Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!uunet!mcvax!hp4nl!botter!star.cs.vu.nl!ast
From: ast@cs.vu.nl (Andy Tanenbaum)
Newsgroups: comp.os.minix
Subject: Re: Re**n:  Future of Minix
Message-ID: <2883@ast.cs.vu.nl>
Date: 15 Jul 89 10:11:56 GMT
References: <194@loligo.cc.fsu.edu>
Reply-To: ast@cs.vu.nl (Andy Tanenbaum)
Organization: VU Informatica, Amsterdam
Lines: 44

In article <194@loligo.cc.fsu.edu> nall@nu.cs.fsu.edu (John Nall) writes:
>I'm not at all sure that I'm going to be able to talk my administration into
>swapping out the pc's in the lab for at's, but then I don't have to throw
>away the version 1.3 that I have, either.
Misconception.  V2.0 will absolutely, definitely, unequivocably run on 4.77 MHz
8088-based PCs.  It's just that to recompile the system, I will probably
assume the presence of a hard disk.  You will be able to run the system
with an 8088, 640K and one floppy.  Bruce did a good job of maintaining
compatibility with the 8088.  He and I have discussed some minor changes,
such as dynamically determining the CPU and system type when the system
starts (partly based on the character you type in, and partly by dynamic
tests).  If it discovers it is on an 8088, it will just set a variable to
88 and continue.  There are only 6 or 7 places where that matters incidently.
Bruce did a really clean job of keeping the 286 and 8088 separate.  Look
at the code.

>A couple of comments, however.  Bruce points out that the Amoeba networking
>code does NOT run on Minix286.  Will official Minix286 have some form of
>networking?  
Hopefully yes.  The problems are (1) some bugs, and (2) need to change the
driver to run in protected mode.  I'll probably do this, as we have a lot
of Ethernet machines at the VU.


>Will Minix286 support arbitrarily large disk farms?
I think most PCs are limited to two drives.  As far as I can tell, there is
nothing in MINIX that is different on big drives than on little drives.
I guess the real limit is the use of 16-bit numbers for block numbers.
That could be changed, or one could use, say, 4K blocks, allowing up to
256M per partition.

>Of course, by making such a decision, Andy is conciously saying that Minix
>is going to be a real operating system, instead of a teaching tool.  
No.  I think I may not list everything in the book, just the parts relevant
to teaching.  People who want to see the rest will have to print their own
listings.

>On the textbook, by the way, let me suggest that the source code listing,
>user guide, implementor's guide, etc., be issued as a separate, optional,
>document.  
It makes it more expensive, so we have been hesitant to do this.  I have a
new editor at P-H now, so maybe we will examine this question again.

Andy Tanenbaum (ast@cs.vu.nl)