systemd-free Devuan Linux hits RC2

Devuan Linux has released its second release candidate. Devuan was created by self-described “Veteran Unix admins” who find Debian's adoption of systemd abhorrent, because they want complete control over the packages that load when Linux boots. “Devuan decided to fork not only the base distribution, but also its governance,” …

[Christian Berger]

It's fascinating that Linux now has the same problem as Windows

I mean Windows has a rather good kernel team, they actually work on making their kernel faster and more secure every day. The reason why people who dislike Windows dislike Windows is the userspace. It's all that crappy software which spies on you or prints PDF-files differently to how it displays them: https://developer.microsoft.com/en-us/microsoft-edge/platform/issues/11896203/

Now we have the same on Linux. There is a team which tries to make the kernel faster and more secure, but there are also teams who make the userspace more complex, buggy, harder to fix and insecure. Often even by doing exactly what Microsoft did wrong 20 years ago, like binary log files and highly opaque service management.

[jake]

Re: It's fascinating that Linux now has the same problem as Windows

No, Linux does not have this problem. systemd is not in the Linux kernel. Rather, several distributions of Linux have chosen to include this problem. There are other distributions which choose not to have this problem. Which distribution to choose is up to the individual. Me, I choose a distribution without the problem. YMMV.

[Christian Berger]

Re: It's fascinating that Linux now has the same problem as Windows

"No, Linux does not have this problem. systemd is not in the Linux kernel."

Yes, you are correct, and you can still choose distributions without systemd. For some platforms you can even go further and use completely different userlands like gokrazy, which attempts to simply boot into a single service on the Raspberry PI3.

However you could also argue that you can get Windows PE which leaves out most of the crap you don't want. The problem there however is that Windows programmers tend to have used every obscure feature there was. You'll still have programs using VBX components. You'll still have programs using OLE and DCOM. Developers simply aren't disciplined enough to not use new features dangling in front of them.

[jake]

Re: It's fascinating that Linux now has the same problem as Windows

Not sure what your point is. WinPE and gokrazy are hardly general purpose desktop computing environments. Devuan is. In fact, the only difference between Devuan and Debian is that one uses systemd, the other doesn't. If you are trying to suggest that a distribution without systemd is somehow crippled ... Well, don't even go there as you'd be very, very wrong.

[Ole Juul]

Re: It's fascinating that Linux now has the same problem as Windows

There really is nothing available which is perfect. However, I agree with the title as do many other people - at least in principle and feel. I note that a lot of *BSD users are Linux refugees who've not liked the Windows-like direction of Linux distros in general. You may disagree with the choice of BSD but the sentiment is common.

[naive]

Re: It's fascinating that Linux now has the same problem as Windows

I could not agree more. Only this year I implemented 15 NetBSD systems for customers, instead of Linux.

It is such a liberating experience. To live without all this complicated stuff they implemented in Linux, not only systemd, also basic issues like configuring network cards gives one the same Voodoo feeling as dealing with windows does.

In NetBSD one can do all in /etc/rc.conf and /etc/rc.local, what you type is what you get, that has always been the idea behind Unix. In NetBSD rm is rm, and not a "are you sure" question.

Since NetBSD allows me to build compact and effective solutions for most customer demands, I can only laugh at the path downhill Linux is going. It would interesting to analyze who hijacked a good idea to turn it into an overly complicated dragon of complexity and an implied massive attack surface.

[DrXym]

Re: It's fascinating that Linux now has the same problem as Windows

"even by doing exactly what Microsoft did wrong 20 years ago, like binary log files and highly opaque service management."

Binary log files are actually there to make the system more secure since they store more data, index data, rollover properly, detect corruption and tampering through signing. If you want text files in addition to or instead of binary it is a simple matter of reading the man page to enable them if you so desire.

[Doctor Syntax]

Re: It's fascinating that Linux now has the same problem as Windows

"Binary log files ... detect corruption"

by becoming unreadable and therefore useless.

"and tampering through signing."

If you're concerned about tampering log to a remote host. That's a problem that was solved years ago.

"If you want text files in addition to or instead of binary it is a simple matter of reading the man page to enable them"

Not quite so simple. One of the times you really need logs is in diagnosing a system that's not booting properly. In such circumstances an original text log is readable by booting from another medium and your binary log probably isn't; how far behind current was the translation to text and how do you know?

Systemd: bringing you problems you don't need to solve problems you don't have.

[HieronymusBloggs]

Re: It's fascinating that Linux now has the same problem as Windows

"If you want text files in addition to or instead of binary it is a simple matter of reading the man page to enable them if you so desire."

The problem if you don't want binary logging is that it can't be removed, only redirected. Corrections welcome if I'm wrong.

[Anonymous Coward]

Re: It's fascinating that Linux now has the same problem as Windows

If you wish to see logging in free text as well as binary, install rsyslog and the binary logging will be duplicated into /var/log/messages (or the equivalent)

[Doctor Syntax]

Re: It's fascinating that Linux now has the same problem as Windows

"If you wish to see logging in free text as well as binary, install rsyslog and the binary logging will be duplicated into /var/log/messages"

Why not do things the other way round, default text, binary an option for those who want it?

[Apprentice of Tokenism]

Re: It's fascinating that Linux now has the same problem as Windows

If you wish to see logging in free text as well as binary, install rsyslog and the binary logging will be duplicated into /var/log/messages (or the equivalent)

Oh really? Wanna see a prime example about the clusterfuck that systemd, journald and all the abominations that come with it are? Here ya go (copied from a real system running rsyslogd):

+++++ snip +++++

May 8 09:28:05 foobar clamd[1147]: Database correctly reloaded (6278511 signatures)

May 8 09:28:36 foobar packagekitd[2145]: PARENT proccess running...

May 8 09:27:16 foobar rsyslogd: [origin software="rsyslogd" swVersion="8.4.2" x-pid="940" x-info="http://www.rsyslog.com"] rsyslogd was HUPed

May 8 09:28:36 foobar rsyslogd0: action 'action 14' resumed (module 'builtin:ompipe') [try http://www.rsyslog.com/e/0 ]

May 8 09:28:36 foobar rsyslogd-2359: action 'action 14' resumed (module 'builtin:ompipe') [try http://www.rsyslog.com/e/2359 ]

May 8 09:28:58 foobar systemd[1]: Reloading.

+++++ snip +++++

Why there is a log line that has a time stamp in the past is completely beyond me and I do not want to hear any "yap yap it is rsyslogd's fault! yap yap".

What I do know though is that I would never have had this with good old plain text logs. So please stop telling all this bullshit. I am so sick of it.

[bombastic bob]

Re: It's fascinating that Linux now has the same problem as Windows

"it is a simple matter of reading the man page"

RTFM noted, not appreciated.

[wayward4now]

Re: It's fascinating that Linux now has the same problem as Windows

Amazing all the down votes you got by merely giving good information. That says much about the anti-systemd folk.

[GreenBit]

Re: It's fascinating that Linux now has the same problem as Windows

Don't you think people are entitled to their own opinions without someone making snarky demeaning insinuations about their personal character? Just saying.

