", not "the people". "The person" drew a line and ignored dissent from anyone who tried to challenge that line. At least, that's what I've read; if it went down differently than it was reported, please enlighten me.
(Score: 2, Insightful) by lizardloop on Friday February 20 2015, @04:42PM
The reason you see so much reinventing is because reinventing is easy. You can look at the existing project, spend a while thinking about how the code might be better organised or a particular feature that might be added and then you can start. Any problems you run in to can usually be solved by referring back to the existing project. It takes a very particular intelligence to come up with a truly novel project and implement it in a manner that isn't completely buggy. Most developers aren't that bright (I'm one of those not so bright developers).
(Score: 5, Interesting) by Thexalon on Friday February 20 2015, @04:57PM
An important difference between systemd and the OSS->ALSA switch: ALSA maintains backwards compatibility tools for OSS to this day. By contrast, systemd very intentionally does not - once your system is going to be systemd-based, you have to switch everything. They use this lack of backwards compatibility to force all userspace software that runs any kind of daemon to be for systemd-based systems only, and if a userspace daemon fails to play ball they create a new one that depends on systemd and bully distros into accepting the new daemon. Then they target any widely used software that depends on userspace daemons to force them into depending on their new daemons rather than the older non-systemd ones.
I'm still not entirely clear why everyone's going along with this. The only answers I've gotten are:
1. Sysvinit was antiquated, systemd is modern. What that tells me is that sysvinit and its derivatives have about 20 years worth of field testing and debugging that systemd lacks. When it comes to software, old doesn't necessarily mean bad.
2. Boilerplate init scripts suck Lots of problems with this one:
(a) the scripts weren't boilerplate, but varied from host to host and daemon to daemon
(b) it ignored the flexibility that you gain from being able to change service startup and shutdown by altering what was usually a pretty clear bash script
(c) that same flexibility decoupled the init system from nearly all other userspace packages, so the init system could be replaced without affecting the daemons
(d) any boilerplate that you did have could be turned into a bash function and added to a library script that you included
3. Systemd means faster boot times This one is easy to knock out too:
(a) Faster startup is basically a pointless task to focus on, because booting up is one of the least common activities a system engages in - sure, if it's 15 minutes it's a problem, but the difference between 15-second and 45-second boot time is basically immaterial
(b) Systemd has not been shown to be consistently faster than sysvinit
As far as I can tell, the Linux development community is not only reinventing the wheel, they're trying to make the new wheels square just for kicks. They've been crippling the init system, the desktop environments, etc, all for no good reason. Now, I realize that since I'm older than 30 that qualifies me to be an old curmudgeony luddite, but if your users are longing for the days of previous versions of your software, you have a serious direction problem.
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 2) by hemocyanin on Friday February 20 2015, @10:14PM
Wasn't the Gnome v. KDE thing originally about some non-free license issues with QT?
(Score: 3, Interesting) by fnj on Friday February 20 2015, @04:32PM
Does this mean PC-BSD will have to find a new DE? If they have a clue, they will throw a lot of resources into Lumina to get it ready faster.
(Score: 2) by doublerot13 on Friday February 20 2015, @06:43PM
They will just use XFCE by default. I think you can select it at install time anyway.
I actually prefer it to KDE.
(Score: 2) by arashi no garou on Friday February 20 2015, @09:17PM
Doubtful. They are building their own desktop called Lumina, based on QT and Fluxbox. I'm pretty sure it's going to be the default if KDE gets dropped.
http://wiki.pcbsd.org/index.php/Lumina/10.1 [pcbsd.org]
(Score: 2) by fnj on Sunday February 22 2015, @12:29AM
Correct; and Lumina looks more important as Razor-Qt and LXQt have slowly soured on the vine.
(Score: 0) by Anonymous Coward on Friday February 20 2015, @09:50PM
From the blog:
Realistically we're talking about the snapshot of the login1/timedate1 specs as of now. I follow what happens in that BSD shim. We tend to have a good relationship with most packagers (certainly I do with one BSD guy) if he tells us something is broken, we'll fix it in the final release.
