Systemd daddy quits Microsoft to prove Linux can be trusted

Linux celeb Lennart Poettering has left Microsoft and co-founded a new company, Amutable, with Chris Kühl and Christian Brauner. Poettering is best known for systemd. After a lengthy stint at Red Hat, he joined Microsoft in 2022. Kühl was a Microsoft employee until last year, and Brauner, who also joined Microsoft in 2022, …

[TVU]

"Linux celeb Lennart Poettering has left Microsoft and co-founded a new company, Amutable, with Chris Kühl and Christian Brauner"

^ Personally, I would be minded to replace 'celeb' with 'controversialist'.

[David 132]

Beat me to it. Ditto the phrase "leading light" used later - if he is, it's only in the sense that smugglers in Ye Olden Days would set up false navigation lights to lead merchant ships onto the rocks!

[Anonymous Coward]

leading lights

Possibly, if the light is red and it says STOP. Not STOPD.

It does not need 6 root bugs and whatever new syntax floated into his brain along with a cloud of his own farts.

It does not need to be merged with 6 unrelated road signs.

It does not need a new command line configuration tool instead of a config file.

It does not need 17 dependencies to bloated packages that also have dependencies on each other.

Linux might have benefited from a secure boot. Now we have a hot mess, because we let Lennart build it. And he did it the way he always does, by going off on his own, ignoring most of feedback and other peoples requirements, and building whatever popped into his head. No he, one of the people I would trust the least to keep my encryption keys, is supposed to architect the basis for cryptographic trust in Linux.

Only it won't be in the Kernel because Linus is fed up with the B.S. (both his, his knackered code, and the permanent hate cloud that follows in his wake bitching when his trail of rabbit raisins crosses the path of a grouchy sysadmin and breaks their deployment in the dead of night).

[benaki]

Re: leading lights

Tony Blair - Middle East Peace Envoy.

[rnturn]

Fix systemd first.

In recent weeks/months I've had to manually fix service unit files that incorrectly try to pass shell environment variable values to service startup command lines. It's now the first thing I look for when services won't start.

LP: Clean up your mess before infesting Linux with yet another one.

[Anonymous Coward]

Perhaps it was a typo - "leading blight"

[stiine]

I would like to know what type of recreational pharmaceuticals the author took before writing this piece.`

[David 132]

...and whether he'll share?

Seriously though, I avoid impugning the Reg writers; I think that in this case, the author was just trying to be even-handed and avoid sparking the usual Systemd/Poettering flame-war. And who can blame him?

[seldom]

And did it work'

[Anonymous Coward]

From available evidence below...

Nope...

Though Poettering leaving Microsoft might (just might) improve people's feelings about him, it's a bit 'too little, too late' for many.

[coredump]

Re: From available evidence below...

I don't really have "feelings" about Poettering the person, so much as opinions about his product. I'd allow that I haven't cared for the way he and his followers interact sometimes with the people who have to deal with their stuff.

Neither his joining Microsoft, no leaving, really factors in to that for me. If he was able to take some nice paychecks out of a big corporation and escape on his own terms, good on him.

One thing that does offend me is the systemd fans claiming those of us who have no use for it are simply old, unwilling to learn, stuck in our ways, or even just plain stupid. As opposed to having enough experience to recognize that some reinvented wheels don't always make things better.

[vtcodger]

Re: From available evidence below...

One thing that does offend me is the systemd fans claiming those of us who have no use for it are simply old, unwilling to learn, stuck in our ways, or even just plain stupid. As opposed to having enough experience to recognize that some reinvented wheels don't always make things better

There isn't a whole lot to be said for getting old. But it does allow those of us who aren't prodigies enough time and experience to develop a feeling for what Colin Powell once described as "Freaking Crazies". Freaking crazy certainly seems an apt term for the Trump administration, cryptocurrency, quantum computing (which will at the current pace of improvement be able to factor an arbitrary 1024 bit number in about 10,000 years)*, many (not all) aspects of AI, folks who are out to save the world from "climate change" using a tool set that can't possibly do what they want it to do**, and systemd.

Square wheels might seem better than round ones in theory. At least to some. But Mythbusters actually tried them once. They cause (as one might expect) a REALLY bumpy ride. Maybe that could be overcome with a sufficiently complex suspension system. But why bother?

*There looks to be some realistic hope that quantum factoring will advance faster than one bit per decade. But for the time being, actual advances are damn slow. They had to fudge the setup in order to factor a four bit number.

**There are certainly problems associated with climate change. But nowhere near the problems caused by failure to understand past climates, how climate(s) work, how power grids work, and why wind and solar can not power modern industrial societies without economically viable long term energy storage we have not the slightest idea how to build.

[Lomax]

Re: From available evidence below...

> wind and solar can not power modern industrial societies without economically viable long term energy storage we have not the slightest idea how to build

Someone should tell the EU.

The EU’s electricity transition reached a new milestone in 2025 with wind and solar generating more power than fossil fuels.

[Lomax]

Re: From available evidence below...

And the reason this is happening has a lot less to do with AGW than other more immediate concerns:

• The belated realisation that we need to kick our dependency on Russian oil & gas
• The simple fact that it is cheaper (not that us consumers will ever notice)

[Anonymous Coward]

Re: From available evidence below...

Who is "we"? The UK has no significant dependency on Russian supplies.

[Lomax]

Re: From available evidence below...

All of us.

[thosrtanner]

Re: From available evidence below...

We'll notice. We just won't benefit.

[Anonymous Coward]

Re: From available evidence below...

You're making some assumptions about how power grids work yourself. Wind and solar have made (and were making) huge improvements to the power grid because running them requires no fuel delivery.

Peaking and storage are important, but they're not unsolvable problems for s and w.

Up until Trump's war on wind, solar, and energy storage, electric generation companies were choosing to use them. Energy costs weren't rising too much. Now that oil/gas/coal are the only approved options for new plants (according to Trump et al), electricity costs are going up. It's not just the new AI datacenters that are driving energy cost. It's "you must build or run non-renewable energy now" as a government policy.

Installing new renewable energy was becoming cheaper than running old thermal plants, or building more of them.

[JLV]

Re: From available evidence below...

Post ended up whining more about climate change than systemd, which it barely mentions.

Yes solar and wind intermittency is a major issue, especially when viewed seasonally. I once saw - can't find it anymore - a paper by California's DoE talking of a 5x difference on sun and wind between high and low periods in the year.

