"we need no permission" - duh, it's an FOSS project that nobody has to use. Distro maintainers have chosen to use it because it has a significant number of advantages over SysV init scripts and the rest.
Don't like systemd? Use another init! Use another distro! Use Arch or any one of the distros that makes it very easy to choose your own components. Or... roll your own distro!
People just like to continue beating a dead horse for no reason.
This is why "The Year of the Linux Desktop Is Just a Year Away!". People belly ache about anything modern... and stunt development for their niche requirements.
As I said, you need to dig my comment history for that, because I have no time to rewrite everything I said over and over again.
Distribution politics is something I’m familiar with enough, and I’m well versed enough in UNIX philosophy and how bending it creates problems (which systemd has gone through, and painfully learnt it again).
Another reason that I’m not writing a wall of text is I like horses and not prefer beating them regardless of they are alive or not.
However, if it comes to that, I can roll my own distribution, which I did in a professional capacity before.
Paralel Sys-V (with dependencies), Mudur and upstart was not slower than systemd, and we were already able to do everything systemd was doing. That’s still true.
I’m using systemd on a thousand machines every day, and yes, it has improved, but it’s neither perfect, nor indispensable, yet developers prefer to lock themselves into it.
Lastly, free software moves with respect to each other, not by “moving fast and breaking things because feeling like it”. Systemd became better because they had to accept that they can’t bend everything the way they want. Systemd doesn’t live in a vacuum and doesn’t have the luxury and authority to break and build things as they wish and expect everyone adapt to them and worship them as demigods.
Your sentiment is ridiculous. If Upstart, Mudur or name any of the other many init's were better or became better, distro maintainers would switch. They are some of the most pragmatic people.
Politics did not insert systemd into all of your widely-used distros... merit did. Even if that merit isn't what you wanted for your niche requirements.
systemd came late to the next-gen init debate and ate everyone else's lunch.
Now, the bar is high to replace init's again - meaning the next init system will need to be a leap forward instead of marginal/niche improvements.
> I’m using systemd on a thousand machines every day
You are a niche user and have niche requirements. Understand that...
One of the primary reasons Debian even considered systemd, was because Redhat's control of the Gnome foundation, and thus Gnome's at the time requirement of systemd as an init, created a major conflict in terms of "do we drop gnome or not".
If you want you can read the proposals, the mailing list discussions, the debates, the votes. There's nothing opaque or hidden here. If you haven't done so, then I suggest you do so, instead of disparaging someone who lived through the debates, read the list daily, and partook in the process.
> Systemd was adopted by redhat because it is their baby
You mean after shipping Upstart in RHEL 6 and immediately regretting it?
> And because redhat controls gnome, gnome created a hard depend upon systemd.
This is some revisionist history right there. Red Hat contributes money to GNOME, along with a lot of other organizations. GNOME is not even close to being the only DE either... are you asserting some sort of evil conspiracy to adopt systemd for profit or something? That is absurd...
> ...are you asserting some sort of evil conspiracy to adopt systemd for profit or something?
Why does it have to be an evil conspiracy? Corporations pay folks to work on these things.
Based on my personal experience: Corporations have things like OKRs and performance reviews and the like. Managers in such companies (at all levels) tend to have a hard time understanding what their people work on and whether or not it was actually important and/or useful. This leads to managers getting seduced by things that are easy to fit stories to... Bold Mission Statements and Grand Visions.
One great story is how you're "Revolutionizing Computing and Paving The Way To The Future With Cutting-Edge Technology".
Another great story is how you're "Accelerating the Desktop Through Tighter Integration With Cutting Edge Components".
Absent highly-clued-in managers who have enough slack time to figure out how useful the things you've been doing are, great stories get you more money, accolades and promotions than folks who don't have great stories. Simple as.
No conspiracy needed. It's just a bunch of folks following "incentive gradients" that don't result in optimal outcomes for the unpaid folks who work on and/or use the product.
You mean they don't officially provide support for them, even though you can choose to use a wide variety of init's.
Regardless, there are other distros that don't use systemd by default. Choose one of those if you have it in your head systemd is causing you some sort of pain (it isn't...).
- Mysterious PID1 crash related to timedatectl, resolved by replacing /etc/localtime with a regular file instead of symlink (to a file on same FS). How do you strace PID1 to get some idea what's going on? You don't. Good thing timezone management needs to be part of init!
And those are just the two that I've had to spend hours on in the past few years.
Not using systemd is barely a choice, even if you don't use systemd itself you nearly have to use things which are designed in exactly the same bad ways (polkit, systemd-udevd), there are several (negative) changes which even systemd-resistant distros have adopted for conformity which were first driven by systemd, lots of software has completely removed their provided init-scripts and replaced them with systemd unit files only so you have to write your own or go dig it up from the git history...
Oh and it's extremely brittle and one little piece breaking means you can't boot at all, not even in single user, since everything is jammed into PID1... so good luck even examining what's going on.
I could go on but of course you've already seen and chosen to ignore any argument I could make, because this dead horse has indeed been beaten, by the people who constantly ask "what's wrong with it" when many, many answers are a simple Google away.
It just werks 99% of the time. The few problems it brings are for the most part basically just annoyances for people used to working with something else.
That's exactly the point. It's so complex that it's impossible to track down (and thus fix) the cause of any of its problems. What works for you does not work for someone else, and vice versa, because of some seemingly unrelated difference in how their system was built or configured (usually by their distro, often even as a result of difference between upgrade-vs-fresh install).
As a guy who has been running Unix machines since the early 1990s, including pre-1.0 Linux kernels, SVR4.x , Sony NEWS and Sun machines... I have stopped hating systemd.
Since I no longer install an OS, if it uses systemd :-)
I wonder if 40 years ago people got as bent out of shape about X as they do now about systemd and find one reason or another why it didn't fit the "unix philosophy".
X has long had critics from Unix fans. There’s an entire chapter devoted to X in The UNIX Hater’s Handbook, and the author of that chapter, Don Hopkins, is a major advocate of Sun NeWS. On the Unix side of things, alternatives to X included Sun NeWS and NeXT’s desktop, which are both Display PostScript-based systems. macOS uses Quartz, which is similar to Display Postscript but uses PDF-based objects instead of interpreting PostScript code for rendering graphics. There are also 8 1/2 and its successor rio from Plan 9, which also addressed issues that the Plan 9 developers had with X.
Of course, X won out over all these systems; while X has been MIT-licensed since its early days, none of X’s competitors were open source at the time of their release. NeWS, NeXT’s display system, and Quartz remain closed source, and while Plan 9 was released under an OSI-compatible license in 2002, it wasn’t GPL-compatible; this was rectified in 2014 when Plan 9 was GPL’d, and the licensing became even more liberal in 2021 when it became MIT-licensed.
Quartz isn't really similar to Display PostScript. Vector drawing is done client-side and isn't proxied out to the window server like NeWS and NeXT did. Sure, you can use Quartz to draw to a PDF, but you can do that in Cairo too.
