Debian Developers Decide On Init System Diversity: "Proposal B" Wins

[muncrief]

you basically right but other init systems and their fanbois are so irrelevant this days they just make a huge thing about any tiny detail to try an maintain this image that non-systemd community is way bigger than the actual 50 guys whining on every space possible. eventually i expect reality will hit them, for now just not reply to them is usually enough to let the thread die out until the next tiny thing will come

Yes, it is fascinating. And I've never really understood why they actually seem to want to fight about it. I sincerely don't care what init system someone wants to use, and in fact whatever it is I say more power to them. Most distributions use SystemD now simply because they've found it's the best init system currently available, and don't have the resources to support more than one by default.

But it's always been the case that people can use whatever init system they choose, and that's one of the many wonderful things about Linux. In fact anyone who feels strongly that another init system should be the default can actually spin their own distro, and then others who agree with them can use it.

Another one of the many fantastic things that make Linux so awesome!

As Austin Powers would say, It's all about choice baby

[ALRBP]

you basically right but other init systems and their fanbois are so irrelevant this days they just make a huge thing about any tiny detail to try an maintain this image that non-systemd community is way bigger than the actual 50 guys whining on every space possible. eventually i expect reality will hit them, for now just not reply to them is usually enough to let the thread die out until the next tiny thing will come

Non-systemd community is most Gentoo people + Devuan, Artix, etc. Gentoo has its own init system, OpenRC, since long. They may not be the majority of Debian's devs but systemd fanatics are not either, since the vote did not end-up rejecting any non-systemd work.

This vote will not end any drama through ; because allowing maintainers to choose while enforcing collaboration with non-systemd downstream (notably Devuan) will definitely end-up in fights between some systemd fanatic maintainers who will refuse non-systemd contributions and people who will propose such contributions.

What I find interesting is that pro-systemd are always complaining about people who still use SysVinit (Devuan) but, if Debian had gone with OpenRC instead of systemd, I do not think anyone would have bothered to fork Debian just to keep using SysVinit. SysVinit would probably be completely dead now if OpenRC has been chosen to replace it instead of systemd. Also, maybe if systemd hasn't tried to be a complete OS instead of just an init system, more people would have ended up liking it.

[Ironmask]

Yes, it is fascinating. And I've never really understood why they actually seem to want to fight about it. I sincerely don't care what init system someone wants to use, and in fact whatever it is I say more power to them. Most distributions use SystemD now simply because they've found it's the best init system currently available, and don't have the resources to support more than one by default.

But it's always been the case that people can use whatever init system they choose, and that's one of the many wonderful things about Linux. In fact anyone who feels strongly that another init system should be the default can actually spin their own distro, and then others who agree with them can use it.

Another one of the many fantastic things that make Linux so awesome!

As Austin Powers would say, It's all about choice baby

Legitimate question to any long-term Phoronix forum users, has anyone had any actual valid criticisms of systemd's init functionality? Systemd is primarily a service manager, I don't know what it has to do with init, in fact init doesn't even do anything. I've always been very confused at why systemd is an init system (I'm sure there are valid reasons, this isn't a criticism, it's a modest question) and why people care so much that it's an init system. People seem to care far more that it's a service system that many pieces of software require to support service functionality, which I disagree with those criticisms since Linux is long overdue for some user-friendly consistency and coherency that Windows and macOS have enjoyed for decades before Linux. But people seem to skip over the entire service system functionality and consider systemd "just" an init system that everything "just so happens" to depend on "for some reason".

[Ironmask]

Also, maybe if systemd hasn't tried to be a complete OS instead of just an init system, more people would have ended up liking it.

Everyone does like systemd, that's why every distro that isn't 5 or so fringe distros use it. Yes, systemd is trying to be a complete OS, because Linux isn't. Windows is a complete OS, it has it's own standard API interface with services, registry, windowing API, everything. Same for macOS. Linux does not have this, it has a fractured, broken community where there are 10 different standards for every API you can think of and none of them are compatible. People do not want this, they want systemd. Linux is not Unix, it is not a hobby OS anymore, it is a real, industry standard with measurable market presence, even in the desktop sector, however meager that is currently. And rejecting systemd is not going to help that, it's going to hurt it. If you want a hobby OS where everything is incompatible, there are plenty out there for you. There's no reason to latch on to Linux to enforce it to stay a fractured mess just as there's equally no reason for you to do that to Windows or macOS.

