Linux: concerns over systemd adoption and Debian's decision to switch
Jay Ashworth
jra at baylink.com
Fri Oct 24 20:25:38 UTC 2014
----- Original Message -----
> From: "Joe Loiacono" <jloiacon at csc.com>
> > The arguments against systemd that I've seen so far:
> >
> > 1) It's different so it's bad.
> > 2) There's a lot of code, there must be some really bad security
> > problems just waiting to happen, so it's bad.
> > 3) It doesn't do things the way we've always done them, so it's bad.
> > 4) The systemd developers are jerks, so it's bad.
>
> Hmmm. It seems that list is missing its most important item.
>
> As an impartial lurker, the primary objection I've seen is:
>
> 1. "Try to do everything" software is not optimal, and will lead to
> heartache.
"Try to do everything *inside PID 1*" is the real problem.
It is a problem because it violates "do one thing and do it well", and
also because it enlarges the privileged attack surface, as made fun
of improperly at 2) above.
3) is not an invalid assertion either, and if you think it is, then
your systems are pretty bone-stock.
4) is not a *direct* impeachment of the code, but developers who can't
work and play well with others are more prone to produce code which also
can't.
The Unix design philosophy wasn't made up out of thin air; it is the
residue of decades of failures, and you flout it at your peril, and
systemd flouts it rather badly.
Cheers,
-- jra
--
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://www.bcp38.info 2000 Land Rover DII
St Petersburg FL USA BCP38: Ask For It By Name! +1 727 647 1274
More information about the NANOG
mailing list