We had a discussion at the Plasma sprint now, we'll be dropping legacy support in 5.5; over 6 months from now. We're trying hard to not screw any hard working distros over.
I can't promise we won't break it accidentally, we do it with the build system. I can say we won't do it deliberately; and certainly this should be done in a way that won't affect users.
However, if a distro doesn't make any effort then they will get left behind with an old Plasma. I don't want to let any one group ruin the experience of anyone else.
(Score: 2) by fnj on Sunday February 22 2015, @12:35AM
I am a hapless Arch user who recently got shafted by updates bringing in horrible KF5 versions of Kate and Konsole[*]. I am so pissed right now, all the old rage at abandoning 3 in favor of a ghastly immature 4 is coming back.
All my painstaking customizations were suddenly gone and the beautiful KDE4 apps turned into ugly KF5 versions. Kate became almost unusable, and Konsole became completely unusable.
I admit I, in a sense, was "asking for it" by using Arch, but at this point I just want to drive a stake through KDE.
(Score: 4, Insightful) by Anonymous Coward on Friday February 20 2015, @07:00PM
It's already over, systemd won a long time ago. It won because most people are willing to complain but only a very small percentage are willing to take action (like ditch Linux for BSD). Hell, even if the whole developer base jumped ship, systemd would still continue on because you're not the customer! Governments and businesses are the customer and Red Hat only listens to those who pay. At this point, meaningful change can only come from system admins educating their clients (businesses and governments) on why avoiding systemd is a good idea. System admins are in the best position to affect change...
I recently rolled out several new BSD systems for one of my clients. I didn't ask them which OS I should pick because they couldn't care less about it. I chose FreeBSD because, well, fuck systemd. It took longer because I'm inexperienced with BSD and had to read up and work around some compatibility issues. Now my client has their new systems and they are happy. I avoided systemd and got paid so I'm happy. If enough IT folks begin silently migrating to BSD when convenient (i.e. replacing aging systems or adding new systems), eventually Red Hat could be forced to care about it (or at least acknowledge it).
Sometimes removing the turd from the punch bowl isn't enough to save the punch. In this case, Linux is the refreshing punch that we all wanted and Red Hat are the drunken ass who just shat in it. Real funny RH... Now please go sleep it off and, oh yeah, you're not invited to the next party...
(Score: 0) by Anonymous Coward on Friday February 20 2015, @10:50PM
Actually, it is BSD that has already one. BSD gained many former Linux users who cares about init system flexibility, the bazaar model, and avoiding dependency hell. These are people who are more likely to contribute back to the distribution and upstream projects, unlike many of the people who stayed with systemd distributions. With BSD's recent gain in users with software development talent, they have more capability than ever to run KDE there, a KDE like desktop environment, Lumina, or the Trinity Desktop Environment. systemd has not won anything of value. systemd has actually lost something of value: They have chased off development talent to other software distributions and sysytemd is building a monstrosity that will be difficult for future development talent to maintain. This will not be realized now, but years from now, it will become more evident.
As this is a win for BSD, this is also a win for permissive free software as opposed to copyleft software. Copyleft is better at protecting the freedom of the user, as opposed to the software publisher.
If you want to stick with Linux and KDE, I suggest Devuan and the Trinity Desktop Environment.
(Score: 1, Interesting) by Anonymous Coward on Saturday February 21 2015, @10:27AM
I am an *nix administrator for a federal agency (and have worked for state and large corporations) and I can tell you that you are 100% wrong. The customers don't know or care about systemd. They aren't clamoring for it. They just want reliable secure machines, RedHat is pushing this all on their own.
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @10:00AM
"Secure" seems to be the operating word here.
Supposedly one reason for Poettering to start on systemd was his growing interest in daemon and network security after developing pulseaudio (it can route sound over the network) and avahi (yep, thats his).
