A lot of the arguments about systemd are annoying because they’re either based on things that are simply inaccurate (like many of the jabs about stuffing things

A lot of the arguments about systemd are annoying because they’re either based on things that are simply inaccurate (like many of the jabs about stuffing things into PID 1, done by people who don’t realize systemd-init is just one component,) are based on a super luddite mindset (I really don’t think shitty Bourne shell sysvinit scripts are better than systemd units, TYVM,) or hyperfocus on relatively small points of contention (Usually Poettering’s attitude, which does suck sometimes. Sometimes more specific things like the UID fail-open bug. Sometimes valid but contentious design choices like binary log storage.) Granted. There is plenty of valid criticism against systemd, and when it was first pushed on all of us I was among the angsty annoyed people. However, I also don’t care that much about init systems, so if it was gonna work it was gonna be fine for me. Systemd today works quite well. It’s a more complete replacement for many of the components it replaces. It has a handful of annoying behaviors, but it also does a lot of things quite well. I really like socket-activated units. User systemd works OK for me too. systemd-networkd is pretty good for systems that don’t need NetworkManager. I initially had some issues with logind, but today it does roughly what I want it to. (Desktop Linux is the only OS that gives me the actual behavior I want with sleep on lid close: it does it only when the last external display is disconnected. I have written software to try to get similar behavior in Windows.) My remaining complaints: - I dislike the way systemd is pushing the world towards dbus even more. Not every UNIX like ships dbus by default and not everyone wants it. It is flawed in some ways and it is terrible that it is being imposed as a boundary for all kinds of system services in a way that will take ages to break free from. - I dislike that instead of building more generalized interfaces or working into existing frameworks, systemd’s interface is quickly becoming a major part of the desktop Linux ABI. Major projects increasingly depend on it, working poorly or not at all without it. - Generally just RedHat not caring about the former concerns as a whole makes me not like their control over the ecosystem. I’m glad somebody takes the mantle of trying to push desktop Linux forward, but I really have a strong dislike towards RedHat in general, and things that are perceived as highly influenced by RedHat. GNOME’s CSD initiative is the worst thing that has ever happened to desktop Linux and libdecor is an ongoing failure of a solution years later. No offense to the folks who painstakingly work to make it happen, but I’m not sorry, it’s crap. My opinion hasn’t changed even slightly since day one. (Though almost all Wayland compositors outside of GNOME’s support server side decorations, so you could always give GNOME users decorations that let them know how you feel ;)