Yes, a lot of current planning is driven more by wishful, romantic, Green views than solid engineering. But climate change as a whole is a big enough existential problem NOT to be put on this list, esp with the snide "understand past climates".

[Lomax]

Re: From available evidence below...

Agreed, but the reality on the ground is that wind and solar are growing rapidly as a share of electricity production, and that's not "wishful thinking" - it just is. I know this simple fact is infuriating to some people, though I struggle to understand why. The energy companies are clearly not charities; they simply do what's best for their bottom line and their shareholders, and renewables are more profitable, cheaper to install and maintain, and easier to deploy locally. This last point is significant, as it offers further savings. There is no question where the trend is going: renewables are for sure going to replace fossil fuels, the only question is how quickly. Energy storage technology (or the lack of it) is probably the most important factor here.

https://www.bbc.co.uk/news/articles/cz947djd3d3o

Record year for wind and solar electricity in Great Britain in 2025

In 2025, wind, solar, hydro and biomass generated more than 127 terawatt hours (TWh) of electricity in Great Britain, according to BBC analysis of provisional Neso data. That beats the previous high of 119TWh in 2024.

This is even more remarkable when you consider that only fifteen years ago the number was in single digits.

[JLV]

Re: From available evidence below...

Oh, I am NOT dissing solar or wind here. But the Greens where I live have long opposed a new dam. And generally display all the economic savvy of a lobotomized goldfish. The Canadian Greens have a policy page that is long on wokeness and direly short on practical measures to reduce emissions.

The ones in Germany were so against nuclear that they retired working plants. Baseload backup was, for a while anyway, provided by Polish brown coal. For all the massive costs of the Wende Germany’s emissions declined at roughly same rate as other EU countries.

In Sri Lanka they pretty much nuked agriculture a few years back, by banning non-organic fertilizers.

So, no, not the least bit a fan. Global warming is too much of a problem to trust them dealing with it. Thats what my “romantic” alludes to.

[Lomax]

Re: From available evidence below...

Personally I'm a lot more concerned about what your neighbour to the south is up to.

[bigtreeman]

Re: From available evidence below...

rust batteries

iron - air batteries

https://formenergy.com/

[Ropewash]

Re: From available evidence below...

Especially when all those wheels are every geometric shape other than round and there's now 7 of them on you car instead of 4.

Pulseaudio was the canary in the coalmine.

[RegGuy1]

Re: From available evidence below...

when all those wheels are every geometric shape other than round

I always remember at school discovering from one of my maths teachers that the highest point on a (UK) 50p coin -- they were new in the 1970s -- when rolled on a flat surface stayed at the same height above the base. In other words it rolled just like a circle! And I've always remembered the name it was given: an epicyclic heptagon.

Wheels don't always have to be round!

[Peter Gathercole]

Re: From available evidence below...

The epicyclic 50p coin is a constant diameter shape, not a constant radius one. The only shape that has a constant radius is a circle, which is why it is he only real shape for a simple wheel attached to an axel.

An epicyclic heptagon could be used to make rollers not connected to axles (think the way that they think large stones were moved for the pyramids or Stonehenge), but not wheels.

[RegGuy1]

Re: From available evidence below...

The epicyclic 50p coin is a constant diameter shape

Yes, that's how it was constructed. Put the point of a compass on one of the seven points and draw an arc across the two opposite points -- do that seven times. And it became instantly clear why it has this property, because it is always rolling on a circle, it's just seven different circles of the same radius. So the diameter of the scape is the radius of these circles. Fascinating!

[Peter Gathercole]

Re: From available evidence below...

It needed to be this shape for coin verifiers and sorters in vending and other machines. These normally work by measuring various properties like weight, diameter, thickness and sometimes by magnetic signature. But most start with diameter and then check weight, so a constant diameter shape is essential for coins.

[gosand]

Re: From available evidence below...

"some reinvented wheels don't always make things better."

Nor does it have to, for every use case. Some people genuinely love systemd and all of its tentacles.

I didn't have a good experience with it, so I went to a distro that didn't require it.

What I don't like is that I had to do that because systemd was the ONLY OPTION for an init system. You could blame the distro for not putting in the work to support multiple inits, but systemd was built and implemented in order to supplant other init systems and shut them out with the intention of becoming the de facto standard. If it was built with choice in mind, distros wouldn't have had to choose one over the others.

[TVU]

Re: From available evidence below...

To this day, I still do not understand why the 2014 Debian conference committee members voted to adopt systemd as init system for Debian.

[coredump]

Re: From available evidence below...

It might be interesting to hear what they think of their decision a decade+ later, after all that's happened.

[Ian 55]

Re: From available evidence below...

To be fair, the vote was to have systemd as the primary init system - the majority expected the other ones to keep working - but some maintainers went down the **only** init system path, including removing code that enabled their stuff to work with other ones.

[NoneSuch]

I guess they'd all prefer Steve Ballmer.

Do you want to see Linux grow or end up paying for a monthly subscription to Microsoft for life?

Anyone with a perfect history of inter-personal relationships and flawless social media presence please reply first.

Outta here for the weekend.

[HereIAmJH]

"Systemd daddy quits Microsoft to prove Linux can be trusted"

The article starts with a bad assumption, the title. My first thought was that leaving Microsoft isn't going to prove that he can be trusted. Locking down Linux in a Microsoft Trusted Computing methodology isn't going to make it more trustworthy. It will just make systems harder to maintain.

[Darkhand]

"It will just make systems harder to maintain."

Which is pretty much the goal of Systemd anyway.

[RAMChYLD]

This reeks of Stephen Elop "leaving Micro$oft for Nokia". We all know how that went.

[R Soul]

I suppose the systemd dumpster fire is the equivalent of Elop's burning platform.

[JLV]

> Microsoft Trusted Computing methodology

Hey, maybe the PC manufacturers want to redo the Windows 11 trick on Linux and obsoletize all current PCs. What better trojan camel than LP?

[Tim99]

You're too kind. No, really, you are much too kind!

[LionelB]

Aren't those terms pretty much interchangeable these days?

[botfap]

Im surprised he wasn't forcibly parachuted into the Devuan camp

[TVU]

Now that I would love to see!

[seldom]

Without the nasty, heavy, old-fashioned parachute.

[David 132]

Parachuted - requires networkd, dhcpd, logind, atmosphered, and Flatpakd. Inexplicably fails to work as expected, #WONTFIX.

