[theclaw]

I don’t have a problem with this. I am seeing systemd as a system management layer and as such it makes sense that it should manage as much of the moment-to-moment operation of the system as possible. It seems like a lot of people are upset because it’s a new thing to learn that’s being forced on them by their distro.

I used to feel the same way. I’m only a casual linux user and I didn’t want to climb this particular mountain but having learned just a little about it, created some of my own units and played with the tools the sense of alienation dissipated and I was fairly happy with it.

[mondoshawan]

I suspect most of us are upset not at learning a new thing, but dealing with brand new bugs in new unexpected ways. Learning a new thing has never really been a problem -- fighting instability in new software has always been the issue.

Speaking as a distribution developer, I can say that systemd has bitten me in multiple ways due to upstream bugs that are either ignored or left to rot for quite a while. I've had to resort to patching them, or working around them, which makes things very difficult to live with, especially given the pace of systemd development and feature additions.

The older software had a number of bugs, yes, but they were well known and battle tested. In short, it's not "I don't want to learn" -- it's "the new shiny has a slew of unknown bugs we can't easily identify and fix".

[edit: removing bits of incendiary text]

[castillar76]

Very much agree. I'd add that the reaction of the systemd developers to security and functionality bug reports has been...decidedly mixed, which is enough to make me very nervous.

For me personally, it's also that systemd keeps expanding and expanding, which both makes more things dependent on their security profile and reduces choice in the ecosystem. The number of different boot/init systems I might need to use is pretty limited and I might be OK with just one (especially since systemd /is/ really good at it). The number of DHCP clients, firewall systems, DNS clients, and so forth is decidedly not so limited, and results in a nasty choice of "use the systemd thing and hope it supports all the features you need" or "rip out the systemd thing and replace it and hope that works and continues to work" or "install something that sits on top of the systemd thing and thus has to interact with it in strange ways that may bite you unpredictably". Each of those is a recipe for a lot of Maalox moments, and it feels like the response to that concern is, "Aww, you're just being a worry-wart."

[eikenberry]

Older software should always have fewer bugs simply due to maturity. Ie. progress isn't free. The best way to mitigate that is where distros like Debian stable come in. They help gate these developments, releasing more well tested versions of the software and fixing any critical issues that come up with it over the time of the release.