Path: utzoo!attcan!uunet!mailrus!ncar!ico!vail!rcd
From: rcd@ico.isc.com (Dick Dunn)
Newsgroups: comp.sys.ibm.pc
Subject: Re: BSD Unix for the PC?
Message-ID: <1989Dec18.180105.1974@ico.isc.com>
Date: 18 Dec 89 18:01:05 GMT
References: <841@crash.cts.com> <28286@amdcad.AMD.COM>
Organization: Interactive Systems Corporation, Boulder, CO
Lines: 47

In article <28286@amdcad.AMD.COM>, phil@diablo.amd.com (Phil Ngai) writes:
[going back a bit...]
> |>Is a BSD version of unix available for the PC (80386)? 
> |We went over this before, there is no such thing.  A lot of System V
> You must not have heard of System V.4.
>...by God, if it's
> not SunOS/BSD, it's good enough for me! 

Well, V.4 is NOT BSD.  It has most of the facilities of BSD systems, but
that's hardly the same.  Here's why I make the distinction:  V.4 attempts
to provide a way to get you all the facilities of V.3, Xenix, and 4.? BSD.
Think just about the kernel level and system interface.  If there are con-
flicting facilities from Sys V and BSD, call them A and B, then there are
several choices:
	- implement both A and B in kernel, provide different ways to get
	  to each
	- implement A in kernel, provide library routine for B which uses
	  kernel call for A and massages results
	- reverse the preceding (implement B, fudge for A)
	- implement something entirely new; implement replacements for both
	  A and B as library routines to use the new facility
I think all of these are used in various situations in V.4.  They all have
implications for performance (space occupied by kernel, time to mess
around with the interface, possibly both).  Now, you *may* be able to make
a lot of SunOS, BSD, or Sys V code work on V.4, although there are
differences which *will* require porting effort (trust me!).  But even
that falls short of the matter of how new code should be written for V.4
specifically--just because an interface happens to exist now (for compati-
bility) doesn't mean you should use it.  V.4 is a different system.  The
compatibility is there, and it's a damned good idea to have it, but it's
still just compatibility, with all the compromise that implies.

>...I don't think it's been ported
> to the PC yet, but there's certainly a 386 version...
...
> Oh, don't worry, I'd bet any sum of money that it will be ported to
> the PC. You can get *anything* on a PC.

It's going to be interesting to see how a kernel > 1 Mb with virtual memory
gets ported to a PC!  (Note that Phil is making a distinction between 386
and PC.)  A gambling sort of person ought to take that bet.  In the PC
world, nothing short of a 386 has enough horsepower or reasonable
addressing capability to make V.4 a believable proposition...I wouldn't
even try it on an AT (i.e., 286).
-- 
Dick Dunn     rcd@ico.isc.com    uucp: {ncar,nbires}!ico!rcd     (303)449-2870
   ...Never offend with style when you can offend with substance.