[Anonymous Coward]

Gory, gory, what a hell of a way to die !!!

Correction:

'Flatpakd' is invoked automatically, if 'any' of the other daemons are NOT running, when 'parachutee' & 'ground' are congruent !!!

Cue rousing chorus of "Blood on the Risers" to the tune of "The Battle Hymn of the Republic" ...

See https://en.wikipedia.org/wiki/Blood_on_the_Risers

:)

[Anonymous Coward]

We could only find testers willing to...

Due to problems finding willing testers, current testing remains with planed on the ground at zero velocity.

[Lomax]

Well played!

[graemep]

[quote]At the risk of triggering a slew of angry comments, it can best be described as software that runs first, then starts other required services and applications.[/quote]

Common misconception. That is one component of systemd, the init system. There are lots of other components. To quote the systemd website "systemd is a suite of basic building blocks for a Linux system. " and these include "Other parts include a logging daemon, utilities to control basic system configuration like the hostname, date, locale, maintain a list of logged-in users and running containers and virtual machines, system accounts, runtime directories and settings, and daemons to manage simple network configuration, network time synchronization, log forwarding, and name resolution. "

The entire objection to systemd is that it is not just an init system. https://systemd.io/

[Anonymous Coward]

> The entire objection to systemd is that it is not just an init system.

Exactly. "Do one job and do it well", not "Do everything, mostly half arsed, and log any issues in a binary format that can't be quickly parsed by simple tools".

[stiine]

...that can't be parsed with a Mark-I eyeball.

[Anonymous Coward]

In other words, make it Microsoft compatible.

Makes you wonder if he left with their blessing as he appears to be planning to make things worse, i.e. even more Windows compatible.

[Anonymous Coward]

It's virtually an OS within the OS.

[ChoHag]

Those who do not understand unix...

[Doctor Syntax]

He certainly didn't reinvent it.

[Anonymous Coward]

I'm sure some Linux people will be upset to find it's virtually an OS inside systemd.

[martinusher]

Not just an OS, unfortunately. If you've followed closely the development of GUI based systems primarily as a developer rather than a user you'll notice that they are (or rather were - I live in hope that they grew up) they're a rather clever way of simulating a lightweight operating system. The snag is that if you swallow the KoolAid and actually think you've got is actually an OS you'll end up with a bloated, increasingly inefficient mess that encourages applications to poll for events. Windows, in other words (for much of its life -- although, again, I hope its grown up a bit over the years).

So this is my complaint with systemd. On paper its a great idea but in practice it appears to be running an inelegant OS on top of an elegant one. Windows, again (at least Win2000). This might explain the prime movers' dislike of POSIX; (I'll admit that its not my favorite interface/standard but compared to the abominations that replaced it in Windows....there are times when its great to be retired.......)

[Doctor Syntax]

It's not even a great idea on paper. It's too tightly integrated. The great principle of Unix was that it was modular. Any component could be replaced by another in a way that could be done without causing shock waves. In regard to init functions I've known that to happen from 7th Ed through System III to SysV 4. They didn't demand that other bits of the userland be modified to fit. They all had plain text files hold their configuration. That was a good idea on paper and in implementation. Systemd is the opposite.

[coredump]

Well explained. "Too tightly integrated" might be more charitably put than some might say, but I believe it's pretty accurate.

We were just reading how it's becoming harder to use certain gui desktop environments (!) without systemd. Talk about "integrated"... how long before login shells start requiring it? </rhetorical, I hope>

And with that tight integration has come the creeping features. It seems as if each new systemd part that comes out is creating a base for some new one coming later. If not directly, then as a systemic replacement of the original Linux bits that were there before.

It's a bit shocking if you ever take a moment to look at all the systemd modules/units/whatever that could be installed if you enabled the lot -- even in a fairly conservative distribution like Debian -- how much of the OS systemd tendrils have gotten into.

[Eric 9001]

Yes, the systemd OS runs on the GNU/Linux OS.

[LionelB]

Perhaps we need to refer to it as systemd/Linux. Or systemd/GNU/Linux (at least until systemd inevitably takes over the GNU userland).

[tekHedd]

SystemD Linux

I suppose, considering, it would be GNU/SystemD. The Linux kernel is really just a module supporting SystemD.

Honestly, I see the headline and all I can think is "oh no what is he going to ruin now."

[Eric 9001]

Re: SystemD Linux

I agree that GNU/systemd is just as appropriate as GNU/systemd/Linux.

What is Linux if not a kernel?

[Doctor Syntax]

You've got it about face. Systemd graciously allows GNU to tun on top of it. For the time being. Also for the time being it runs on top of the Linux kernel but I suspect he has plans to fix that.

[LionelB]

He, he. At some point it'll be rebranded as SysDOS.

(I'm not even sure I'm joking.)

It's all a bit reminiscent of the old Emacs joke: "Emacs would be a great operating system if only it had a decent text editor".

[onefang]

I was waiting for someone to mention the emacs OS. So what will happen first, emacs OS gets a decent text editor, or systemd OS gets a decent init system?

[Eric 9001]

Emacs already has a decent text editor (you can even have vim keybindings via evil mode if insane).

You also have the option of running whatever editor you want in Emacs.

Meanwhile, don't hold your breath on a decent init from systemd.

[JLV]

I’ve come to wonder. Assuming that a successor to init is desirable, couldn’t people fork systemd, keep its init core -making it compatible with various service consumers - and ditch the rest of the Borg?

I’ve never written an init script, but I did hook up some services with runit. and systemd. Systemd, at least on VMs built with Ansible, tended to flake out more than runit, but was otherwise easy to use.

[HereIAmJH]

Assuming that a successor to init is desirable, couldn’t people fork systemd, keep its init core -making it compatible with various service consumers - and ditch the rest of the Borg?

Possible, but no telling how much cutting and massaging of code would be required to resolve inter-dependencies. And it wouldn't solve the bulk of the problem, apps being written to rely on systemd and expecting all it's components to exist.

[RAMChYLD]

Sadly.

Like KDE. They're losing BSD support because of a new DM that replaces SDDM from what I understand. the new DM is going to rely on logind exclusively, which is part of systemd.

I totally hate that. I thought the folks at KDE would be smarter than the folks at gnome...

