Subject: Re: A report on implementing runlevels in NetBSD
To: NetBSD Userlevel Technical Discussion List <tech-userlevel@NetBSD.ORG>
From: Greywolf <>
List: tech-userlevel
Date: 12/04/1999 20:33:37
On Sat, 4 Dec 1999, Greg A. Woods wrote:

# As I recall Motorola UNIX and Amiga UNIX are both very faithful ports
# that were not official reference ports (Motorola used GCC instead of
# AT&T's compiler, for example, but otherwise stayed true to the spirit).

Actually, Motorola started out using the Green Hills compiler -- they
stayed away from GCC as long as possible due to the inclarity of the
licensing stickiness at the time.

# Even Pyramid's DC-OS/x was a reasonable facsimile of the original (and
# was certainly geared to being a true mainframe class production system,
# true to the "Data Center" part of its name).

Yes, but OS/x at least let you have a choice between which universe
you wanted to live in.  [Guess which one we picked.]

# What are not good examples are those concoctions made up of various
# generations of SysV releases all munged into one, often with lots of
# more native BSD stuff thrown in on top.  SCO, the various incarnations
# of AIX, what I know of HP/UX, DG/UX, etc. are such examples.

Atrocities, all of them.  About as effective as trying to glue together
two sheets of Teflon.

# SunOS-5
# started out being what AT&T took on as SysVr4, but as I understand the
# history the political pressure on AT&T to divorce itself from Sun early
# on caused the Sun folks to continue on their own, not wanting to restart
# with the Fujitsu reference port when all the dust settled...

That settling took a full two+ years to happen; it's a good thing Sun
didn't wait.

Personally, I think Sun ought not have been bedded by AT&T in the first
place.  Solaris 2.0 was a fscking disaster.

# Obviously a
# lot of independent development has happened to SunOS-5 since its first
# release too, both on the outside and the inside.

...and Solaris is actually turning out to be a half-decent platform on
which to work.

On a tangent:  I'm tired of seeing these desktop-in-a-window suites.  What
we need on that front are separate programs which can then be integrated
in any way desired.

Unfortunately, to complete this, we'd need the atrocity that would be
DND in the kernel (and we really don't want to go there).

NetBSD: the cathedral versus the bizarre.