4 ms·

I really sympathize with people who try to avoid systemd but I think it is a futile exercise. Every piece of software these days assume you are running systemd

I really sympathize with people who try to avoid systemd but I think it is a futile exercise. Every piece of software these days assume you are running systemd and make your life miserable when you try not rely on systemd, and sometimes, it is even not possible to run without systemd at all. Unfortunately, people don't remember how was the systemd adoption and why it was so controversial. systemd was promoted as a init system, but in reality, it is more than that, it is a suite of software; we have journald, firewalld, udevd, logind, networkd, resolved and many more. People forgot how systemd was forced to everybody when Gnome made systemd-logind a hard dependency and logind was extracted to elogind so people could initialize their desktop without using systemd. The same happened with udev, forcing Gentoo to fork it into eudev to avoid systemd dependency. People complaint about portability and compatibility because you cannot interact with systemd-* without systemd; you are coupled. The problem is not systemd as a software, is the attempt to hijack the entire ecosystem, replacing existing services with services dependents of systemd, forcing systemd adoption. And the question I ask to you is: who is behind this?

kaba0 4y ago
You forget the tiny bit of detail how the absolute majority of the linux world voted in favor of systemd, often literally. Debian for example had multiple votes (in a system much more democratic than what we are used to in politics) on the issue where systemd won hands down in each case. Arch got to the same conclusion independently. You know, the actual people that have to maintain packages, so they actually know what does it entail. So you can’t even claim that they don’t know what they voted on. So if you call the democratic absolute majority’s choice hijacking, I have to believe you are the one doing the pushing.
viraptor 4y ago
> Every piece of software these days assume you are running systemd and make your life miserable when you try not rely on systemd This is silly. Apart from desktop environments, barely anything cares about what session manager is being used. There's no connection between almost every software these days and the init system.
eternityforest 4y ago
I'm not sure if anyone is truly behind it. Systemd is open source. The problem with FOSS is people make stuff they want, then they share it. Others only complain if it affects them. And if you have no interest in alternate init systems, then compatibility with anything else sometimes just means stretching your team thinner. From my perspective, and I'm sure some others, the hijacking is a benefit, it standardizes all the systems. If you're a corporate IT person, or a home user who wants to not have to think about their OS, this is great. It means you won't run into incompatible stuff, there's one standard "Linux platform". Systemd isn't a DRM, it doesn't cost subscription fees, so I don't see why you would push the technology other than really believing that it is best for your dev team and your users. From the point of view of the companies doing this stuff, Gentoo and Void and Devuan don't exist. They are like forks of the entire Linux platform. There may be a bit of corporate competition happening, I'm sure Canonical doesn't want too much variety, but it's hard to separate it from the real benefits it has for non-hobbyist users to have less fragmentation. There may be some dirty political stuff happening to push it with dependencies, but most of the community seems to be behind it. Perhaps this should be seen as a tyranny of the masses kind of situation, and the better question is what obligations do the rest of us have for the other systems we will likely never use. It does seem like there is some minimal level of effort that should be made, in the name of inclusiveness, and compatibility with some possible future QuantumHyperSystemD thing, at least to ask yourself and sure that it's all good faith and you aren't actively sabotaging anything. But beyond that, unless you're being paid, not many are going to support a system that isn't relevant to them.
- d0mine 4y ago
  Isn't systemd a Red Hat thing (IBM now)? How much influence does it have?
  - baisq 4y ago
    Complete influence: they have shoved systemd down our throats by making GNOME depend on it.
    - 4y ago
_-david-_ 4y ago
>systemd was promoted as a init system, but in reality, it is more than that, it is a suite of software; we have journald, firewalld, udevd, logind, networkd, resolved and many more. Several of the things you mentioned are completely optional. You can use some of the systemd programs but not others. Many distros don't used networkd or resolved for example. GNU was promoted as an editor. Just because an organization or project expands into multiple different projects doesn't mean that is a bad thing. Does it matter to you that GNU makes 100 different things? >People forgot how systemd was forced to everybody when Gnome made systemd-logind a hard dependency This is not an issue with systemd, but Gnome and whoever else. You should complain to them, not the systemd folks. It doesn't make systemd bad or not a useful piece of software. Gnome only has a limited amount of resources (both developers and money). You can't really expect them to support an alternative. I'm sure if you talked with them they would support an alternative to logind if you were willing to provide enough money. If gnome won't support it then you could pay Gentoo developers and I'm sure they would create additional alternatives like you mentioned with eudev. Ultimately I think unification of Linux distros is good. It is unfortunate for BSD and whoever else, but when it comes to Linux this is a good thing. I am currently helping a coworker migrate to Linux and trying to explain apt vs yum/dnf vs pacman vs whatver, rpms vs debs vs snaps vs flatpak vs appimage, etc is just tiresome. When they look up how to do something on Linux and it just doesn't work it is a bad experience. Maybe systemd isn't the correct choice, but unifying on one init, network manager, firewall just makes a better experience, especially for new people.
- mid-kid 4y ago
  > GNU was promoted as an editor. Just because an organization or project expands into multiple different projects doesn't mean that is a bad thing. Does it matter to you that GNU makes 100 different things? You're comparing an organization consisting of projects headed by different people with very few strict inter-dependencies and whose primary goal was being compatible with as many Unix systems as possible at the time, to be able to slowly start replacing their utilities, to a single umbrella project headed by the same people and parent company, whose utilities are deeply interwoven (can't run a systemd tool, say, resolved, without systemd) and are expressly incompatible with the ecosystem they're trying to replace.