[Warm Braw]

Re: It's fascinating that Linux now has the same problem as Windows

binary log files and highly opaque service management

I'd agree with you about the opacity of the service infrastructure (what are all those svchost processes actually doing?). However, binary configuration files and binary log files are inevitable: most of the world does not speak English and a bug chunk of it doesn't use a language that can be represented with the ASCII character set. Text belongs in the localised user interface: saying it's somehow "more convenient" to have the computer parse, or serve up, English is just a form of cultural imperialism.

[Paul Crawford]

Re: It's fascinating that Linux now has the same problem as Windows

"However, binary configuration files and binary log files are inevitable"

No they bloody well are not!

If, as you seem to think, that English is some imperial conspiracy then why do we not program in binary? Why do all major languages use it? Why is it the most common language in the world (mostly as the 2nd spoken choice)?

And if you need to translate binary to/from some local readable format, why not translate English/ASCII in the same way? Fundamentally providing a language-agnostic system is very hard work and you then lack any simple way to interact with it for development with just a text editor.

[jake]

Re: It's fascinating that Linux now has the same problem as Windows

You call it "cultural imperialism", I call it lingua franca.

Like it or not, just as Arabic drove early science, Latin (Koin Greek, Aramaic, et al) drove early Christianity, Deutsch drove later science (etc., I won't continue. You are quite welcome), American English is the lingua franca of tehintrawebtubes.

Trying to make it otherwise is probably similar to swimming against the Boston Molasses Disaster.

[PNGuinn]

Boston Molasses Disaster @ Jake

And note that this is a moving target.

'Merican is rapidly giving way to Indlish and Chindlish.

Advice from your former masters: Learn to play cricket - it's the only way.

Or realise the error of your ways, say sorry and return to the fold. Downside - you'll have to replace POTUS Trump with PM TM.

Hmmm - Joke or troll icon - Methinks I'll go for the latter.

[jake]

Re: Boston Molasses Disaster @ Jake

I play cricket. And rugby. And snooker. I drink bitter. I even have O and A levels, and spent my first two years of Uni at Kings College ... I've lived roughly 20% of my life in Blighty.

Bit late to "return to the fold", methinks ;-)

[Bronek Kozicki]

Re: It's fascinating that Linux now has the same problem as Windows

@Warm Braw I call BS. UTF-8 character encoding of text files is both long established and universally supported in Linux, so you can have national characters in text without resorting to binary format.

[Paul Crawford]

Re: It's fascinating that Linux now has the same problem as Windows

UTF-8 is great, but please don't be a muppet like the systemd lot and have your program crash if a non-UTF-8 character is used in a *comment*...

[Warm Braw]

Re: It's fascinating that Linux now has the same problem as Windows

you can have national characters in text without resorting to binary format

You can, but it then makes no sense to someone from another cultural background

providing a language-agnostic system is very hard work

It isn't really that much more work - you simply need a level of indirection to allow language-neutral tokens to be translated appropriately to the locale. You could even provide that in a common system that applications could share. Rather like Windows, MacOS, ...

American English is the lingua franca of tehintrawebtubes

This is just a circular argument. If you have to speak some level of American English in order to work with computers, then only people that speak some level of American English will work with them. That hasn't solved the problem.

why do we not program in binary

The last time I looked, executables were architecture-specific binaries, not human-readable. A compiler translates one into the other and there's no reason at all why the compiler cannot compile a local variant of a programming language (eg АЛГОЛ 68) in exactly the same way that a local variant of a configuration file can be compiled into a culturally-independent one. There's a fuller list of non-English programming languages in the usual place.

[jake]

Re: It's fascinating that Linux now has the same problem as Windows

It hasn't solved the problem because the problem clearly doesn't actually exist in the real world.

[Bronek Kozicki]

Re: It's fascinating that Linux now has the same problem as Windows

Any time I see pathetic attempts to use localized resource files with ID in place of actual readable message I am thankful my parents started teaching me English when I was a child.

The only good programming language with "Polish" in it is RPN. And it's not even a language.

(mine is the one with Polish passport in the pocket)

[HieronymusBloggs]

Re: It's fascinating that Linux now has the same problem as Windows

"A compiler translates one into the other and there's no reason at all why the compiler cannot compile a local variant of a programming language (eg АЛГОЛ 68) in exactly the same way that a local variant of a configuration file can be compiled into a culturally-independent one."

How would you apply this to a multi-national collaboration like the Linux kernel (or any other large FOSS project for that matter)? Sometimes a lingua franca is the only practical solution.

[Uffish]

Re: It's fascinating that Linux now has the same problem as Windows

@ Warm Braw - I think you are ignoring the facts and history behind the term 'Lingua Franca' .

My experience of working with 'foreigners' is that British, Australian, Nigerian, American, and just about anyone's English is good enough if you can't use anything else. In other words people use their own language (Spanish, Chinese etc) where possible and English when that is not effective. Nobody seems to get upset by this arrangement except politicians and bureaucrats.

My experience with Devuan RC1 is that you can set it to use the keyboard nationality that you want but the login page stays resolutely in US keyboard layout, which makes strong passwords somewhat fiddly to type.

[Doctor Syntax]

Re: It's fascinating that Linux now has the same problem as Windows

"English is just a form of cultural imperialism"

So should programming languages be translated from English - "if", "else", "break" etc translated into your local language?

[Zolko]

Re: It's fascinating that Linux now has the same problem as Windows

So should programming languages be translated from English - "if", "else", "break" etc translated into your local language?

Actually, it has been done, with BASIC, to French: it's horrible, and unreadable. A very good candidate for "bloody Hell, did you really think this will be used ?"

[GrumpenKraut]

Re: It's fascinating that Linux now has the same problem as Windows

> So should programming languages be translated from English - "if", "else", "break" etc translated into your local language?

Believe it or not, this has been done. "for" was translated to the German "für". This at a time when everything beyond ASCII was largely an undefined minefield. Much non-hilarity ensued.

[Ramazan]

Re: "if", "else", "break" etc translated into your local language?

They used to do it in the USSR, but this practice is now obsolete. So the answer is, "You can do it but eventually no one will use it".

[Ramazan]

Re: It's fascinating that Linux now has the same problem as Windows

"I mean Windows has a rather good kernel team"

Maybe, even most probably it's true, but their kernel isn't so good. It's not even that the kernel itself is no good, it's its Win32 API to userland, especially when compared to POSIX.

[Jim 59]

Re: It's fascinating that Linux now has the same problem as Windows

I could upvote Christian's post until my mouse key wears out.

[Adam 52]

The question I keep asking myself is if systemd is so bad, why have all the major distos except Slackware and Gentoo adopted it?

[Graham Dawson]

Politics.

[jake]

Easy answer.

Children running the show. No seasoned un*x hack would look at systemd and say "This is a good idea! I want THAT!".

[Pompous Git]

Re: Easy answer.

"Children running the show."

