Path: utzoo!attcan!uunet!mcsun!hp4nl!star.cs.vu.nl!ast
From: ast@cs.vu.nl (Andy Tanenbaum)
Newsgroups: comp.os.minix
Subject: Re: New MINIX available (UNIX clone with full source code)
Message-ID: <7581@star.cs.vu.nl>
Date: 17 Sep 90 13:31:01 GMT
References: <90257.232245UH2@psuvm.psu.edu> <7571@star.cs.vu.nl> <1990Sep15.182737.22534@lavaca.uh.edu>
Sender: news@cs.vu.nl
Organization: Fac. Wiskunde & Informatica, VU, Amsterdam
Lines: 24

In article <1990Sep15.182737.22534@lavaca.uh.edu> jet@karazm.math.uh.edu (J. Eric Townsend) writes:
>This will not be true in the very near future.  The next major version
>of Mach will be free of AT&T code, thereby making it publicly available
>to those of us w/o > $100K to spend on a UNIX source license (what
>a Sys V license costs last time I priced them).

This is true, although one can have a nice discussion about "very near future".
I was at an OSF meeting in Grenoble recently, and we were treated to several
talks about the future of Mach and OSF/2.  The current version, 2.5, has
most of Berkeley UNIX inside the kernel.  The next release of the kernel, 3.0,
which is indeed almost ready, has no Berkeley code in the kernel.  Instead,
all of it resides in the top of the user space.  When you issue a system
call trap, the kernel vectors the trap to high core, where good ole Berkeley
takes over and carries it out.  The step beyond this is to rewrite the
Berkeley UNIX part as a bunch of separate servers.  Then the task of rewriting
all the compilers, libraries, utilities, etc. must be dealt with.  No doubt
all this can be done.  Informally, I was told to expect it in the 1992/1993
time frame, if no slip-ups happen.  Since I am an ex-astrophysicist, I agree
that 2 or 3 years is nothing.  For an astrophysicist, 100 million years is
like the blink of an eye.

Also, I'd count on having an 8M 80386 as the bare minimum.

Andy Tanenbaum (ast@cs.vu.nl)