[_wojtek]

erm, only new login manager will rely on systemd and you can still run any other login manager, SDDM included, on BSD.

please kindly avoid spreading false claims ...

[Peter Gathercole]

The comment was about KDE. If KDE becomes reliant on logind, and logind is part of systemd, and systemd does not run on BSD, then following the chain of dependencies, without systemd, you can't use KDE.

KDE used to run on BSD. It won't with that dependency chain.

[jfm]

This commit, titled "Drop code for FreeBSD support", would appear to suggest there's some truth in the statement.

[Anonymous Coward]

Yes, people have written several, but also no

Because what LP is really engaging in is more akin to campling a wiki and engaging in edit wars.

Without going to deep into how he and systemd got their foot in the door(aka Debian screwing up) Lennart worked for Red Hat and that is the upstream of most of Linux. Since it was and is the big dog on corporate Linux, and it had money to spend, as long as they backed the systemd teams hatchet job re-writes everyone else choked it down and avoided making comments about the moose turd pie in public. Some didn't know till it was too late how bad the sprawl would get.

So even though init technically would still work, and other init replacements have been made, replacing systemd now means rewriting literally dozens of other parts of the operating system. Or writing an maintaining a reliable comparability shim that LP the the rest of the systemd team will break a couple of times a week. Then you have to convince every other project and maintainer to ALSO do twice the work to maintain compatibility with YOUR init replacement and AND systemd's ongoing brokeness.

Or like so many here, put down the bag of rocks and just lend your support to BSD, or to projects like Slackware which would make a great clean core on Torvald's kernel, and could be reliable with a bit of funding and a few extra bodies.

So yeah, there are at least two replacments for init that aren't systemd that support deterministic loading order(the problem systemd was actually supposed to solve), and a whole non-linux OS that also powers every Mac and iphone on the planet.

[Michael Strorm]

Ah, you got there first. Systemd is indeed not, and was clearly never intended to be, a simple replacement for init.

That was merely the excuse the camel needed to get its nose in the tent.

[seldom]

Unfortunately, it was not even the nose end that entered the tent.

[John Robson]

It was a bit that smelled though

[JLV]

So not so much inside the tent, pissing out.

Rather, inside the tent, shitting in.

Pardonnez mon Français, svp.

[Anonymous Coward]

The arse end of a camel is far more pleasant than systemd.

[DS999]

It is a system

Designed to take away text files for configuration and logging that were easy to use/modify with common tools, and hide them away in databases requiring special software to access that makes previously simple tasks much more difficult.

[JLV]

Re: It is a system

... and not easily version-controlled, which is a pretty useful feature when dealing with configurations.

[Doctor Syntax]

""Other parts include a logging daemon,...."

The real objection is that all those existed previously but, in true NIH fashion Poettering felt the need to replace those as well.

[coredump]

I've theorized before in similar threads that replacing syslog, resolv, ntp, etc. isn't so much Poettering NIH, rather that he couldn't be arsed to work out using them the way he wanted, so he set about writing his own (incompatible, IMO) replacements. End result is the same as NIH though, regardless of motivations.

It might be fine if that's as far as it went, but it didn't stop there. And he doesn't seem to care if he breaks existing setups ("works fine here, WONTFIX"), or shifts blame ("upstream is doing it wrong"), or sometimes even dismisses the reporting person ("you're just doing it wrong").

Intended or not, it tends to come off as rather "my way or the highway". Unsurprising that some of us ended up choosing the latter; I doubt he's noticed, and probably wouldn't care if he did.

The other theory I've heard round these parts and elsewhere, is the Ship of Theseus plan, replacing bits of Linux until it's SystemDOS or the like. Dunno how plausible that might be, but I'm not waiting around for it.

[Fruit and Nutcase]

Succession?

Just thinking out aloud...

By jumping ship now, would it improve the chances for Agent P in the event of vacancy for grand fromage of the Linux Kernel?

The work on making Linux trustworthy sort of making amends/pennance for the divisiveness of systemd?

[steelpillow]

Re: Succession?

Remains to be seen how divisive he can make crypto/verification. Cut his teeth on PulseAudio, grew of age with SystemD, will this be his disruption pièce de résistance?

[martinusher]

Re: Succession?

I'm obviously 'too thick' or 'not worthy enough' but this thing called "Pulse Audio" has always given me a pile 'o trouble except when its used in the most simplistic way (i.e. one pair of speakers attached to computer etc.)

I can see what its trying to do but like a lot of systemd things it appears incomplete......its the 'works in my system' mindset.

Obviously if I want to devote my life to system voodoo I could use Windows. (This is the poster child for bugs masquerading as features.)

[Michael Strorm]

Re: Succession?

Bearing in mind that Poettering used the pretence that systemd was merely a replacement for init as a Trojan horse to get it where it is today, how much trust and assumption of good faith do you think he deserves this time round?

[wolfetone]

Re: Succession?

I'd trust him.

Not with anything important. Maybe he can keep guard of the paper straws at McDonalds. I'd trust him to do that. Anything else? No.

[David 132]

Re: Succession?

Gods, no.

Within a week he'd have (badly) reinvented the straws, with his new implementation being 16' long, porous above 220 kelvin, and requiring a complete reformulation of all the drinks in order to maintain compatibility. Reports of severe gum & lip damage from users would be brushed off.

[Michael Strorm]

Re: Succession?

And then it would turn out he'd got himself a suspiciously cushy job at arch-rivals Burger King...

[steelpillow]

the systemd ecosystem

And that's the problem.

Now it's going to grow crypto/verification management on top of everything else.

Whatever happened to "do one thing and do it well"?

[Arthur the cat]

Re: the systemd ecosystem

Whatever happened to "do one thing and do it well"?

Discarded as old fashioned. The new fashion is try to do everyhing, fail regularly, brick the machine occasionally and claim AI will solve all the problems which is why you're forcing it on everybody even if they don't want it.

[bombastic bob]

Re: the systemd ecosystem

Shhhh - don't mention AI in the same sentence with "the init system that must not be named"...

[Darkhand]

Re: the systemd ecosystem

He literally stated that his goal is to make Linux less Unix-like.

[wolfetone]

Re: the systemd ecosystem

It's when it finally got it's own text editor, that was the moment I knew he was full of shit.

[steelpillow]

Re: the systemd ecosystem

Is it a clone of the AI-enabled Notepad?