[ALRBP]

Everyone does like systemd, that's why every distro that isn't 5 or so fringe distros use it. Yes, systemd is trying to be a complete OS, because Linux isn't. Windows is a complete OS, it has it's own standard API interface with services, registry, windowing API, everything. Same for macOS. Linux does not have this, it has a fractured, broken community where there are 10 different standards for every API you can think of and none of them are compatible. People do not want this, they want systemd. Linux is not Unix, it is not a hobby OS anymore, it is a real, industry standard with measurable market presence, even in the desktop sector, however meager that is currently. And rejecting systemd is not going to help that, it's going to hurt it. If you want a hobby OS where everything is incompatible, there are plenty out there for you. There's no reason to latch on to Linux to enforce it to stay a fractured mess just as there's equally no reason for you to do that to Windows or macOS.

So, what you want is just an open source Windows. I definitely not agree with this. Even if Windows was open source, I would not use it, and this may actually happen, with the direction MS is taking toward open source. Happily, systemd/Linux is still not Windows. Also, you talk about standards, but systemd has caused more compatibility issue than it has solved. Creating a new "standard" do not solve standards fragmentation issues.

[Ironmask]

So, what you want is just an open source Windows. I definitely not agree with this. Even if Windows was open source, I would not use it, and this may actually happen, with the direction MS is taking toward open source. Happily, systemd/Linux is still not Windows. Also, you talk about standards, but systemd has caused more compatibility issue than it has solved. Creating a new "standard" do not solve standards fragmentation issues.

I don't understand this common criticism of calling systemd "Windows" as if it were some demon to be expelled, it's quite dismissive. No, I don't want Linux to be Windows, nor is systemd anything like Windows. What I want is Linux to have what Windows has: usability. After that, market demand will follow. Systemd is a stepping stone in that direction.
Yes, systemd is "yet another standard", the difference with this one is, it's a standard everyone is actually following. And that's a good thing, it's something to be celebrated. Everyone using GNOME libraries should be more outrageous than everyone using systemd functionality.

[Niarbeht]

systemd fanatic

How, exactly, is a volunteer deciding they want to write systemd units fanatical, but someone who insists that writing and maintaining init scripts for multiple init systems is required is not fanatical? Why are you insisting on two different standards, here?

So, what you want is just an open source Windows. I definitely not agree with this. Even if Windows was open source, I would not use it, and this may actually happen, with the direction MS is taking toward open source. Happily, systemd/Linux is still not Windows. Also, you talk about standards, but systemd has caused more compatibility issue than it has solved. Creating a new "standard" do not solve standards fragmentation issues.

I'm not seeing how you managed to go from "We need a standard and complete set of system APIs" to "you want open-source Windows".

That's... That's a big leap, right there.

[brad0]

Init System Diversity
This is so cringe...

systemd is so super cringe.

[ALRBP]

I don't understand this common criticism of calling systemd "Windows" as if it were some demon to be expelled, it's quite dismissive. No, I don't want Linux to be Windows, nor is systemd anything like Windows. What I want is Linux to have what Windows has: usability. After that, market demand will follow. Systemd is a stepping stone in that direction.
Yes, systemd is "yet another standard", the difference with this one is, it's a standard everyone is actually following. And that's a good thing, it's something to be celebrated. Everyone using GNOME libraries should be more outrageous than everyone using systemd functionality.

You are the one who used the word "Windows" first ; it is not me.
What you are saying is absurd. systemd is only made of non-GUI components, it has nothing to do with end-users usability. For the server market, Linux is already dominating. For home user market, the blocker is compatibility of end user software and pre-installation, this has nothing to do with systemd. Nearly no such software is linked to any systemd component. Standardized software distribution is possible with Flatpak, which is totally independent of systemd (before that, some even used good old bash scripts, yes, like init scripts, but they are definitely not as good as Flatpak). Linux itself is technically ready for domination, the issue is the market itself. The market will definitely not follow just because of internal design changes that nearly nobody knows about.
And saying everyone is following the systemd standard is just wrong. You just can't keep ignoring non-systemd distributions because they do not fit in your systemd world.

[Ironmask]

You are the one who used the word "Windows" first ; it is not me.
What you are saying is absurd. systemd is only made of non-GUI components, it has nothing to do with end-users usability. For the server market, Linux is already dominating. For home user market, the blocker is compatibility of end user software and pre-installation, this has nothing to do with systemd. Nearly no such software is linked to any systemd component. Standardized software distribution is possible with Flatpak, which is totally independent of systemd (before that, some even used good old bash scripts, yes, like init scripts, but they are definitely not as good as Flatpak). Linux itself is technically ready for domination, the issue is the market itself. The market will definitely not follow just because of internal design changes that nearly nobody knows about.

systemd is already encompassing networking and device functionality and will probably further extend from there. That is definitely a standardization end-users can appreciate. Systemd already has the capabilities to essentially replicate the Windows Registry, which, no matter how much criticism that architecture gets, is preferable to a few hundred XML/INI/conf files scattered throughout a labyrinthine filesystem (that was the entire reason the Registry was invented).