Subject: Re: Precompiled vax packages anyone?
To: None <port-vax@NetBSD.ORG>
From: Michael Sokolov <sokolov@alpha.CES.CWRU.Edu>
List: port-vax
Date: 02/22/1998 13:30:25
Dear Ty Sarna,
In my posting about the differences between traditional Berkeley UNIX(R)
and "modern BSD" (archived at http://sclwww.scl.cwru.edu/~mxs46/bsd10.txt)
I have forgotten to make one point and to ask one question.
The point that I have forgotten to make is that 4.3BSD vintage OSes have
one additional advantage over 4.4BSD vintage ones on platforms where the
hardware vendor supplies a BSDish commercial OS. These OSes are almost
always 4.3BSD vintage. OSes that are of the same vintage as the hardware
vendor-supported one always have some advantages. Developing non-profit
software is often difficult. There are problems with documentation,
hardware resources, and just plain time and effort requirements. It's
always nice to be able to borrow something from the vendor-supplied OS. For
instance, the authors of the sparc port in 4.4BSD didn't have the time,
resources, documentation, or whatever to write a standalone system.
4.4BSD/sparc kernels are loaded by the SunOS boot program. Guess what, the
filesystem format in SunOS is 4.3BSD vintage. In 4.4BSD the native
filesystem format is different enough to be incompatible. The result is
that using the native format for the root filesystem makes the system non-
bootable. 4.4BSD supports 4.3BSD filesystem compatibility, but it's awkward
to use it on a permanent basis.
Although the documentation for the pmax port in 4.4BSD silently skips
this issue, I'm sure that the situation is the same. On PMAXen the boot
program is cast in stone in the system ROM. Of course, it's designed for
Ultrix. I'm sure that the latter has 4.3BSD vintage filesystems. It
supported PMAXen at least as early as v3.0 which much predates the new
filesystem format. Since the system ROM is not tied to any particular
version of Ultrix, the filesystem format has to be the same in all
versions. Therefore, the pmax port of 4.4BSD is probably hit by the same
problem as the sparc port.
Also having an OS look close to, say, SunOS for the user is sometimes
beneficial. SunOS is one of the most widespread Unices on our campus. The
CES department's UNIX "cluster" (which I currently use for my E-mail) is a
SunOS/Solaris mixture. alpha.ces.cwru.edu (the mail server where my mailbox
lives) is a SunOS box, omega.ces.cwru.edu (the disk server where my home
directory lives) is a SunOS box, and there are both SunOS and Solaris boxes
for people to log into (I use the SunOS ones). Therefore, if my system
looks close to SunOS, users are likely to be comfortable with it.
The question that I have forgotten to ask you is whether your statements
about 4.4BSD are based on 4.4BSD-Encumbered, 4.4BSD-Lite1, or 4.4BSD-Lite2.
In case you have the same question for me, my statements are based on the
4.4BSD documentation which corresponds to 4.4BSD-Encumbered.
Sincerely,
Michael Sokolov
Phone: 440-449-0299
ARPA Internet SMTP mail: sokolov@alpha.ces.cwru.edu