[TrevorH]

Feature creep

systemd-encrypt-everything.target

[David 132]

Re: Feature creep

Bug report: using the command unexpectedly encrypted all my data, my possessions, and even my cat!

Developer Response: Works for me. #WONTFIX.

[Neil Barnes]

Re: Feature creep

Obligatory xkcd: https://xkcd.com/26/

[David 132]

Re: Feature creep

Knew which XKCD that was before I even clicked on it. Good one!

[steelpillow]

Re: Feature creep

Quite lucky the butterfly algorithm has been replaced, really.

[CrazyOldCatMan]

Re: Feature creep

and even my cat!

My cat(s) are already sufficiently cryptic/cryptid thankyouverymuch..

(6 of them - ranging from 19 years old to 8 months)

[Michael Strorm]

Re: Feature creep

The "feature creep" isn't a bug, it was the entirely intentional point from the start.

As I mentioned elsewhere, the pretence that Systemd was simply meant as a replacement for init was merely the excuse the camel needed to get its nose in the tent.

[Anonymous Coward]

Re: Feature creep

More its rear end, judging by the amount of manure it managed to dump.

Makes you wonder if he hasn't left with a brief from Microsoft to really put an end to Linux, as he's well on his way.

Time to check out FreeBSD, I think.

[pitrh]

Re: Feature creep

> Time to check out FreeBSD, I think.

Or OpenBSD, see eg my now no longer quite recent (2021) piece, Recent and not so recent changes in OpenBSD that make life better (and may turn up elsewhere too), possibly supplemented by the long-ish What every IT person needs to know about OpenBSD and the positively short and snappy You Have Installed OpenBSD. Now For The Daily Tasks. which is about daily life with the system.

[steelpillow]

Re: Feature creep

Linuxes without Mini-Me are doing quite well - I am writing this on one. My Android phone has a Linux kernel with yet another init system.

It's more a case of whether, in an enshittified world, Enshittified-Linux distros can survive against competitors where even the kernel is enshittified.

[doesnothingwell]

Maybe?

The trio of terror started bossing around the windows kernel team, that would be their style.

[Anonymous Coward]

Re: Maybe?

Or he failed their tests...

[Anonymous Coward]

Re: Maybe?

Nah, they no longer do testing, that's left to the users now.

[ComputerSays_noAbsolutelyNo]

Anti Microsoft

"Every system starts in a verified state and stays trusted over time."

This sounds like a working definition of 'not Microsoft '

[ChoHag]

Re: Anti Microsoft

If it hasn't been verified and signed off by Ken Thompson I don't want to know.

[keithpeter]

Re: Anti Microsoft

"The moral is obvious. You can't trust code that you did not totally create yourself. (Especially code from companies that employ people like me.) No amount of source-level verification or scrutiny will protect you from using untrusted code."

The Ken Thompson who won the Turing Award and made a speech?

[Joe Gurman]

“The” other?

*cough*

[Anonymous Coward]

Sounds familiar

Didn't someone leave Microsoft and run Nokia (into the ground) ?

[m4r35n357]

Re: Sounds familiar

Yep. To kill Symbian.

Returned to base on project completion.

[Lomax]

Re: Sounds familiar

I think MeeGo was what Microsoft found so scary, seeing as it was by far the best mobile phone operating system the world had ever seen. Several parsec ahead of Windows Phone, and Android - arguably iOS as well - in terms of functionality and polish.

https://en.wikipedia.org/wiki/Nokia_N9

Despite a limited release, the N9 received widespread critical acclaim, with some describing it as Nokia's finest device to date. It was praised for both its software and hardware, including the MeeGo operating system, buttonless 'swipe' user interface, and its high-end features.

[powershift]

Dig it

Big slam in a professional way. Says a lot about working at Microsoft.

[Not Yb]

This isn't a move forward towards better personal computing.

Too much of this "trustworthy computer" business is aimed at making it possible for the corporate DRM to trust your computer, not the other way around.

This is not a move forward in my book, it's a move towards MORE control of other people's personal computers by making them trustable by Poettering and his company's customers.

[Anonymous Coward]

Re: This isn't a move forward towards better personal computing.

I mean, I get the paranoia the Pott head induces at his very mention. I absolutely don't want him writing this, but that doesn't mean that it shouldn't be written.

Root kits have been pretty evil for a while now, and state actors are playing games with supply chain attacks and wartime hacking things like hospitals and power plants. So if I am in the ICU getting open heart surgery, I'd rather the maker of the heart monitor have access to a signed and trusted LTS build of some flavor of unix then windows CE 5 off an old thumb drive that is shared between 12 factories in Shenzen.

Only down side is that the us three letters will have a harder time remote hacking the Iranian and North Korean nuclear programs again.

As to personal computing, I'd find some peace of mind if the open source community had a trusted platform they could hand to journalists that isn't just a thumb drive with TOR and old fork of Firefox on it. But that probably won't be running Red Hat, even if it is Unix.

Even then, trustworthy hardware is pretty hard to find.

[Not Yb]

Re: This isn't a move forward towards better personal computing.

Thing is, I don't think trustworthy hardware means what you think it means. Who originally backed the TPM that Microsoft, et al use these days? It wasn't users clamoring for it, it was corporations that wanted DRM that truly worked that were the loudest voices.

The computer user's personal security was an afterthought and a good marketing excuse for requiring it, but definitely not the original reason.

[tekHedd]

TPM

Exactly this: "trusted" computing is one of those inverse names. What it really means is "you are no longer trusted."

Adding this to Linux is just another way to take away control over your OS and your personal computing. SystemD was a fairly obvious and successful play to take Linux away from small developers. This is a move to take control from the end users.

Obvious power play is obvious.

[C R Mudgeon]

Re: This isn't a move forward towards better personal computing.

"[...] if the open source community had a trusted platform they could hand to journalists that isn't just a thumb drive with TOR and old fork of Firefox on it."

Maybe Qubes is what you're looking for.

Or, since you mention journalists, perhaps SecureDrop, which is a much more specialized to tool. (It runs on top of Qubes, I think).

[Doctor Syntax]

Re: This isn't a move forward towards better personal computing.

The problem with trust is who you decide to trust. If you're looking for a signed OS then you have to trust those who signed it. On the whole this is not corporates; even corporates who have been trustworthy for years or decades can be ruined with a change of ownership or leadership. As for the rest you have to trust them by what you see and what we have seen of Poettering does not, for many of us engender trust.