Hmmm... I kinda like Cinnamon Mint. Perhaps developed by very clever children, but then I'm only 66 years young meself ;-)

I seem to recall reading that motor cars would never replace horses.

[i1ya]

Re: Easy answer.

Pomous Git, heads up! Downvotes incoming: you left pro-systemd comment on TheRegister!

[Pompous Git]

Re: Easy answer.

@ i1ya I get around one downvote for every 6 upvotes. Think how boring it would be if we all thunk the same :-)

Frankly, I have no real idea what the fuss is about. When it comes to OSs I view things from the perspective of an end user. Does it run stable and does it run my applications? Implying I'm childish for wanting that seems rather petty. Now the downvotes will come thick and fast [evil grin]

[jake]

Re: Easy answer.

The "fuss" is that systemd doesn't solve any problems. It doesn't actually fix anything. What it does do is add complex code unnecessarily, which is totally against the entire design philosophy of un*x. I speak as both a jaded old sysadmin, and as an end user.

Linux is stable, and runs your applications just fine without systemd.

Note I never said the users of systemd based distros were childish.

[Anonymous Coward]

Re: Easy answer.

> When it comes to OSs I view things from the perspective of an end user. Does it run stable and does it run my applications?

The lack of downvotes on your post is probably because it's not a pro-system post, even though you seem to be thinking it is. ;)

[Pompous Git]

Re: Easy answer.

"The lack of downvotes on your post is probably because it's not a pro-system post, even though you seem to be thinking it is. ;)"

I am neither pro, nor anti systemD. I am however quite entertained by the discussion :-)

[jake]

Re: Easy answer.

Motor cars have not replaced horses.

[Pompous Git]

Re: Easy answer.

"Motor cars have not replaced horses."

Really?

Late 19th C London Omnibus

[jake]

Re: Easy answer.

Yes, really. I took my Percheron & buckboard to the feedstore to purchase chicken feed just this afternoon. Or, rather Clyde (the horse) took me to the feedstore ... so no, horses have not been replaced.

[Pompous Git]

Re: Easy answer.

"Yes, really. I took my Percheron & buckboard to the feedstore to purchase chicken feed just this afternoon."

You're a man after my own heart, Jake :-) However, just because some few of us preserve the old ways, doesn't mean that the vast majority give a damn.

[Ramazan]

Re: Easy answer.

> > "Motor cars have not replaced horses."

> Really?

When had they started to put car's meat into суджук, махан and казы?

[John Robson]

Re: Easy answer.

"I seem to recall reading that motor cars would never replace horses."

And look at the mess they have caused.

Clearly we just don't know what's good for us...

[Pompous Git]

Re: Easy answer.

"And look at the mess they have caused."

Motor cars solved the mess problem.

"The main problem, however, was manure. A horse produces between 7 and 15 kilos of manure daily. In New York in 1900, the population of 100,000 horses produced nearly 1,200 metric tons of horse manure per day, which all had to be swept up and disposed of. In addition, each horse produces nearly a litre of urine per day, which also ended up on the streets.

....

Writing in the Times of London in 1894, one writer estimated that in 50 years every street in London would be buried under nine feet of manure."

[jake]

Re: Easy answer.

You see manure. I see next year's veggies.

[Pompous Git]

Re: Easy answer.

"You see manure. I see next year's veggies."

No, we both see next year's veggies. For a decade I made 25 cubic metres of compost a year for my market garden. It's just that in the late 19th C there was a vastly greater amount of manure in New York, London, Paris, Sydney etc. Far more than was needed for vegetable production. The piles of manure on vacant blocks in NYC were reported to be up to 20 metres high.

[John Robson]

Re: Easy answer.

"The main problem, however, was manure. "

Whereas now the main problems are:

- Death through impact (several thousand a year in the UK, many more globally)

- Death through obesity/diabetes (and other symptoms of a sedentary lifestyle)

- Death through air pollution

- Disease through air pollution

- Disease from sedentary lifesytle

- Injuries from collisions.

- Communities isolated from each other

- Kids never seeing the sky

The old problem was too much free fertiliser?

That's not actually that bad a problem - it might take some collection, but that's not too difficult - a manure bag whilst out and about and semi regular 'deposition' points for emptying them.

[Doctor Syntax]

Re: Easy answer.

"Death through impact"

People have been known to die in horse-related accidents. An ancestor of mine died falling from his horse. Unfortunately re-engineering the horse to make it intrinsically safer is a bit trickier than re-engineering a car.

[Pompous Git]

Re: Easy answer.

"People have been known to die in horse-related accidents."

The death rate from horse accidents in the major cities in the 19th C was considerably higher than that from motor cars today. Possibly because falling off a motor car is considerably rarer than falling off a horse. The only horse I know of that carried humans internally was built in Troy...

[Doctor Syntax]

Re: Easy answer.

"I seem to recall reading that motor cars would never replace horses."

And look at the mess they have caused.

If you watch some period drama from the days of the horse look at the nice clean streets and ask yourself if the historical reconstruction was accurate.

[Pompous Git]

Re: Easy answer.

"If you watch some period drama from the days of the horse look at the nice clean streets and ask yourself if the historical reconstruction was accurate."

What's even more amazing about those period dramas is there's not only no horse manure, there's not a sparrow starver in sight!

[wayward4now]

Re: Easy answer.

""I seem to recall reading that motor cars would never replace horses."

And look at the mess they have caused."

At least motor cars don't shit in the street for you to step in.

[Doctor Syntax]

Re: Easy answer.

"I seem to recall reading that motor cars would never replace horses."

News must have been delayed where you were brought up. I'm a good bit older than you and cars had already replaced horses here.

[Pompous Git]

Re: Easy answer.

"News must have been delayed where you were brought up. I'm a good bit older than you and cars had already replaced horses here."

I was brought up in Nuneaton, Warks. Working class so new books were a luxury denied. Most of the books I read were published in the 19th C. Heck, the words of Aristotle I read were written ~2,500 years ago!

[DrXym]

Re: Easy answer.

"Children running the show." Yeah all the people in Red Hat, Ubuntu, SUSE, Debian et al are children. What on earth do they know about producing enterprise ready, stable operating systems? Oh.

[Paul Crawford]

Re: Easy answer.

"What on earth do they know about producing enterprise ready, stable operating systems? "

Like RHEL 6 that is without systemd, perhaps?

[DrXym]

Re: Easy answer.

"Like RHEL 6 that is without systemd, perhaps?"

RHEL 6 was first released in 2010. RHEL 7 was first released in 2014 does use systemd. Failing to grasp your point.

[jake]

Re: Easy answer.

Yes, children. From a technical perspective. RedHat is run by management. Ubuntu is a derivative distro run by management. SUSE is run by management. Debian is being taken over by management. I'll stick to a distro where the decisions are made by technical folks, thanks.

Perhaps ask yourself "Why isn't BSD jumping on the systemd concept?". Or maybe "Why did those veteran Debian developers leave to start Devuan?" I mean, if it's the great technical panacea that you seem to think it is, surly the BSD crowd would be all over it, and the veteran techs wouldn't have left Debian, right?

