Systemd 237 Released With WireGuard Support, Keyboard See-Saw/Rocker Changes

[starshipeleven]

Pictures from the last systemd developer's conference:
And you call yourselves trolls... amateurs.

Damn, that's a quality trollpost there. I'm happy to see the USA propaganda machine is still top-notch.

[unixfan2001]

Damn, that's a quality trollpost there. I'm happy to see the USA propaganda machine is still top-notch.

If he lives in the US, he ought to be deported to Russia.

Comparing Poettering to Hitler is also in fairly bad taste.
I mean. One of them murdered 6 million+ people. The other maintains a system management framework some people don't like.
Totally the same thing...

[Guest]

Damn, that's a quality trollpost there. I'm happy to see the USA propaganda machine is still top-notch.

Impressive effort indeed.

[ypnos]

Comparing Poettering to Hitler is also in fairly bad taste.
I mean. One of them murdered 6 million+ people. The other maintains a system management framework some people don't like.
Totally the same thing...

It is even more cynical than that. Mr. Poettering's biggest sin is being exceptional at his job. In a way that reaps huge benefits to the GNU/Linux desktop. While the silent masses profit from his work daily without even knowing his name, a small bunch of bitter sysadmins lets out their frustration and targets him personally. Would these guys also shit on Stallman for establishing GNU or GCC?

[starshipeleven]

While the silent masses profit from his work daily without even knowing his name, a small bunch of bitter sysadmins lets out their frustration and targets him personally.

Most systemd detractors aren't actual sysadmins... just saying.

[cybertraveler]

It is even more cynical than that. Mr. Poettering's biggest sin is being exceptional at his job. In a way that reaps huge benefits to the GNU/Linux desktop. While the silent masses profit from his work daily without even knowing his name, a small bunch of bitter sysadmins lets out their frustration and targets him personally. Would these guys also shit on Stallman for establishing GNU?

That's a misleading way of framing it. Systemd very obviously does bring advantages over popular alternatives, but it also has disadvantages. Some of them being:

• An additional single, central point of attack is created that a person or organisation could target to cause damage to large numbers of GNU/Linux users. So if you're someone who wants to gain non-consensual access to large numbers of GNU/Linux systems, you are no longer limited to attacking Linux (the kernel) or GCC (the most popular compiler), you can now also target the init system and the other near-ubiquitous systemd components.
• As the decentralised process of building conventions, protocols and best practises around init and userland are replaced by a more centralised, permissioned process managed by the gatekeepers and leaders in the systemd project, GNU/Linux will have its evolutionary pressures restricted or directed. Large centrally planned systems usually yield inferior results to smaller, independent systems that freely compete. Of course, users will still have the choice to use other complete, competing systems (eg Windows, Mac OS X, FreeBSD, NetBSD etc), but there will be a net reduction in experimentation, choice and thus evolution.
• Distros that don't like particular aspects of systemd which are imposed by the architecture of systemd, will likely accept those aspects in order to retain the benefits of using the rest of systemd. Prior to systemd, they likely would not have had to make this compromise: they could have selected all the best components they wanted with far fewer restrictions. Anyone who claims this particular point is wrong is likely being disingenuous, because one of the widely claimed benefits of systemd is that it introduces a number of standardized interfaces to GNU/Linux allowing developers targeting GNU/Linux to use these interfaces and expect their code to run on a very large number of distros. This benefit is a double edged sword: when you standardize something like this, developers come to expect the interface to be there and thus distros are pressured to include it or risk becoming incompatible.

I know all of my listed points are very abstract. I think the fact that the problems created by systemd are so abstract, is partly why many GNU/Linux users don't understand what all the fuss is about. Thinking in a very abstract fashion is hard. The immediate, non-abstract benefits of systemd can be comprehended by almost everyone (faster, more standardized, new features added on every system, less duplicated efforts). The abstract issues (like the ones I highlighted above) are less easily understood / comprehended.

I'm definitely not saying that everyone who fully supports systemd is less capable of more complex abstract thought. I am saying though that there are likely a large number of systemd full-supporters who do struggle to comprehend this stuff (we are not all born equal).

[arokh]

^^

s/abstract/clueless/g

Distro's are free to implement the parts they want or none at all. As always, the anti-systemd crowd is unable to bring up real problems with systemd. Now we even have "abstract" problems.

[Staffan]

I hope they have added a feature where system logs are rendered into jpeg files who are then assembled into pdf documents stored in a database. That will make it so much easier to access and read logs when troubleshooting issues.

[cybertraveler]

Distro's are free to implement the parts they want or none at all.

When an interface is defined by a single implementation and that implementation changes the interface over time as and when its developers see fit, distros cannot simply create an alternative implementation as doing so risks their distro being incompatible with applications which depend on subtle implementation details of the "official" implementation.

When you have a separate standards group or even just a set of popular shared conventions (as was the case prior to systemd), you have much more assurance that applications will be portable and less reliant on the implementation details of a single implementation. Developers are forced to be more liberal in what they accept and conservative in what they send in order to be maximally compatible.

