I don’t think that’s true at all. I’m ok with systemd, but I don’t really like it, and find much of the criticism valid. At this point the reason I use it, and am more-or-less fine with it, is that it has become the de facto standard and is very well supported.
Which is also one of the reasons I dislike it — it is such an integral part of modern Linux systems that it can be hard to change, which reduces a lot of the appeal of Linux — flexibility and freedom.
I’m fine with systemd, but I really do get the feeling that it exists more because of Red Hat’s NIH syndrome than anything that was actually bad about upstart.
Nah, upstart actually sucked. I used it in an embedded project and getting things to start consistently was a nightmare. We had to put all kinds of sleeps in because there was no way to tell when something actually started, only when it was told to start (i.e. start on started x would start both at essentially the same time).
With systemd, that all went away. It magics away sockets and whatnot so things just work properly. Also, our startup time went way down because things could start just a bit earlier, and the config files were more intuitive.
Upstart was a poor solution in search of a problem, and items sysvinit was honestly better imo because sysvinit didn’t hide little gotchas all over the place. Systemd is an over engineered solution to a real problem, but it works really well. Oh, and socket activation is magical.
That said, I still prefer the FreeBSD way, which is just a slightly fancy sysvinit. It works well, though it won’t win any awards for fanciness. Maybe launchd, if it ever comes to FreeBSD (maybe it has? I’ve been OOTL since 12), will be cool, IDK.
I don’t think that’s true at all. I’m ok with systemd, but I don’t really like it, and find much of the criticism valid. At this point the reason I use it, and am more-or-less fine with it, is that it has become the de facto standard and is very well supported.
Which is also one of the reasons I dislike it — it is such an integral part of modern Linux systems that it can be hard to change, which reduces a lot of the appeal of Linux — flexibility and freedom.
I’m fine with systemd, but I really do get the feeling that it exists more because of Red Hat’s NIH syndrome than anything that was actually bad about upstart.
Nah, upstart actually sucked. I used it in an embedded project and getting things to start consistently was a nightmare. We had to put all kinds of sleeps in because there was no way to tell when something actually started, only when it was told to start (i.e.
start on started xwould start both at essentially the same time).With systemd, that all went away. It magics away sockets and whatnot so things just work properly. Also, our startup time went way down because things could start just a bit earlier, and the config files were more intuitive.
Upstart was a poor solution in search of a problem, and items sysvinit was honestly better imo because sysvinit didn’t hide little gotchas all over the place. Systemd is an over engineered solution to a real problem, but it works really well. Oh, and socket activation is magical.
That said, I still prefer the FreeBSD way, which is just a slightly fancy sysvinit. It works well, though it won’t win any awards for fanciness. Maybe launchd, if it ever comes to FreeBSD (maybe it has? I’ve been OOTL since 12), will be cool, IDK.
On the other hand, fragmentation makes software hard to support all of them. It seems like a dilemma.