To return to your heart monitor - what if it had to be connected to the net to confirm its trusted status with the mothership on boot - or every hour and promptly shut down if it failed? Would you be happy with that?

[shodanbo]

Somehow this all feels like the Turing Halting Problem and the Godel incompleteness theorem both slapping "Unix" in the face

Either that or good, fast, cheap ... pick two.

[An_Old_Dog]

The Obvious Approach

... for Agent P and his unindicted co-conspirators to take to add "determinism and verifiable integrity" to Linux is simple. They will add functionality to systemd!

[Not Yb]

Re: The Obvious Approach

You've got this marked as a joke, but... it's very likely what they'll be doing.

LP starts from the assumption that OpenRC is old and buggy, and thus insecure and untrustworthy (not that I agree with them, but that's the basic assumption they make). If their new software is compatible with anything other than 'full stack systemd', it'll be surprising everyone.

[Qwerty44]

And here we have systemd in its natural habitat and the evolution just carry on, in this matter via the wrong path. So the disaster just keeps getting better and bigger...

First of all when I saw the story of systemd I was in shock for a week or two. With a right mind a would not allow a person to touch anything in Linux core whom was involved with M$crosoft! This is a matter of principle! And now they carry on "deliverying" the new disaster after systemd and "fixing" cripto...

To be frank the ssl lib was not really a great stuff latelly the swap from old to new libraries and incompatibility is far from great. But asking any person who worked for "them" is itself an abomination! Leave alone cripto and security, that part is tricky and deep, you already made enough mess in systemd disaster!

[_wojtek]

systemd is not that bad

I'll probably be downvoted to oblivion but having to prepare startup scripts for all distros because there were ever AI tiny differences of that is required or supported and actual handling of service live lines was abysmal I do like simplicity of systems and that "it just works" and works everywhere the same way...

[Doctor Syntax]

Re: systemd is not that bad

"having to prepare startup scripts for all distros"

Go and visit LibreOffice download. Check out how many options there are for Linux. There are just four. Two each (rpm & deb) for two architectures: aarch64 and amd-64. How do that manage that? Simple. They install into /opt (which is something that would probably have to go in a more potty version of Linux).

Now check out a lot of distros. You'll find that a good many of them have LO in their repositories and that these install into the distro's usual places. How does this happen if the developer didn't provide that? The distro has a maintainer for LO who looks after that.

The developer having to prepare installs for all distros is just another straw man - either that or the developer's not understanding the possibility of using that option.

[Lomax]

Re: systemd is not that bad

Also, as someone who doesn't use SystemD, I'm happy to own that and write my own init.d scripts for those rare cases where a) something must run as a service and b) that something does not have an init.d script already. It's not hard once you've done it a couple of times, and boilerplate code is readily available. Sure, it's more lines than a typical SystemD unit but very readable if you speak a minimal amount of Bash/Dash (which you really should do if you're using Linux). The only show stoppers would be packages that have hard dependencies on other parts of SystemD, but I can't remember encountering any of those. Even things that are typically part of SystemD can be run without it in some cases, e.g. Pulesaudio, though why you'd want to I have no idea. Also, Debian and its SystemD based derivatives can be switched back to SysVinit without too much trouble; I have done so with (headless) PiOS Bookworm for example. So the door is still ajar, if only just. The best way to ensure it doesn't shut completely is to continue to use and support the alternatives.

[sabroni]

Honest question

I asked google "what popular linux distros use systemd?"

it said:

Most major, mainstream Linux distributions use systemd as their default init system.

Popular Distributions Using systemd

Ubuntu: Switched from Upstart to systemd in 2015.

Debian: Adopted systemd as the default starting with Debian 8 "Jessie".

Fedora: Was the first major distribution to implement systemd as its default.

Arch Linux: Officially supports only systemd for its default installation.

Red Hat Enterprise Linux (RHEL): Uses systemd for service management and booting.

openSUSE: Implements systemd across its Leap and Tumbleweed releases.

Linux Mint: Based on Ubuntu/Debian, it uses systemd by default.

Manjaro: A popular Arch-based distribution that uses systemd.

Pop!_OS: Developed by System76, it relies on systemd for system initialization.

So, if it's so awful, why does everyone use it? It's not like the CEO of the corporation told them to. All I hear on here is people complaining about it, you seem to have a valid point (it does too much which isn't in line with linux philosophy) so why is it everywhere?

[_wojtek]

Re: Honest question

you will be told that's a global conspiracy or whatnot...

yet noone wants to admit that systemd tackled the cluaterfuck that was sysvinit...

[R Soul]

Re: Honest question

No! systemd replaced the clusterfuck of sysvinit with an even bigger and uglier clusterfuck. FTFY.

[Michael Strorm]

Re: Honest question

Even arguing against systemd in terms of whether or not it's a good replacement for sysvinit plays into their hands and misses the bigger picture.

Systemd has grown far beyond- and long since ceased being- a replacement for init alone. And, as noted elsewhere, even that pretence was only ever the excuse it used to get its foot in the door before taking over.

[Missing Semicolon]

Re: Honest question

Init needed sorting out. Something like systemd v1 might have been a candidate for the representation of the init requirement. It was seen as an improvement. But now it's in, the requirements of a small set of users is driving the development, so that the rest of us keep finding stuff is surprising.

[Doctor Syntax]

Re: Honest question

I've been using System V and then Linux with SysV init since the late '80s & not found this clusterfuck of which you write. If need be I can dive into /etc/init.d and find out exactly how something's started up. It's just a shell script. Alternatively I can go to the run level directories and see the sequence in which things are started if that's critical. It's transparent.

I don't, of course, usually need to do that unless there's some deep problem and need to put some sort of debugging statements in there.

[_wojtek]

Re: Honest question

ah yes, the joy of constant fiddling with init scripts, which usually are monstrosity. yes, not missing having to look at them or make them.

[Anonymous Coward]

Re: Honest question

If you're 'constantly fiddling with init scripts' you're doing something very unusual. I've been running openRC for years, and the only time I've had to fiddle an init by hand is... never.

[coredump]

Re: Honest question

I can't say I've *never* fiddled with init scripts, but it has been a very (!) long time since I actually needed to. Circa RHEL3 or 4 or perhaps even Red Hat (no bloody E) Linux.

