@david_chisnall
> Some are caused by his other projects.
Ah, the good, ol' `systemd-fault-proxy`.
@dzwiedziu @waltman @ryanc @xabean
His project before systemd was pulseaudio. Pipewire has largely undone the damage it did, but it drove a lot of people to FreeBSD before then.
@david_chisnall
I know. At the time PA was something welcomed by probably anyone using consumer audio on Linux, me included.
Narrator: the traffic light controller runs embedded Linux and uses systemd
@ryanc Today I learned how to deal with userland services in a system on sysvinit (or anything init really) because screw all that.
I can still say it's Poettering's fault because guess why I'm having to switch to sysvinit? It wasn't just because I randomly wanted to see what it would be like...
@ryanc I mean, if I had a choice I'd lean towards OpenRC, but right now I need the least effort thing closest to a base distro, which is probably MX Linux which lets you choose sysvinit and has a lot of stuff (like Pipewire-Pulse) fixed to work with it.
But really, the biggest problem is just distros in general need to stop relying on systemd. Like even if they continue to use it, packages shouldn't be built to explicitly call it a dependency, should also support init, etc. That seems basic, but it just doesn't even occur to most to do such a minimal thing. So if one switches from systemd, a lot of stuff uninstalls and/or breaks on most distros.
SystemD gave my cat fleas
@ryanc And PulseAudio ...
I approve of you hobby.
@ryanc When I lived in Redmond I used to pray that I ran into Pottering in a pub/cafe/dark alley.
Alas, it was not to be. Damnit.
@ryanc after overcoming kernel issues I was / will be dealing with net booting RHEL installer and using an NFS root without using an initrd at work.
Weird memory thing on an IBM Power 10 server which still has the 256 MB memory region not able to hold kernel and initrd in memory.
Something about Power 10 changes the way memory is allocated. Seems as if the kernel needs to negotiate more memory. But GRUB doesn’t do that.
IBM’s answer is to change a frame configuration setting. But that requires a full frame outage. And there are production LPARs on this frame.
@ryanc
The problem being you'll be usually right…
<sad trumpet>