[Anonymous Coward]

Systemd is a solution to a messy and fragmented process that was well overdue an overhaul, modern software requires modern solutions which is something rabid luddites don't seem to understand.

[Anonymous Coward]

Do one thing and do it well

Principal of least privilege

Change for change's sake

[wolfetone]

"Systemd is a solution to a messy and fragmented process that was well overdue an overhaul, modern software requires modern solutions which is something rabid luddites don't seem to understand."

Quite right. And it's solution is to solve the messy and fragmented process you described by providing a bloated, horrible, vastly more complex system to remedy all that. And to make sure it solves all the problems, it doesn't provide proper compatibility with the system it replaces.

Wait until systemd comes with a word processor. Then you'll wish it never existed.

[Doctor Syntax]

"Wait until systemd comes with a word processor."

You mean it doesn't?

[dajames]

Systemd is a solution to a messy and fragmented process that was well overdue an overhaul, modern software requires modern solutions which is something rabid luddites don't seem to understand.

Rabid luddites aside ... the problem with systemd is that it isn't a good solution to the perceived problems that it was intended to solve. The old sysvinit system may be messy and fragmented, but it works and is well understood.

Systemd is less messy, in many ways, but it doesn't (yet) work as well. If that were the only problem we could put it down to teething troubles and hope that future versions would be better, but systemd also brings problems that sysvinit doesn't have.

Systemd is a single, monolithic, system rather than a set of interoperating modules, so it goes against the Unix/Linux philosophy of making each job the responsibility of one tool that does that job well. Systemd does some jobs fairly well and others fairly badly, but doesn't allow any possibility of using its good parts and replacing its bad ones.

Systemd also has its dirty fingers into other parts of the system. As a replacement for sysvinit is is supposed to be an init system, but because its scope goes far beyond the initialization phase (and it doesn't let you take the good without the bad) it has become a dependency for many userspace programs that should never have any reason to interact with the init system at all, making it harder to use those programs on a non-systemd system.

Systemd is an insidious piece of malware, and the community needs to recognize this and expunge it, before it is too late.

Yes, the sysvinit system is getting old and creaky, and needs an overhaul or possibly a replacement ... but systemd is not an improvement.

[DrXym]

"Systemd is a single, monolithic, system rather than a set of interoperating modules,"

This sentence right there demonstrates the ignorance surrounding systemd. Systemd is actually a collection of processes each with specific, isolated function that are designed to run together with minimal privilege.

[wolfetone]

"This sentence right there demonstrates the ignorance surrounding systemd. Systemd is actually a collection of processes each with specific, isolated function that are designed to run together with minimal privilege."

And this statement right here demonstrates someone blindly reading the crap that's written about Systemd.

It's supposed to be compatible with v int scripts and commands. But guess what? In the real world: that doesn't happen.

[DrXym]

"And this statement right here demonstrates someone blindly reading the crap that's written about Systemd."

No, it's a statement of someone who's bothered to read about systemd, use it, and has the ability to type

dnf list systemd\*

rpm -ql systemd | grep bin

[wolfetone]

"No, it's a statement of someone who's bothered to read about systemd, use it, and has the ability to type"

How much is Lennart paying your blind affiliation to something which doesn't solve any problem anyone was suffering from yet is setting the scene for catastropic problems in the future?

[DrXym]

"How much is Lennart paying your blind affiliation to something which doesn't solve any problem anyone was suffering from yet is setting the scene for catastropic problems in the future?"

Yes it's all a vast conspiracy. Meanwhile back on planet earth...

[HieronymusBloggs]

"No, it's a statement of someone who's bothered to read about systemd, use it, and has the ability to type"

DrXym, what makes you think that doesn't also apply to those here who don't want systemd on their systems? You keep ignoring rational and legitimate reasons for not using it, and banging on about how "ignorant" everyone who has reservations about systemd is. Is adult discussion really that difficult?

[DrXym]

"DrXym, what makes you think that doesn't also apply to those here who don't want systemd on their systems? You keep ignoring rational and legitimate reasons for not using it, and banging on about how "ignorant" everyone who has reservations about systemd is. Is adult discussion really that difficult?"

Show me one of these "rational and legitimate reasons" that I've responded to that you consider my response was unfair please.

[HieronymusBloggs]

"Show me one of these "rational and legitimate reasons" that I've responded to that you consider my response was unfair please."

This and other threads have plenty of examples. You've ignored them. Not only that, you repeatedly dismiss the concerns of experienced people who have to run reliable systems to get their jobs done. What makes the opinion of a small group of developers following the corporate agenda of their employer more valid than those of numerous experienced sysadmins and independent developers?

This isn't the black-and-white issue you seem to think it is. If someone wants to use systemd good luck to them. Just don't try to force me to use it.

[chain]

You seem to be one of the people who thinks that because systemd is modular it can't also be monolithic.

It being modular and it being monolithic aren't mutually exclusive.

Please feel free to further educate yourself here: http://judecnelson.blogspot.com/2014/09/systemd-biggest-fallacies.html

Let me quote an excerpt from the article that I linked above:

"A piece of software is modular if it is decomposable into distinct functional units such that each unit addresses a specific concern. Systemd, the Linux kernel, and X.org are all examples of modular software. Systemd addresses its concerns with its binaries, the Linux kernel with loadable kernel modules, and X.org with its drivers and extensions.

Now, a piece of software is monolithic if its components (if it has any at all) are tightly coupled--that is, components logically depend on one another to the point where using them in different contexts requires re-implementing the missing ones. Examples include Linux and X.org--in Linux's case, you can't use a kernel module without the kernel, the kernel can't run without the requisite kernel modules to interface with the hardware, and you can't use a Linux kernel module with other kernels or as a stand-alone program. Similarly, you can't use an X video driver without the X server, you can't use the X server without at least one video driver, and you can't use X's video drivers with other graphics managers or as stand-alone programs.

Under these definitions, systemd qualifies as both modular and monolithic. You cannot run journald without systemd, and cannot run systemd without journald (at least, not without losing logging for systemd-supervised programs). None of the *ctl programs work without systemd, nor do its collection of systemd-*d daemons. It used to be possible to run logind separately, but not anymore. According to the systemd developers, udev will likely be next to hard-depend on systemd. The point is, despite the fact that systemd is comprised of multiple binaries, the hierarchical logical coupling between them means that it is more accurate to think of them as extensions to systemd-PID-1 that just happen to run in separate address spaces. They are not truly independent, composable programs."

[arnolf]

Strawman arguments from 2014, great - lemme think, is that the one featuring the Great Fear of Binary Files?

Linux kernel is monolythic - and for solid reasons. Why systemd shouldn't be?

You're running a single "kernel" doing the bare metal things.

On top of that, you need a single "init" doing the bare init/start/stop/management.

Using systemd means having a well written piece of software to do all the boilerplate/checks for you instead of having to fight with locks, dependencies, privileges, capabilities, temporary files, environment variables and so on, while collecting all stderr and dmesg and whatever in a single timely ordered source. You just don't need to reinvent the wheel every time.