And even then, I think "fiddling" is up for discussion. E.g. temporarily tacking a "-x" on the end of the interpreter to see what it's getting up to, or a couple strategically placed echo for some printf style debugging; but nothing which fundamentally changed what the script was doing, and reverted when I was done.

Now, I have written a couple init scripts for homegrown tools at a previous $JOB; fiddling or not, it wasn't difficult. The script itself was nothing special, working out how and where the K/S symlinks needed to appear, playing nice with 'chkconfig' et al, even absent the man pages it wasn't a huge task.

By contrast, trying to do something which I thought should be pretty straightforward ("run this thing after the network is up") was apparently considered out of bounds by the systemd folk -- there is/was a fairly well-known wiki page about the topic back then, and not just a few bug reports which were presumably WONTFIX'd. Maybe things are better now, but back then examples were scarce on the ground, or so simple/trivial as to be not useful.

So IME it's not necessarily about how often you need to do something; to me it's also about how much of a chore it is to work it out when you need to.

[David 132]

Re: Honest question

Well, I guess we've found the person responsible for anonymously downvoting all the comments above that dare to criticise Systemd. Keep tilting at those windmills!

[Lomax]

Re: Honest question

I've exclusively used Linux for the last 15 years, on my desktop, laptop, various servers and embedded systems that I build and maintain for a living - the majority of which still use SysVinit. I can honestly say that I have never, ever, had any kind of trouble with SysVinit of any kind whatsoever. In my experience it just works, and it's (almost) as easy to add custom services to as SystemD. I totally fail to see what "problems" SystemD solves - though I know well the new problems it creates, for example the inexplicable pain of binary log formats, and the confusing mess of dependencies. I prefer SysVinit and find it easier to understand and configure. Am I allowed to?

[Michael Strorm]

Re: Honest question

"systemd tackled the cluaterfuck that was sysvinit"

Systemd may have got its foot in the door under the pretence that it was merely a replacement for sysvinit.

But whatever one thinks of it in that role, it's grown far beyond that, and should be judged as the monolithic takeover for large swathes of the OS it now is, and was always intended to be.

[tatatata]

Re: Honest question

From Arch-linux, I remember that there was a justification on Reddit why they switched to systemd. Basically, they stated that they made a mess of the initscripts an when systemd came along, they stopped thinking and went like sheep with the crowd.

[Claude Yeller]

Re: Honest question

"So, if it's so awful, why does everyone use it?"

It isn't awful, just a complex answer to a complex problem.

Every big distribution uses it because it helps them maintaining a healthy Linux installation. There have been quite a number of systemd-free forks. But as they don't get the users and developers, they don't get any traction.

Systemd is a repeat of pulseaudio, where Lennard was vilified for his design, which most people ended up using because it was better than the dumpster fire preceding it

[Will Godfrey]

Re: Honest question

Funny that. On the few occasions I need to do a complete install, one of the first things I do is delete the pulseaudio server. After that all my audio software works reliably, exactly following the paths I want it to, not some brain-dead autoconnect nightmare.

[Claude Yeller]

Re: Pulseaudio

I remember more than 2 decades of fighting to get audio working on Linux, recording&playing. Starting anew with every install.

I welcomed auto-everything for audio.

Anyhow, anyone who doesn't want to use pulseaudio still is free to use something else. But the majority just uses pulseaudio.

However, vilifying someone for delivering an extra choice seems to be the FLOSS way of Free&Open.

[steelpillow]

Re: Honest question

Because it does solve some endemic problems and makes life easier for both the distro builder and the app/utility developer. At least for now.

However for power users, it has replaced an eatable plate of spaghetti (calls) and meatballs (utilities) with an indigestible interpenetrating tangle that nobody except the chefs are able to understand or unravel. So customising, troubleshooting, or working round bugs, become impossible.

Looking forward, it will become so overblown, buggy and unmaintainable that it becomes unreliable, unfixable, and vulnerable to exploit, taking down so much of your system functionality when it barfs or borks that the whole system explodes.

Since most of us here are experienced power UNIX/LINUX users, our attitude is a foregone conclusion.

[coredump]

Re: Honest question

Except it's not entirely an honest question, is it.

I mean, you started with a Bandwagon Fallacy, and followed with the claim that a corporation wasn't driving systemd adoption, when that's what Red Hat did by virtue of being the proverbial 800lb gorilla in the Linux room. I think you'll find that many things about today's Linux started with Red Hat in some fashion.

I'll grant that Debian (probably) made their own decision, and I've bemoaned that choice in these forums before. But to think they weren't at least influenced by Red Hat's path seems naive.

If you're genuinely curious how things got to this point, there are stories and anecdotes about what went on inside Red Hat in the early days of systemd, which talk about the internal debates, project managers and marketing motivations etc. Some Debian discussions are out there too.

I've read some of it over the years, it's not too hard to find; I can't speak to the veracity, but I will say they're interesting reading from a historical perspective.

[Claude Yeller]

Re: Honest question

Here is an answer from someone who actually knows what they are doing:

Why I'm Sticking With systemd-based Linux Distros

Traditionally, [with SysVinit], if you added a piece of hardware, even something like an external disk drive, you would have to shut down and restart the system. SysVInit was also convoluted, with shell scripts corresponding to "runlevels." This approach was inadequate as Linux became more widespread. With modern machines, you might plug in a USB drive or move between Wi-Fi and wired networks with a laptop. systemd can respond to such "hotplugged" hardware instantly.

There is much more at the link.

[sabroni]

Re: Honest question

Sweet, thanks, I'll have a read!

And more thanks for not being a dick about it!

[Claude Yeller]

Re: Honest question

Your welcome. It was the relevant question to ask.

[Dan 55]

Oh dear

Now he's free to mess around with Linux again.

[Doctor Syntax]

Re: Oh dear

I'm sure he was fully encouraged to do so while working at Microsoft.

[Apocalypso - a cheery end to the world]

Don't ask MS why he left...

> It is unclear why Poettering decided to leave Microsoft. We asked the company to comment but have not received a response.

I'm not really that interested in why he left Microsoft but I would be interested in what Microsoft thinks was his main contribution while he was there.

[Wargasm]

Leading torch doesnt always mean "leading lights"...

[Blackjack]

I don't like systemd but I guess Microsoft current focus on AI down everyone thoats was finally enough for him?

He seemed perfectly fine lwith everything else Microsoft is doing.

[Michael Strorm]

He can't shove systemd down our throats if MS is already doing so with AI.

