Newsgroups: comp.os.minix
Path: sparky!uunet!europa.asd.contel.com!darwin.sura.net!udel!minix
From: Marcelo Pazzini <CEC3MPZ%BRUFSC.bit...@uicvm.uic.edu>
Subject: Matters in Minix-L
Message-ID: <1992Jun4.041636.14712@udel.edu>
Originator: m...@nigel.ee.udel.edu
Sender: use...@udel.edu (USENET News Service)
Nntp-Posting-Host: nigel.ee.udel.edu
Organization: University of Delaware
Date: Thu, 4 Jun 1992 03:22:30 GMT
Approved: usenet=re...@louie.udel.EDU
Lines: 18
I feel something weird happening to this newsgoup. While bozos as me
keep writing foolish and asking the same old FAQs, the old experienced
writers as Bruce Evans, Mike Irons, Christoph van Wuellen, Franz
Meulenbroeks (sorry for typos :-) Fred (terrible) van Kempen, Glen
Overby, Peter Holzer, Wim van Dorst, (and many others I forgot)
got all them quiet. (Ah, I forgot Andy..., but he were propositedly
always quiet :-)
Isn't there no more worth themes to be discussed here? Is it to be
left for the bozos? Don't tell me it's a bandwidth saving process...
Marcelo Pazzini Bitnet: CEC3MPZ@BRUFSC
Internet: pazz...@vortex.ufrgs.br
CTC-DCEC-CPGCC
Uni Fed Santa Catarina pa-pa-pa, pa-pa-paba, pa-pa-pa, pa, ba
Florianopolis, SC, BRAZIL (from Smoke on the Water, Deep Purple)
--
Mail System (MMDF)
Newsgroups: comp.os.minix
Path: sparky!uunet!darwin.sura.net!mips!pacbell.com!network.ucsd.edu!
celit!equalizer!timbuk.cray.com!hemlock.cray.com!overby
From: ove...@cray.com (Glen Overby)
Subject: Re: Matters in Minix-L
Message-ID: <1992Jun5.190531.11919@hemlock.cray.com>
Summary: The world is a changing!
Lines: 63
Organization: /dev/null, that's where I am
References: <1992Jun5.160822.26463@udel.edu>
Date: 5 Jun 92 19:05:31 CDT
BURGESS%HRLEIS.dec...@hqhsd.brooks.af.mil (Dave Burgess) writes:
>Marcelo Pazzini Bitnet: CEC3MPZ@BRUFSC writes:
>>I feel something weird happening to this newsgoup. While bozos as me
Well, MINIX now has, shall we say, competition.
I think a lot of people who were active in doing things with Minix have moved
on; some completely while others (like myself) are still reading the Minix
world but writing elsewhere.
First the was Coherent. Remember how it was supposed to be (or at least
HOPED to be) so much better than Minix? After all, it was a commercial
product and would surely have better support. Binary only.
I found going from a source-available environment (4.3BSD on a VAX) to
binary-only (SunOS) to be simply frustrating. I definately didn't jump the
boat for Coherent.
Next came Linux and it's missionaries (they won't leave you alone until
you agree with them that Linux is the greatest operating system ever created;
I find this behavior both on the net and in person).
Linux seemed to promise to be the Minix without Andy's hesitations about adding
all kinds of bells and whistles. I think it's good that there's a place for
the bells and whistles to be put :-)
Linux seems to have taken most of the MINIX people who really didn't want to
play with Unix (internals or user level) but just wanted to run Unix on their
own machine (or, rather, have Unix run itself on their own machine). I'm sure
it also grabbed a few good hackers from Minix.
Then there's 386BSD (as well as BSDI/386) for the hardcore. It might boot on
your machine, and it definately will take over like a virus. It demands the
extreme sacrifice: your WHOLE hard drive. No DOS. No half-way commitment:
either say yes or no.
This happens to be where I've spent much of the past several months' worth of
spare time (doesn't run on my hardware yet, but I'm having fun getting there).
Lastly, there's the recently announced MACH + Unix server (based on 386BSD).
That'll thin the crowd out some more.
If anyone tries mentioning GNU HURD, they should start with the FTP site we
can all retrieve a usable copy of it from :-)
Yes, I think it has hurt Minix to have competition for bright minds and busy
fingers. It also hurt AT&T to have competition, but it also gave the US a
digital telephone system a LOT sooner than if we'd have waited for AT&T. I
remember Andy saying something to the effect that 'if Minix gets replaced, so
be it'. Minix has already made it's mark on the Unix world; that can't be
taken back. Overshadowed, maybe.
I'm not sure it's all bad. Minix's roots were as a SIMPLE OS that could be
taught and understood in an undergraduate OS class (or at least in a grad OS
class). If you know where your market is and address it well, you'll do OK.
You can't be everything to everybody.
As always, the vaccum created has opened up a place for a new set of Minix
gurus.