Once upon a time Mr. Famous Professor wrote that his Minix was the future because Minix was not monolythic like Linux kernel. And we know what happened right away.

Today a number of punypenor obsoleted "veterans" are desperately trying to save their bloated initscripts boycotting systemd and three years later they didn't yet manage to demonstrate a single compelling technical reason to make anyone follow their luddism.

[Graham Dawson]

@arnolf

If systemd only handled init, there wouldn't be an argument. The fact that it has swallowed logging, interprocess communication and device management, to name just three formerly independent subsystems, is precisely any there is so much hostility. Each new subsystem it absorbs is one less that can be replaced with alternatives and one more dependency on systemd that did not previously exist.

Devuan has non-defaulted the big desktops because they have dependencies on systemd thataren'teasilly resolved. Why in the hell should kde have deps on the init?

[HieronymusBloggs]

"they didn't yet manage to demonstrate a single compelling technical reason to make anyone follow their luddism."

They have not yet managed to demonstrate a single technical reason comprehensible by certain participants in this thread. That's no loss IMO. Why do you assume that anyone would want others to "follow" them? That seems quite a strange perspective to me.

A reminder of history: the Luddites broke things to make their point. That is occurring now, but not by the people you think.

[oiseau]

Hello:

I ran on MS-Windows OSs for many years, from DOS 6.0 onwards through 3.0, 3.11 and all the versions of W95/98/NT/2000/XP with the exception of Me.

I've seen and had to wrestle with quite a bit from and about MS systems over the past 20+ years.

Now I only run Linux on my rigs with XP on a VirtualBox (isolated from the web) for when I need it.

Not being a specialist or developer, a technical opinion I can back up and argue for is beyond me.

That said, this systemd issue in Linux brings back memories of my difficult transition from W3.11 to W95, with the end of the familiar *.ini files I understood (and could tweak when things went foul) to the dark and undocumented workings of *the registry*, which took me a few years to get a minimal hold of.

This quote from djames' post seems to be a perfect definition of what the registry in W95 seemed to me at the time:

"Systemd also has its dirty fingers into other parts of the system. As a replacement for sysvinit is is supposed to be an init system, but because its scope goes far beyond the initialization phase (and it doesn't let you take the good without the bad) it has become a dependency for many userspace programs that should never have any reason to interact with the init system at all, making it harder to use those programs on a non-systemd system."

So I have the gut feeling that systemd is a step in the wrong direction.

I am still a far away from being able to get around Linux (Mint 64bit) as I was with the MS OSs I used for many years, but I think it will be better for me to shift to a non-systemd distribution as soon as I am able to.

Just my $0.02, YMMV.

Have a good week-end.

[Criminny Rickets]

@oiseau

"I ran on MS-Windows OSs for many years, from DOS 6.0 onwards through 3.0, 3.11 and all the versions of W95/98/NT/2000/XP with the exception of Me."

I was reading your post and could not help think that your experience and mine are almost exactly the same, including the running of Linux Mint 64bit. The only notable difference from what you wrote and myself, is that I've also used ME, Vista and Win 7 (which I have in a VM for the very rare instance it may be needed).

[Anonymous Coward]

Hello Lennart

[HieronymusBloggs]

It's software, not a fashion statement

"modern software requires modern solutions which is something rabid luddites don't seem to understand."

See title. Something inexperienced people who can't see past the latest shiny distraction don't seem to understand.

[Anonymous Coward]

Me an' all the other lewd rabbits seem to understand something else.

[bombastic bob]

you forgot the troll icon

"modern software requires modern solutions which is something rabid luddites don't seem to understand."

(ok you posted as an AC which is why you couldn't show the troll icon)

luddites. use of that term is a dead giveaway. just like the [ab]use of 'modern'.

[Doctor Syntax]

"modern software requires modern solutions"

The difficulty I have about accepting this argument as favouring systemd is that I don't see systemd as modern.

I take this as a good principle of modern design: "A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away."

On this basis systemd fails to qualify as a piece of modern design. It may, by definition, be newer than that which it sets out to displace but new does not equal modern. In fact I see it as a throwback to the sort of messy complexities that the modernism of Unix set out to replace.

[Pompous Git]

"I take this as a good principle of modern design: "A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away.""

This is of course the canonical translation of Antoine de Saint-Exupéry's words: "Il semble que la perfection soit atteinte non quand il n'y a plus rien à ajouter, mais quand il n'y a plus rien à retrancher". (Terre des Hommes, 1939). Missing are the words "It seems that" from the beginning of the aphorism and that quite changes the meaning of what Saint-Exupéry wrote.

Perhaps we can set against it another aphorism: Appearances can be deceptive.

Some of the best examples of "good, modern design" occurred in the UK after I left in the mid-60s. Towering concrete edifices to house the workers that Bill Bryson described as the "Fuck You school of architecture".

[Doctor Syntax]

Some of the best examples of "good, modern design" occurred in the UK after I left in the mid-60s. Towering concrete edifices to house the workers

OK, here's another variation on the same theme: “Everything should be made as simple as possible, but no simpler.”

The 60s town blocks failed on the "no simpler" criterion, especially those that shared Ronan Point's construction.

[DrXym]

They've adopted it because systemd is demonstrably better in virtually every respect. Most of the complaints about it are seriously misinformed nonsense.

[jake]

DrXym, you can say that until you are blue in the face, and then you can threaten to hold your breath until you are sick. It still won't make it true. But follow your bliss.

[DrXym]

I can keep on saying it because it is true. People moaning about systemd are acting like conspiracy theorists parroting the same debunked nonsense over and over again

[jake]

You can keep saying it simply because nobody is going to stop you. But that doesn't make it true. I'm no parrot, I'm a jaded old sysadmin. And with that comes a healthy dose of skepticism. Over the decades, I've seen fads come and I've seen fads go. In my eyes, after what I hope was a fair evaluation, systemd is a fad that will run its course. Feel free to use it if you see fit, no skin off my teeth. Just don't try to baffle me with bullshit. It won't work. There are reasons why systemd will never be ported to BSD, and they are the same reasons why Devuan exists, and why Slackware & Gentoo don't use it.

[DrXym]

"There are reasons why systemd will never be ported to BSD, and they are the same reasons why Devuan exists, and why Slackware & Gentoo don't use it."

Those "many reasons" applies to lots of low level Linux software. Porting a pid 1 process between two disparate kernels is hard, especially when compounded by philosophical and licence issues. Hardly an argument really. I'd also point out that OS X uses launchd for some of the same reasons that systemd exists - because sysvinit sucks for modern desktops.

Gentoo also allows systemd to be used. So your argument against systemd boils down to slackware. Seriously.

[HieronymusBloggs]

"So your argument against systemd boils down to slackware."

Classic straw man. If you bothered to read jake's posts you would see he has several valid arguments against systemd. Slackware is merely his chosen method of systemd avoidance (and a sensible one given long term experience with the distro).

Don't mind me though, your debating style is doing the anti-systemd crowd a lot of favours.

