[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: On init in Debian

]] Svante Signell 

> On Wed, 2012-03-21 at 12:11 +0100, Tollef Fog Heen wrote:
> > ]] Svante Signell 
> > > And how do you expect non-experts be able to solve problems when they
> > > pop up. Buying consultant services from the experts?
> > 
> > I don't expect non-experts to be able to solve problems with init
> > scripts any more or less than I expect them to be able to solve problems
> > with systemd units, so I fail to see why this question is relevant.
> As mentioned before set -x can be used and somebody knowledgeable about
> the output can help the person having problems. This is just an example
> of possible debugging for a non-expert.

Yes, and systemd --test is useful to debug systemd problems.  I don't
know what point you are trying to make?

> Can you explain what features are so much better with systemd/upstart
> (event-based) compared to sysvinit? If you are interested in faster boot
> times, then coreboot would help much compared to commercial BIOSes,
> regardless of the init system.

I have no idea why you are constructing a straw-man which I don't think
I've ever argued for, and then propose a different solution to it (and
to boot, one that does not generally work on most systems).

As for a few features that are better: services are properly contained
and are actually managed rather than relying on the assumption that
since you've run a shell script that happened to exit 0, that service is
up and working.

Related, services are properly contained and there's no way for the
environment of the user who is starting the service to leak into the
service's environment.

> Regarding who is expert or not, can the people who considers themselves
> as such (others shouldn't bother) do a _scientific_ comparison of the
> three alternatives with respect to important features. First step would
> be to write down which aspects are important, and continue from there.
> Also, practical experiments are needed to verify statements made.

I'd rather work on making systemd better and a better init system for
Debian than to satisfy some academic desire for a comparison of init

Tollef Fog Heen
UNIX is user friendly, it's just picky about who its friends are

Reply to: