Mostly what Eoghan said, but with a couple of additions.
So, how insane would people think I am for, if this works, proposing that we ditch upstart?
I think very. In a test VM, sure. Across all of Redbrick's servers? Nuts.
Insane. Completley, batshit, throw that man in a mental institution insane. We might as well just install linux from scratch on our machines. Think about it - it's not even production ready in the opinion of _the 2 distros that came up with it_. The main reasons against upstart seem to be centred around three arguments: * "Wahhh, there's not enough shell scripts for me to hack up anymore" * Its dependency graph seems to be a mess. This is an issue of evolution more than a critical fault with upstart, and I'll bet you a small sum of money that systemd will come with similar head wrecking-isms. * Documentation sucks and it could do with a couple more stanzas for corner cases. Replacing init is getting very, very far into "building a Redbrick Linux" territory. And not just an Ubuntu reskin, a whole separate descendent of Debian. Take a look at the bugs list on Launchpad if you want something to scare you away from doing that :) Having said that, systemd does sound fun to play with and like it has excellent potential in Fedora/OpenSuSE land. I look forward to tinkering with it just as much as you seem to. Just, not on production redbrick machines. -Andrew