[HieronymusBloggs]

"People moaning about systemd are acting like conspiracy theorists parroting the same debunked nonsense over and over again"

Two questions:

1 - Do you own a mirror?

2 - Are you a Red Hat employee?

[Anonymous Coward]

Then, it isn't trying to be damnear everywhere like glibc and X? Because that's what I'd moan about... along the same lines, my fundamental problem with politicians is that they all decided to be politicians at some point and I don't quite trust the sort of person who wants that. Also I have a severe allergy to marketing campaigns and it seems systemd is riding one like UNIX never had to.

[Anonymous Coward]

Most of the complaints about it...

... are because those sysadmin need to learn something new and they really hate it, and fear it.

They like a fossilized Unix that must not change since it was written on stones in 1970. They're happy with it because they are frightened anything new will make them obsolete like Unix is, unless they update their skills.

It's far too comfortable to keep on using the same commands and scripts over and over.... and keep everything stuck in 1970. My Linux sysadmin is a perfect example of them....

[Anonymous Coward]

and ahead on your left you can see

I quite happily learned to fight with CMake instead of autotools. The syntax is kind of completely different (command lines are still command lines) and the workflow is all weird by comparison-- but I crammed it in here and I'm basically happy with it because it makes sense to me and it works, actually pretty dang often, and with rather few surprises. It's real easy to accuse people of being afraid of change & the future, innit? I am merely turned off by a number of errant fragrances near a certain cabal. You want to talk about motivation? It's just as easy for me to accuse them of inordinately enjoying the power trip from having millions directly or billions indirectly relying on their code, (gasp) almost like Linus!

[Doctor Syntax]

Re: Most of the complaints about it...

"... are because those sysadmin need to learn something new and they really hate it, and fear it."

Sysadmins are paid to keep things working. They know through experience that changes break things. Therefore they realise that the only changes that should be made to something that works are those that are absolutely essential. This means that changes should be evolutionary not revolutionary. I've yet to hear about one essential thing that systemd's supposed to have fixed.

I think the resistance to learning something new is on the part of those coming from the Windows side being reluctant to learn how Unix works.

"They like a fossilized Unix that must not change since it was written on stones in 1970."

I can't claim to have used Unix back in 1970. However I have used it since the early '80s so when I tell you, as I do, that Unix has evolved greatly over that time and has not become fossilized I do so from long experience.

"My Linux sysadmin is a perfect example of them...."

Your Linux sysadmin appears to know what he's doing. You should learn from him.

[HieronymusBloggs]

Re: Most of the complaints about it...

"... are because those sysadmin need to learn something new"

My complaint isn't about learning something new, it's about being expected to learn something I consider technically inferior. I don't hate or fear systemd. I have just chosen, after research and testing, not to use it. If it becomes unavoidable on Linux I'll switch to *BSD. I can't understand why that generates such disapproval. My systems are mine. Use what you like on yours.

[Androgynous Cupboard]

Re: Most of the complaints about it...

I was installing a new server recently with systemd. I was mucking around with the mounts in /etc/fstab, but for some reason the mount command wasn't picking them up. Turned out I had to request systemd reload the file by running "systemctl daemon-reload".

This isn't a question of a "fossilized admin" such as myself having to learn new things - it's simply poor design. Caching the contents of a small text file that is likely to change is the kind of idiocy I would associate with sendmail - it's pointless from an optimisation point of view, and only add to the complexity of the process. A simple text file read on demand has now become an on-disk representation of some internal datamodel which I have to manage manually. There is no problem solved by doing this, only a problem created.

That's one example, I'm sure there are others. Maybe systemd does solve some great horny issue but it has, and indeed necessarily must, introduce a bunch of others that have been ill thought out. Change for change sake, as someone has already pointed out.

[Steve Graham]

why have all the major distos except Slackware and Gentoo adopted it?

Dependency.

The way systemd is designed means that a developer who wants his software to work well on a systemd box has to work much harder to make it also work well (or at all) without systemd.

The obvious result is that most developers don't bother, so there's a rolling increase in the amount of software that needs systemd.

[aerzurumluyan]

Politics indeed

When people raised systemd issues on mailing lists, the answer would be (when there's no flame) is "but everyone/fedora does it too"

[Mozzie]

Because the distros that the major distros are built on adopted it.... mainly.

Debian switched to systemd which in turn pushed distros like Ubuntu on to systemd, which then resulted in Linux Mint (based on Ubuntu) shifted on to it as well. Red Hat (and thus CentoOS and Fedora) are responsible for systemd in the first place, with Lennart Poettering, Kay Seivers and Harald Hoyer all being Red Hat engineers and the authors of systemd.

Have a quick check on Distrowatch to see how many of the top 20 distributions are based on these two base distros, it pretty much covers the large majority of installs out in the World.

There is also the issue of desktop environment dependencies. As the article mentioned Gnome, KDE and Cinnamon somewhere along the line depend on a library that is systemd dependent, with Gnome going full steam ahead with systemd reliance.

Arch Linux is systemd based but also offers a way to remove it and is also available systemd-free via Manjaro-OpenRC, otherwise I personally am just ever hopeful a number of traditionally Debian-based distros will move over to Devuan.

[DrXym]

"Debian switched to systemd which in turn pushed distros like Ubuntu on to systemd"

Kind of absurd rationalization given that Ubuntu is downstream of Debian, not upstream. Just like Devuan in fact. Also, Ubuntu didn't switch from sysvinit to systemd. It switched from upstart.

[Graham Dawson]

And I'm sure the fact that systemd rolled a couple of very important daemons (such as udev) into its monolith, forcing dependency on it where none previously existed, has absolutely nothing to do with its forced adoption.

[g00se]

Kind of absurd rationalization given that Ubuntu is downstream of Debian, not upstream.

Err .. isn't the fact that it's downstream the point?

Just like Devuan in fact.

That's disingenuous. It's not 'downstream' in the normal sense - in essence it's a fork

[SImon Hobson]

As others have stated - politics and entanglement.

In the case of Debian, there was a long and "heated" discussion. In the end there was a vote which looks to have been designed to produce the desired answer rather than get an accurate count of opinions. If rationally analysed, it comes out with a clear majority (by a long way) of people not wanting systemd - but because of the way the options were chosen*, and some of those were interpreted*, the outcome was decided as being a majority in favour of systemd.

* Eg, "we don't want to debate it any more" was taken as "we want systemd".

But AIUI, part of this came about due to the ever greater effort needed to disinfect upstream packages - particularly desktop environments. As others have pointed out, systemd goes way way way beyond what it was originally sold as - and it appears to be a deliberate ploy to infect as much software as possible and thus make it ever harder to keep that software disinfected. Part of this is deliberately deprecating old but reliable and well known interfaces and replacing them with "new shiny" ones. Thus you get the problem that unless software has been written to use the new shiny (but really, in most cases no better) interfaces then it doesn't work well/at all on a systemd system - but if it is written to use those systemd interfaces, then it doesn't work well/at all on non-systemd systems.