As always, the anti-systemd crowd is unable to bring up real problems with systemd. Now we even have "abstract" problems.

It's interesting that you have associated me with an anti-systemd crowd. I am neither anti-systemd nor do I speak for that crowd (assuming there is such a thing). Your comment makes me wonder why you'd say that. I expect you don't have a strong rebuttal, so you are falling back on primitive sophistry (mis-characterization, poising the well and transfer of values by association). Your "clueless" comment is a simple ad hominem attack.

[starshipeleven]

That's a misleading way of framing it.

A good FUD post, kudos to you. Looks technical and has the right amount of logic in it to fool people that don't know enough of the topic.

But it is not fooling me. Now let's debunk it.

[*]An additional single, central point of attack is created that a person or organisation could target to cause damage to large numbers of GNU/Linux users. So if you're someone who wants to gain non-consensual access to large numbers of GNU/Linux systems, you are no longer limited to attacking Linux (the kernel) or GCC (the most popular compiler), you can now also target the init system and the other near-ubiquitous systemd components.

I don't know why you say "additional" (well, I know, this is a FUD post, but let's pretend I don't).

I mean, before distros still had an init system, usually with init scripts. So far they have shown to be MUCH easier to exploit than systemd init. Other systemd components are not terribly exploitable since they don't run with root privileges anyway, and are usually sandboxed to some extent with namespaces and other linux kernel features (which other programs never really used seriously because the init itself was unable to configure this easily).

[*]As the decentralised process of building conventions, protocols and best practises around init and userland are replaced by a more centralised, permissioned process managed by the gatekeepers and leaders in the systemd project, GNU/Linux will have its evolutionary pressures restricted or directed. Large centrally planned systems usually yield inferior results to smaller, independent systems that freely compete.

Facts would seem to contradict this statement, before systemd involvement many critical system components were woefully inadequate (init systems mostly) or barely maintained at all (Dbus). Still now not much has changed, the only serious init system around is OpenRC, which is still inferior to systemd by a long shot.

The truth is that while your statement is technically correct with the appropriate context, if used ignoring context becomes invalid FUD bullshit. A great choice for a good FUD post.

If the smaller systems competing freely are too small (i.e. their developers have too little resources) then what they can realistically do even when competing in a hypotetical free market will still be negligible when compared to a larger centrally planned system that actually has significant amounts more resources than all of them combined (and a decent leadership with a sound plan too).
It is just a matter of scale. If all projects have more or less the same resources, then we can call of competition being useful, if one of the projects has massively larger resources than the others, it will be as good as its leadership allows.

Which is the case here. RedHat simply has much more resources and access to more talented people, which are working fulltime on that. Sure it does receive significant contributions by developers from Devian and Arch, but the core RedHat developers are still enough to roflstomp anything else.

Of course, users will still have the choice to use other complete, competing systems (eg Windows, Mac OS X, FreeBSD, NetBSD etc), but there will be a net reduction in experimentation, choice and thus evolution.

Again this is technically correct with the right context, but again the lack of context makes this look like a one-sided thing.

As said above, experimentation, choice and evolution that happen in small projects is still not going to yield comparable results to a single bigger project.

The obvious example would be the Linux kernel, or GCC. We can apply the same broken FUD bullshit arguments of yours there. Focusing all efforts on a single larger project is depriving users of freedom of choice, experimentation and stifles OS evolution, ok (maybe, I guess?)

But would having many different kernels worked on by a fraction of the developers working on Linux kernel yeld anywhere near the results we are getting with Linux? Nope, it sure won't.

Same for GCC. There is only ONE similar-sized competitor (LLVM/Clang) and even that got massive investment from multiple major companies, and it is still trading blows with GCC.

[*]Distros that don't like particular aspects of systemd which are imposed by the architecture of systemd, will likely accept those aspects in order to retain the benefits of using the rest of systemd.

This is plain false. Systemd init can work much alone, you can use its logger or keep using the pre-existing logger program, Dbus works fine also without systemd (yes it pulls in a libsystemd but that's just a shared library), and most other systemd project binaries are standalone tools like the systemd-boot, or have a systemd-less equivalent like logind (which is made possible by the fact that there is significant modularization going on).

I know all of my listed points are very abstract. I think the fact that the problems created by systemd are so abstract, is partly why many GNU/Linux users don't understand what all the fuss is about. Thinking in a very abstract fashion is hard. The immediate, non-abstract benefits of systemd can be comprehended by almost everyone (faster, more standardized, new features added on every system, less duplicated efforts). The abstract issues (like the ones I highlighted above) are less easily understood / comprehended.

Very abstract statements also greatly help FUD arguments. For example, if you actually made examples of your statements it would be far harder to sound believable.

I'm definitely not saying that everyone who fully supports systemd is less capable of more complex abstract thought. I am saying though that there are likely a large number of systemd full-supporters who do struggle to comprehend this stuff (we are not all born equal).

Beware of people that not only comprehend what you post and can also see through your disguise and help lesser minds to do the same.

As someone said "there is always a bigger fish".