kaba0 4y ago

You forget the tiny bit of detail how the absolute majority of the linux world voted in favor of systemd, often literally. Debian for example had multiple votes (in a system much more democratic than what we are used to in politics) on the issue where systemd won hands down in each case. Arch got to the same conclusion independently. You know, the actual people that have to maintain packages, so they actually know what does it entail. So you can’t even claim that they don’t know what they voted on. So if you call the democratic absolute majority’s choice hijacking, I have to believe you are the one doing the pushing.

viraptor 4y ago

> Every piece of software these days assume you are running systemd and make your life miserable when you try not rely on systemd This is silly. Apart from desktop environments, barely anything cares about what session manager is being used. There's no connection between almost every software these days and the init system.

eternityforest 4y ago

I'm not sure if anyone is truly behind it. Systemd is open source. The problem with FOSS is people make stuff they want, then they share it. Others only complain if it affects them. And if you have no interest in alternate init systems, then compatibility with anything else sometimes just means stretching your team thinner. From my perspective, and I'm sure some others, the hijacking is a benefit, it standardizes all the systems. If you're a corporate IT person, or a home user who wants to not have to think about their OS, this is great. It means you won't run into incompatible stuff, there's one standard "Linux platform". Systemd isn't a DRM, it doesn't cost subscription fees, so I don't see why you would push the technology other than really believing that it is best for your dev team and your users. From the point of view of the companies doing this stuff, Gentoo and Void and Devuan don't exist. They are like forks of the entire Linux platform. There may be a bit of corporate competition happening, I'm sure Canonical doesn't want too much variety, but it's hard to separate it from the real benefits it has for non-hobbyist users to have less fragmentation. There may be some dirty political stuff happening to push it with dependencies, but most of the community seems to be behind it. Perhaps this should be seen as a tyranny of the masses kind of situation, and the better question is what obligations do the rest of us have for the other systems we will likely never use. It does seem like there is some minimal level of effort that should be made, in the name of inclusiveness, and compatibility with some possible future QuantumHyperSystemD thing, at least to ask yourself and sure that it's all good faith and you aren't actively sabotaging anything. But beyond that, unless you're being paid, not many are going to support a system that isn't relevant to them.

d0mine 4y ago
Isn't systemd a Red Hat thing (IBM now)? How much influence does it have?
- baisq 4y ago
  Complete influence: they have shoved systemd down our throats by making GNOME depend on it.
  - 4y ago

_-david-_ 4y ago

>systemd was promoted as a init system, but in reality, it is more than that, it is a suite of software; we have journald, firewalld, udevd, logind, networkd, resolved and many more. Several of the things you mentioned are completely optional. You can use some of the systemd programs but not others. Many distros don't used networkd or resolved for example. GNU was promoted as an editor. Just because an organization or project expands into multiple different projects doesn't mean that is a bad thing. Does it matter to you that GNU makes 100 different things? >People forgot how systemd was forced to everybody when Gnome made systemd-logind a hard dependency This is not an issue with systemd, but Gnome and whoever else. You should complain to them, not the systemd folks. It doesn't make systemd bad or not a useful piece of software. Gnome only has a limited amount of resources (both developers and money). You can't really expect them to support an alternative. I'm sure if you talked with them they would support an alternative to logind if you were willing to provide enough money. If gnome won't support it then you could pay Gentoo developers and I'm sure they would create additional alternatives like you mentioned with eudev. Ultimately I think unification of Linux distros is good. It is unfortunate for BSD and whoever else, but when it comes to Linux this is a good thing. I am currently helping a coworker migrate to Linux and trying to explain apt vs yum/dnf vs pacman vs whatver, rpms vs debs vs snaps vs flatpak vs appimage, etc is just tiresome. When they look up how to do something on Linux and it just doesn't work it is a bad experience. Maybe systemd isn't the correct choice, but unifying on one init, network manager, firewall just makes a better experience, especially for new people.

mid-kid 4y ago
> GNU was promoted as an editor. Just because an organization or project expands into multiple different projects doesn't mean that is a bad thing. Does it matter to you that GNU makes 100 different things? You're comparing an organization consisting of projects headed by different people with very few strict inter-dependencies and whose primary goal was being compatible with as many Unix systems as possible at the time, to be able to slowly start replacing their utilities, to a single umbrella project headed by the same people and parent company, whose utilities are deeply interwoven (can't run a systemd tool, say, resolved, without systemd) and are expressly incompatible with the ecosystem they're trying to replace.