KEY POINT systemd is incompatible by design. it is clearly a key design tenet that it must break as much compatibility as possible with "legacy" systems. That is a key driver to making it herder and harder to avoid it.

This last anti-choice point should alone be good enough reason to avoid it. We saw a decade or two ago what happens when one dominant group (in that case, Microsoft) is able to eliminate choice. Things stagnate, choices are made for the wrong reasons - ie if your only choice is Windows Server then you have to run Windows Server regardless of whether it's the right/best choice for your application. In the browser sphere, we saw how once they'd seen off the competition and eliminated choice, IE got stuck in a mire of mediocrity and dragged the whole web down with it for several years.

If systemd were so great, then people would adopt it by choice. With a few exceptions, most of the adopters have done so because they've given up fighting it (or at least, just run out of resources to fight it).

And I'll finish with :

PID 1 is (after the kernel) THE most critical component of your GNU/Linux system. It MUST be secure and stable. Basic software engineering principles dictate that it should be as small and simple as possible, with as few interfaces as possible, and basically minimise all avenues for bugs and vulnerabilities as is reasonably possible. systemd is the opposite of that - it may be multiple different binaries, but they are close coupled, with ill defined interfaces between them, and above all written by a team with (to be very polite) a less than impressive track record in producing "good" code.

Do you really want your system to be 100% reliant on software produced by people who deliberately break things and then leave others to clean up after them ? Do you really want your system to be reliant on people who re-create historical dangerous "features" and them blame the user when it hoses their system ? Do you really want your system to be reliant on people who's code is so "good" that they've been banned from contributing to the kernel ?

[Tim99]

"The question I keep asking myself is if systemd is so bad, why have all the major distos except Slackware and Gentoo adopted it?"

Dreams of riches? It seems to be a way of locking users into a "Big Company Distribution" - Even If the distribution is open source, but becomes becomes convoluted and opaque, only the anointed "Big Company" people will be able to support it; and the user will be locked into that particular distribution. Are we revisiting the Unix Wars?

[bombastic bob]

"why have all the major distos except Slackware and Gentoo adopted it?"

a) "new, shiny"

b) it makes them *FEEL* like they're doing something to implement 'the new thing'

c) peer pressure (lots of 'how come we don't have systemd yet' mail, etc.)

d) same reasons why they implement gnome 3 instead of MATE as a default - right Debian?

I noted that FreeBSD kept gnome 2 for the LONGEST time in ports, and when Mate was ready for prime time, THEN they brought in both gnome 3 AND mate. I'm still using gnome 2 from a couple of years ago and it works pretty well. In fact, NOT having "bleeding edge" all the time made the userland a LOT more stable.

Thing is, when the userland practically *DEMANDS* support from systemd [a continuing problem on BSD systems, which will *NEVER* have it] then you have to put in hacks and patches to compensate. Or put in systemd. Guess which one they did?

We all know why Gnome 3 developers did what it did: they're IDIOTS trying to be Micro-shaft and swallow the touchy-fondly-fat-finger-friendly "trend" coolaid. For desktops, that's been nothing but *IRRITATION* for several *YEARS*. THey had to DUMB DOWN several gnome 2 features that were *THANKFULLY* retained by Mate, like "how many icons can I cram into the icon bar". Try 50. Or maybe just 20, ut you have the system monitor and you organize the icons into groups with variable white space between them. In Gnome 3 you have "3 groups" and you have to engage in odd key+mouse combinations to get to the menu to re-position them (yeah changing THAT was a real "pleasant" surprise, thanks Gnome 3 dweebs). In any case, Gnome 3 carried on in the wrong direction, just like Micro-shaft with the 2D FLATSO FLUGLY "the metro" and now UWP and Win-10-nic.

So now, these same MILLENIAL "it's OUR turn now to do it OUR way" types have created a *MONSTER* that feeds itself and grows stronger due to "all of that".

Fortunately we have Mate, and now Devuan. Thanks, Devuan! Looking forward to the preservation of X11 and *NOT* Wayland, as the SAME PEOPLE drive everything off the cliff like a bunch of lemmings.

[herman]

...and through all your ranting and raving over the years, there was Slackware just quietly and efficiently getting on with it.

[oiseau]

"The question I keep asking myself is if systemd is so bad, why have all the major distos except Slackware and Gentoo adopted it?"

Well ...

Possibly lack of foresight and/or sheer lazyness.

One enlightening example from as far back as 2015:

---

"The reason why we used systemd is because we didn't decide to use it; we just went with the default."

---

See https://forums.bunsenlabs.org/viewtopic.php?id=131

I suspect that this "going with the tide" attitude has been the norm among most Debian or RHEL based distributions.

[ttkciar]

Systemd adoption was scant before they absorbed the udev project, making udev dependent on systemd. Udev is how most Linux distributions load device drivers, and it is an essential component of a working operating system. Newer versions of udev require systemd to be installed, so many distributions installed systemd.

Soon thereafter, developers forked eudev from udev and have been providing it as a systemd-free alternative to udev. Slackware and many other distributions switched to eudev to avoid the systemd dependency.

This is a list of distributions which have switched from udev to eudev. It is rather extensive:

https://forums.gentoo.org/viewtopic-p-7648392.html

[Mage]

Good news

Choice is good even if you want systemd

It's good if you don't want systemd

[wolfetone]

"GNOME, KDE, and Cinnamon have been removed from tasksel, but can still be installed although they “are known to suffer from some glitches due to the lack of systemd.”"

So what window manager is recommended for use with Devuan? XFCE I suppose?

[jake]

@wolfetone

Slackware defaults to KDE ... but not a cutting-edge KDE. It gets the job done.

[fukudasan]

Re: @wolfetone

Dang, that sounds like a good choice t'me... :)

[Bronek Kozicki]

XFCE is my impression as well, but there is nothing wrong with it!

[Steve Graham]

So what window manager is recommended for use with Devuan?

I don't know what's recommended, but I use Openbox.

I don't have a "desktop environment" because I've never been able to work out what useful extra functionality they provide.

[chuckufarley]

Re: So what window manager is recommended for use with Devuan?

I like LXDE myself. It's very small and has few dependencies. Unless the distro in question has bloated itself I can usually boot to a desktop with less than a 1GB of RAM used. Sometimes (ie. Gentoo) it's less than 256 MB! This makes LXDE a great choice for virtualized desktop systems or if you need to keep the total number of installed packages low for maintenance and/or security reasons.

[chuckufarley]

Re: So what window manager is recommended for use with Devuan?

With LXDE, no ssh server, and no print server installed Devuan boots to 345MB of RAM used. Not bad at all. After opening Firefox and loading up El Reg to write this it's at 753MB with 222MB cached. It's a real shame when your browser out weighs your OS.

[Ramazan]

Re: what window manager is recommended

blackbox

[Anonymous Coward]

Strategic Point of Control

Linux is a wild animal and SystemD is the collar that will be used to tame it.

