News for hackers. Stuff that matters.
Re:Systemd, ten years later: a historical and technical retr... by BiteCode_dev on Monday May 04, 2020 @05:08AM (#23065894)
> SystemD feels like the new grad hire who decided the first thing the would do after joining — the opening power play — is to convince the PHB we should rewrite everything in Haskell.

I've seen this scenario played out, and it always fails. The product never ships, of ships in such a bad shape the users have a terrible user experience.

That's not what happened with systemd. It shipped. And after a few iterations, it actually provided a better user experience.

I don't know about sysadmins, but as a simple end user of Ubuntu, systemd is the best user experience I ever had for what matters to me: setuping my services.

It's quite easy to create your unit files, declare service dependencies, stop, restart, and so on. It's declarative, conf is reasonably straighforward to read an understand, documentation is there, and my system boot is stable on laptop and on my servers.

Maybe for "advanced case X", there is problem. Apparently, a lot of power users say there is. But I also know the FOSS community has some people advocating purity over practicality, don't care much about the people with less skills than them, don't understand user friendliness or will fight for their ego. So I have no way to be sure if the problem is real, or just another free software drama.

But at this point I don't care: for the average Joe like me, it's great, and I'm very happy it has been chosen by Centos and Debian as the default.

Apparently nobody in the entire history of init system have managed to provide something decent for, you know, initializing the system. I never heard a herd of devs claiming that one is the fantastic, and they were all terrible to use when I tried.

Well, I don't care for choice in my init system. I want it to work.

And systemd works fine for me.

Re:Systemd, ten years later: a historical and technical retr... by dijit on Monday May 04, 2020 @08:47PM (#23075119)
> Well, I don't care for choice in my init system. I want it to work.

We Agree.

> And systemd works fine for me.

But it doesn't for me, and I am left with no choice but to suffer with it.

Re:Systemd, ten years later: a historical and technical retr... by tifadg1 on Saturday May 16, 2020 @08:50AM (#23202976)
so what should be done, in your opinion, if it works very well for 51% of people, good enough for 48% and breaks the work flow of 1%?

easiest answer is to say get rid of the inter-dependencies, but that's throwing money and time at a problem affecting 1% of users, at no benefit to the rest. Maybe it's still the right answer, but who's to pick up the slack?