The course +Debian
is taking (i.e. that, which could lead it to ending up with upstart) is really worrying.
- Not only that two Canonical employees are members of the tech-ctte... and I do not believe that either of them would act in bad faith but at least this will have the effect that if
upstart is chosen, people will start questioning that.
- Also, I personally, would really love to see *buntu’s (IMHO rather bad than positive) influence on Debian shrinking, instead of growing even more.
- Not everything with systemd is fine (yet),… but upstart really seems to have a broken core design.
And in general, I’d consider system in basically every technical way superior over upstart.
It seems that if Debian decides for upstart, it may end up on a very lonely island together with Canonical, when basically all other major distros and many upstream projects go for systemd.
Hopefully they will just decide on a default init system and not outlaw any others. IMHO open source is about the freedom of choice and users (which doesn’t mean that the developers are the user’s slaves, but it means that the mustn’t intentionally restrict their freedom).
This in turn is the downside of systemd.
It is (or is going to be) quite strongly tied to many other projects - not necessarily always technically, but in terms of people, politically, organisationally.
There seem to be quite close ties to the Linux kernel, D-Bus, GNOME, KDE may follow, udev is already in the repo, etc. pp..
While this has sometimes obvious benefits, it also has long term drawbacks.
As these projects are more and more integrated, one gets a more and more “monolithic” system landscape.
Till know, when someone didn’t like a component it was easy to fork and replace it - in the future, when all this is more and more tied together it will simply no longer be possible to easily get rid of something.
In a way this is what we‘ve always had with Windows/Mac OS ... one big (organisationally) monolithic system - absolutely contradicting the ideas and principles of FLOSS.
This development cannot only be seen at the low system level,... look at desktop environments, +GNOME
most notably, where we have more or less the same.
Anyway… a major drawback, which seems to quite influence the decision of the tech-ctte, is it’s non-portability.
Sure, if you want to make a powerful init system, you’ll want to use all of Linux’ new features... and from that POV the decision was surely right.
On the other hand, having the BSDs and others around is IMHO very important for the FLOSS eco-system - not only in terms of improving code quality and finding bugs, but also in terms of diversity.