[Ramazan]

Re: Linux is a wild animal and SystemD is the collar

not even funny

[Anonymous Coward]

Re: Strategic Point of Control: Linux is a wild animal and SystemD is the collar

"not even funny"

Sorry, it wasn't meant to be... I was going for something along the lines of:

https://devcentral.f5.com/articles/what-is-a-strategic-point-of-control-anyway

No more 4am replies. Now to get moving on my switch to Devuan.

[Alistair]

systemD wars are back on again.

a) I've had issues. many issues. systemd is *on occasion* one of them

(FD - enterprise *systems* admin, 20+ years experience, covering SGI, HP, IBM, Sun, Caldera products. Once in while I still poke at mainframe (MVS/JCL/IMS stuff). Sometimes *cough* windows admin. I've done Networking, DB, code, hardware work along the way. I can still wire up a rack and make it look like a work of art if I want.)

RHEL7 is systemd. Our automation tools choked on it because we'd let them get older and out of date. Updating the automation tools on something in the neighbourhood of 4,200 systems is a serious ass pain. But it got done - in reality there were about 120 to 130 cases where the automation variables, the config for the automation or the actual automation code had to be altered to the case where it worked for both. Our automated build needed to have some language altered to make the systemd instances behave better in *error* cases - the binary logs are seriously #$%@#$% annoying in the case of (did the device actually come online and do its shit) - but again - once the language was fixed and the tests run we got there.

I seriously prefer SysV. Simple, straightforward, *do it in this order* if it fails log the failure and keep going. SystemD *still* has some nasty nasty misfires, including (most recenty) loosing the ability to differentiate kernel cli rootvg= swap= and root= (which, let me tell you was a fuckup of a different order) in cases where *the entire volume group* was encrypted. - one of the things I've seen it do was roll back to a kernel cli option set that grub didn't even have ..

What systemd does do is parallel inits, and that *does* help in some cases - mostly not in the *desktop* world but in the *server* world this has cut some of our app server boot times WAY down. It took *me* about 5 months to get the hang of *just* the basics of systemD. I'm nowhere near a specialist in the territory, but I'm capable of figuring out if it's systemd's fault and *usually* can quickly find the culprit. Hell, I've even written systemd inits for a few weird apps.

What I despise about systemd is that my desktop KDE can, *FROM MY USER INSTANCE* alter system level settings. (I said can, not has or does). This ... right there ... is the single largest violation of the unix principles. -> on that front, they are *working* on adding different security methods to limit that capability <-

What I will note is that some of us systems folks are going to resist, some of us will adapt, and some will just lie back and complain. About *everything*. I believe that as a Hooman Bean, there will never be a day when I can sit back and say "I know it all, you can't teach me something new". So. I don't like it much, but it works (mostly) and when it breaks I can figure it out and fix it. Much like my car. Or the dishwasher. Or my NBSD box, or that HPUX thing I've been working on decomming for the last 8 months. Or that Cygwin instance that #$%@# fails to start because some jackass in the windows domain *once again* changed *that* rule in AD. etc etc etc.

Why? - its what I do. I fix #$%. Usually when its broken.

/Penguin. 'Cause its Friday. And because I saw the quarter stats, and according to those, we now have more active linux instances in the DC than any *other* single OS. (and with O365 much of MS is now *sigh* outsourced to MS. I can't wait.....)

[Outer mongolian custard monster from outer space (honest)]

Well done chaps and chapesses on the gaining momentum front. I've been tinkering with devuan the past week out of interest because I want to run debian on a long term server, couple of tiny niggles but its running a lamp stack happily and we have another box trailling dovecot and some other mail stuff. The only minor pain so far has been certbot not recognizing it as a supported program for a automated letsencrypt cert which I could work round anyway.

Ain't choice brilliant?

[Anonymous Coward]

PulseAudio

I'm surprised that nobody seems to have mentioned PulseAudio yet. I can't speak for anyone else, but that was a complete and utter shitshow for me at least for a LONG time. A buggy, crappy piece of software sat in front of ALSA and not showing any real improvement - I still have nightmares about it

Other than the good arguments about systemd going against the *NIX philosophy of 'do one thing and do it well' etc., an important thing to consider is that systemd has PID1 and it hogs it to an insane degree. There was a bug fairly recently where if the systemd journaling subsystem failed on boot, the entire system could hang. That kind of amateurish coding, ESPECIALLY In an init manager, is unacceptable. I'm no kernel-level guru but even I know shit when I see it.

PulseAudio and systemd have (at least) one very detrimental thing in common. I'll let you figure out what it is...

[Jim-234]

Couldn't they just use an older version of the Cinnamon Desktop interface?

The Cinnamon Desktop interface worked fine before SystemD in the older versions of Linux Mint.

I would think that simply using an older version of it would work fine in the packages?

[herman]

KDE is fine without systemd

KDE works fine on my OpenBSD and Slackware laptops. So I don't see what the issues are.

[Graham Dawson]

Re: KDE is fine without systemd

The KDE packages they use are derived from Debian, where a dependency exists between KDE and systemd through, amongst other things, systemd-logind, which manages users and userspace access to hardware. They're also implementing more dependencies on systemd's various integrated subsystems.

I imagine they'll eventually repackage KDE to remove the dependency, but that's going to be increasingly difficult as time passes. KDE is already moving to make use of more of systemd's "features".

[chuckufarley]

Thank you RNGesus...

...for my thrice weekly dose of xkcd.com!

https://www.xkcd.com/1833/

[hayzoos]

My thoughts on the systemd thing

A major symptom of the problem of systemd is the fact that it takes so much effort not to use it. Here you have *nix veterans having to establish a major fork/project/distro which shows the effort goes beyond even what many here can do. I'm not talking the one-off, "I removed systemd from xyz distro and everything is working for me except maybe x and sometimes y, but that is typical *nix anyway so it works".

I finally left Windows last year as my daily driver. I dabbled in *nix of various flavors for many years so I have no fear of leaving Windows. I was trapped by "certain application(s) are only available in Windows". I solved that issue by a POV change. I don't use an OS because of applications anymore. I have things to do, not I use a particular app. When a part of my system exhibits objectionable "features", I can change the entire toolset if I have to ,to get my stuff done.

Now, (over a many months' time) I find Linux distros beginning to exhibit objectionable "features" mostly due to systemd. Including the distro I chose to replace Windows. It seems I have few distros to choose without systemd. Fortunately for me, I have dabbled with some of those before, so I am more comfortable considering them. Unfortunately, the Graphical environments are also exhibiting systemd issues. And udev is in the future?

I propose for clarification that non-systemd distros continue to be known as Linux distros. Those that have adopted/(been assimilated by) systemd be known as systemd distros instead of Linux distros. It would make things alot easier in choosing one. There already are compatibility issues between the two types, so it seems a new species has been created. Should also be extended to anything else with a systemd dependency, so you know the package is incompatible with non-systemd Linux.

Is there a niche for creating a systemd to legacy translation package to make things like (systemd udev), (systemd KDE), et. al. work in legacy Linux? Probably more like a compatibility layer, but also a sanity layer to insulate kernel and other low levels from userland. Nevermind, it's starting to sound more and more like how Windows (d)evolved.