Yes, he specifically goes out of his way to refer to init as generically as possible. But then drops this line at the end of the post:

And yes, a large part of this may be that I no longer feel like I can trust "init" to do the sane thing. You all presumably know why.

It's not hard to understand what he meant and for some of us it's articles like this on HN which give an opportunity to talk about systemd without being off topic.

IMO the take away from the article is that systemd has become enough of a moving part that it would be unwise to rely on it for default anything. Maybe I read it wrong but there's my take on the technical argument against using init to set default limits on processes. There are already defaults in the kernel so use them and allow an override for admins that need one.

It's impossible to misinterpret that he's referencing systemd, but that's not the point of his post. He's saying, "Kernel over here, Init over there, One way transaction". You're talking about whether Systemd is qualified to be init. He's talking about whether the Kernel-side or Init-side is responsible for defining a given value. He declares his bias clearly (trusts Kernel, doesn't trust Init, especially doesn't trust Init via Systemd) and then makes his point. I'm just deeply saddened that literally two comments out of a hundred bothered to discuss his point, that divide, at all.

So talk about that divide. Clearly a lot of people care a lot about Sysemd and want to talk about it. It's even the title of the post.

Complaining that people aren't having the conversation you want to have is just weird. Start the conversation you want instead of complaining that others are failing to have it. No one is obliged to hold the conversation you want.

I came here to learn about a technical issue that HN felt worthy of the front page, because I'm not certain how I feel about systemd yet. I ended up learning what a bunch of people feel about systemd, but nothing about whether this is a reasonable divide strategy, or whether anyone has any arguments to counter his, or .. anything whatsoever. I'm glad the thread benefited the wellness of the community by providing an outlet for unresolved rage, but y'all missed a key opportunity here to consider a new semantic about the Kernel-Init(Systemd) interaction before it's fixed into stone by commits or whatever.

Was the title different when you clicked on this story? If you clicked a story about how Linus no longer trusts init and expected some deep discussion of init/kernel separation, I kind of feel like you were destined to be disappointed. The topic was clearly Linus's distrust of init, not architectural purity.

The suckless community maintains a page about SystemD's flaws (Why it "sucks"), and what they think an init system should do, and what it shouldn't. It's a relatively concise summary of the position of the people opposed to SystemD: http://suckless.org/sucks/systemd

So far I haven't found a page like this by the SystemD advocates. I haven't really found much of an argument by them, to be quite honest. It would be nice to hear of one, of course.

Yes, exactly. Great post.

> for some of us it's articles like this on HN which give an opportunity to talk about systemd without being off topic.

I would argue it is off topic because it's barely relevant to the submitted link. Just post "Tell HN: systemd sucks" and you can talk systemd all day long.

You could even include some actually real and concrete concerns, like trying to bloat the kernel with some linux-on-teh-desktopz IPC which will probably be replaced by the next big thing by 2025, bundling some weird syslog replacement, bundling a half-assed DNS cache (it was originally vulnerable to spoofed answers right off the bat, besides the recent remote code execution oopsie), polluting dmesg with systemd log messages, asking the kernel to workaround systemd crashing when the kernel is booted with debug output, Poettering's dismissal of the UEFI bricking fiasco, the recent failure to drop privileges on usernames starting with 0, ...

You guys behave as if some overblown pretext were needed to flame systemd :p

I don't understand your complaint, especially given that your last two links are more discussion about problems with systemd instead of the divide you think is the interesting point.

I apologize, I'm not sure how I could improve it to make it more clear.

It's really confusing that you complained that no one is discussing the architectural decisions separating kernel from init but then you link to a comment that discusses "1) systemd's bad programming taste and complexity, and 2) Lennart Poettering's inability to admit when he is wrong." Your (second) link is 75% about the interpersonal failings of Poettering and 0% about the separation of kernel and init.

Also with respect to your comment: "I am disgusted to report that ... the 2 above-linked comments are the sum total of actual replies to the content Linus was trying to discuss." Linus did not post this here and in no way was attempting to discuss this with the HN crowd. cnst posted this with the intention of discussing Linus's apparent distrust of init.