He was also involved with consolekit for a while, and pretty much the person that declared it dead in favor of logind. This because of some apparent corner cases with the consolekit seat tracking.
"Security" seems to also be what is pushing the development of containerization in Gnome (depending heavily on systemd for handling the nitty gritty stuff) and in part Wayland.
And all of that seems to be germinating in Fedora, that happens to be what RH is skimming ever so often to spin a new RHEL.
Frankly all this brings to mind the old computer security joke about the secure computer being one encased in concrete at the bottom of the ocean. And likely the netsecs driving these changes don't get the joke...
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @07:14PM
Red Hat develops new tech based on their own experience working with clients. It is very likely that they see an opportunity for systemd to solve some real issues AND as a way to disrupt (or exert some control) over a wider selection of distros. To me, systemd makes a typical distro more Centos-like; which is good if you're Red Hat. It also slowly locks software to systemd/Linux as dependencies grow stronger. Remember, the community was chastised when it first speculated that unnecessary dependencies would be introduced to further adoption. Then Gnome, Gimp, and now KDE (among others) have all announced dependencies of some kind. More will follow...
(Score: 3, Interesting) by r00t on Saturday February 21 2015, @12:00PM
Keep an eye out for the Debian fork 'Devuan'[1]. It's coming along nicely. A sneak-peek pre-Alpha ISO was posted to the mailing list this month to demonstrate the Devuan system booting into the desktop. It was beautiful to see the geen '[OK]' sysVinit messages again during boot and the XFCE desktop settling in at the end.
A well-done video on the Devuan/Linux problem. Probably NSFW due to language:
https://www.youtube.com/watch?v=VSbNumR9Z8k [youtube.com]
(Score: 0) by Anonymous Coward on Sunday February 22 2015, @09:16PM
I don't know what to think of that video... However, Devuan is a knee-jerk reaction that likely won't succeed. Systemd represents death (or in this case, change) by a thousand paper-cuts (dependencies). As projects like KDE adopt systemd, they also abandon or remove legacy code supporting alternatives. As more projects follow, Devuan will find itself overcome with patching them all to work without systemd - or they'll be forced to continuously update shims to maintain the systemd-like bits that allow dependent apps to function. As complexity continues to increase, the shims will begin to look more like full replacements for systemd until there are a bunch of clones floating about; none of them being a perfect solution. Eventually people will say 'fuck it' and just adopt the real thing.
You don't stop systemd by forking distros because that just weakens one project while also creating an understaffed new project. In fact, the time to stop it has already passed... If you disagree with systemd and are in a position to make decisions then take the plunge and choose BSD. Most customers won't notice or care and you may actually enjoy working with a different, yet oddly familiar, operating system.
Linux was once appealing because it was developed by 'the people' rather than the corporations. Sadly, that romantic ideal hasn't been true for a very long time. Aside from smaller projects, Linux development is very much driven by corporate interests today. As others have mentioned, systemd is another step in the 'Windows-fication' of Linux (think svchost) which actually makes some sense if the roadmap is influenced by companies. Systemd hits a lot of birds with a single stone: consolidation of power, unification of services, a single standard to support, competitive advantage (RH has access to undocumented features and can easily add whatever they want; others must wade through hoops and red tape), and it also disrupts the other distros; allowing RH an opportunity to surge ahead while the rest scramble to adopt without losing users. A Debian install offered no direct benefit to Red Hat. A Centos install was less-bad because it closely tracked RHEL and providers an easier transition path to RHEL than other distros. In this regard, all systemd distros will soon be more like Centos (easier to support, quicker path to RHEL for up-sells)...
Init worked fine before so that's not the real reason for systemd. Many addon cards for servers have their own BIOS that often delay system boot by up to several minutes (such as RAID cards) so faster boots also are not the real reason either. Swapping out the core guts of every major distro with bits developed by Red Hat (and securing that position with weak dependencies), well... that's a really good reason for systemd to exist. It almost borders on embrace, extend, extinguish; only this time it's Red Hat instead of Microsoft...