[Anonymous Coward]

systemd and its journey towards taking over the world !!!

I never understood what 'problem' was being solved by systemd !!!???

The reality was that 'some' people had a problem using init (couldn't get their head around how it worked or something !!!) ... LP pushed systemd as a 'solution' to replace all the init scripts and solve the issues.

When attempting to solve a problem you can do one of two things:

1. Identify the 'real problem' and solve it !!!

2. Change the 'real problem' to something else then attempt to solve that !!!

LP did option 2. then when it was obvious that the 'real problem' had not been solved option 2. was applied again ... repeat until bored !!!

Each time the 'real problem' evolved and the solution grew encompassing more and more of the functionality of the OS.

That is why systemd is so big and has tendrils running throughout everything ... (or so it seems).

The idea that 'If I could just encompass enough, I would be able to finally solve the problem' is really going to work, seems to never die.

Systemd's scope has grown to be huge, yet it is never quite enough !!!

After all this time the actual solution is 'staring you in the face' ... go back to the init system as originally written and simply teach people how to use it properly !!!

I live in fear of what the next excitement from LP will be, and what 'problem' it will attempt to solve this time !!!

:)

[Philo T Farnsworth]

Re: systemd and its journey towards taking over the world !!!

I've got an idea: Let's all go back to hand editing rc.local.^1,2

__________________

¹ That's a joke, son. Why aren't you laughing?

² Apparently that is still a thing. See here, here, etc. Who knew? Obviously, not me.

[coredump]

Re: systemd and its journey towards taking over the world !!!

Thanks for the Friday flashback. :-) I think the first (and maybe last?) time I directly edited /etc/rc.local (or equivalent) was SunOS 4.1.3 . As I recall, there really weren't many other mechanisms back then to change e.g. daemon startup flags and args and such.

We didn't have that many Suns though -- it was mostly an SGI shop -- so I'll freely admit I may not have been following BCP for the time.

[Philo T Farnsworth]

Re: systemd and its journey towards taking over the world !!!

SunOS 4.1.3 is about where I last touched rc.local myself. About the time Solaris¹ came out, we hired an actual system administrator, so I gratefully backed away from all of that as quickly as humanly possible.

We had a mix of SGI and Sun in our lab and when I arrived there, the rc.local on the various machines was a bit of a hacked up nightmare, having been largely cobbled together by a succession of grad students and undergraduate interns. Not something to be approached or modified without proper HAZMAT gear.

___________

¹ The operating system, of course, not the 1972 Tarkovsky film of the same name.

[Bebu sa Ware]

Re: systemd and its journey towards taking over the world !!!

Especially in places where there were a mixture of Unixes SunOS, SGI Irix etc admins used rc.local to recreate the System V start up (without runlevels.†)

Meant if were sufficiently cunning clever you could use the same script on the different platforms. I even bothered to implement Irix's chkconfig.

Later FreeBSD did something similar (but different.)

_{† typically start/stop scripts in /etc/rc.d with active links in /etc/run.d lexically ordered by numerical prefixes to the link's name.}

[Lomax]

Re: systemd and its journey towards taking over the world !!!

I like sysv-rc-conf, but to be honest most of the time I just use update-rc.d.

[steelpillow]

Re: systemd and its journey towards taking over the world !!!

There have been many forks of the various Linux init systems, sysvinit and systemd being no exception.

AIUI there are issues around how you manage dependencies, to ensure that A is properly up and running before you start up the B which depends on it.

We can't agree how this should be done, be it by package managers or init systems.

IMHO we are heading to a place where there are three major clades of Linux-kernel based systems; GNU/Linux, Android, and SystemD/Linux. I mean, why include vim, emacs or gedit when SystemD already has a text editor in there?

[Bebu sa Ware]

Re: systemd and its journey towards taking over the world !!!

"I never understood what 'problem' was being solved by systemd !!!???"

More concisely Poettering &co dosn't seemed to have understood the basic problems with systemv init.

Having constructed countless start script over four decades I found that you expend a lot of effort on basic boiler plate code:

• checking the service's binaries and configuration files are present
• checking the service isn't already running
• checking any prerequisites are installed, and if services, that they are running
• after starting the service taking care that the process is demonised, checking that is actually running

Even at that point if the service has to be monitored and restarted attempted on failure you have run another service.

A minimal systemd by itself did more or less address these issues. I found constructing unit files under RHEL7 a lot less labour and not unpleasant. Once systemd took over autofs (automount) the lunacy seems to have set in.

Alternatives for looking after the system start up and services weren't new. Before systemd - DJB's daemontools had been around since the Ark; Solaris' service management facility (smf) dates from 2006. Linux has had quite a few init replacement contenders before systemd and still has.

My main objection to the systemd circus is that it is not easily decomposable - if take one bit you mostly have to take the lot; including, unfortunately, the clowns.

[Anonymous Coward]

Because clearly the world needs another Linux distro.

[Philo T Farnsworth]

Sure, why not?

While I understand the viewpoint of the Linux advocates¹ to make "next year the year of the Linux desktop,"² my personal view is Why not? The more the merrier.

After all, for example, there are many different brands of automobile, refrigerator, or toaster, all with different features but, at the core of it, they all get you from Point A to Point B, keep your food cold, or toast your morning bagel, respectively (one hopes), and I don't see anyone complaining about that.³

I think different distros are actually a sign of a healthy ecosystem. If someone wants to do something in a unique or idiosyncratic way, great, maybe we'll all learn something new. Even if it's just an ego trip or a petulant snit, if nothing else they can still serve as a cautionary tale, an example of what not to do.

Evolution in action.

If you don't like a distro, don't use it, in the same way that if you don't like a car's styling (or whatever), you don't have to buy it.

Buy something else. Or run something else. Who's going to stop you?

Finally, consider this: The corporate world is largely a desktop monoculture and we've seen what happens to half the world when a bogus update gets pushed out.

_________________

¹ I was going to write "evangelists" but, sorry, an operating system is not a religion nor should it be.

² In the same Red Queen's race manner that practical fusion power is 20 years in the future.

³ Maybe the Soviet Union had just a single brand of each but that didn't turn out especially well, did it? I'll have to ask my Russian colleague the next time we talk.

[Oh Homer]

Please God No!

I just started a GoFundMe campaign to pay Poettering to go back to Microsoft.

[pts1967]

Some say he's the Thomas Midgley Jr of computer science.