Minix is far from dead.
Glen Overby <ove...@plains.nodak.edu> (among others)
Path: sparky!uunet!mcsun!uknet!icdoc!puffin.doc.ic.ac.uk!ajt
From: a...@puffin.doc.ic.ac.uk (Tony Travis)
Newsgroups: comp.os.minix
Subject: Re: Matters in Minix-L
Message-ID: <1992Jun6.231057.23959@doc.ic.ac.uk>
Date: 6 Jun 92 23:10:57 GMT
References: <1992Jun5.190531.11919@hemlock.cray.com>
Sender: use...@doc.ic.ac.uk
Organization: Department of Computing, Imperial College, University of London, UK.
Lines: 38
Nntp-Posting-Host: puffin.doc.ic.ac.uk
X-Newsreader: Tin 1.1 PL3
ove...@cray.com (Glen Overby) writes:
: [...]
: Linux seems to have taken most of the MINIX people who really didn't want to
: play with Unix (internals or user level) but just wanted to run Unix on their
: own machine (or, rather, have Unix run itself on their own machine). I'm sure
: it also grabbed a few good hackers from Minix.
I think you have hit the nail on the head there Glen.
I 'played' with Minix 1.2 -> 1.5.10 but I *used* Minix/386 to run Unix
on my own machine. After a careful evaluation of 386bsd, I am now
using Linux instead of Minix/386.
I think hackers like me have migrated to Linux reluctantly because
Minix enjoys good cross-platform support and represents 'good' design.
The problem is that Minix is subject to design constraints imposed by
Andy so that it will fit on minimal PC/XT hardware.
The Monochrome 8MHz CGA PC/XT I ran Minix 1.3 on cost me double what I
paid for a 25MHz 386SX with Colour SVGA. Minix/386 suited me fine and
I've really enjoyed using it, but it has fossilised. I find that
Linux, and the active comp.os.linux group are what I want.
: [...]
: Minix is far from dead.
True but Minix, like Pascal, was designed to *teach* principles of good
design. I've learnt a lot from Minix, but Andy has made it clear that
he does not want Minix to evolve into a 'real' Unix and probably
welcomes the migration of the Free Unix supporters to other groups.
Tony
--
Tony Travis <a...@uk.ac.sari.rri> | Dr. A.J.Travis
| Rowett Research Institute,
| Greenburn Road, Bucksburn, Aberdeen,
| AB2 9SB. UK. tel 0224-712751
Path: sparky!uunet!mcsun!sun4nl!star.cs.vu.nl!ast
From: a...@cs.vu.nl (Andy Tanenbaum)
Newsgroups: comp.os.minix
Subject: Re: Matters in Minix-L
Message-ID: <14862@star.cs.vu.nl>
Date: 9 Jun 92 10:10:43 GMT
References: <1992Jun5.160822.26463@udel.edu>
<1992Jun5.190531.11919@hemlock.cray.com>
Sender: n...@cs.vu.nl
Organization: Fac. Wiskunde & Informatica, Vrije Universiteit, Amsterdam
Lines: 30
In article <1992Jun5.190531.11...@hemlock.cray.com> ove...@cray.com (Glen Overby)
writes:
[Whole bunch of things]
By and large, I agree with what Glen said. A lot of folks really wanted free
BSD, and tried to hijack MINIX in that direction. Then they successively
tried to use Coherent, Linux, BSDI, HURD, and no doubt more in the future.
Fine. MINIX started out as a teaching system, and I intend to try to keep
it that way, although more bloat has crept in than I would have liked. The
format of the previous book--text + code in an appendix--isn't really
feasible any more, with the code now at 600+ pages. This is going to be a
real problem, the solution to which I don't know yet. The German version
has the code in a separate book, shrink wrapped together with the text.
I am not sure if I like that, with or without shrink wrap.
A lot of the silence of late has been related to the fact that I have worked
on MINIX very little in the past 18 months, for various reasons. Nevertheless,
there has been some progress. Version 1.6.19 has most of P1003.1 in it,
except for the terminal stuff. This version is now being tested on the Atari.
The code is largely ANSI-fied, and everything compiles and runs fine with the
ACK ANSI C compiler. I will try to get it working with the K&R compiler
before releasing it to the newsgroup (hopefully the beta test list can help
here)
On the other hand, Amoeba is almost at the point of the first public release.
Although Linux may work well on a 386, I'll bet Amoeba is better if you have
50 of them working together :-) I'll announce more here in due course.
Andy Tanenbaum (a...@cs.vu.nl)
Path: sparky!uunet!mcsun!Germany.EU.net!ira.uka.de!smurf.sub.org!nadia!
aragon.stgt.sub.org!tcl!jmozdzen
From: jmozd...@tcl.stgt.sub.org (Mozdzen, Jens-Uwe, 7033 Herrenberg)
Newsgroups: comp.os.minix
Distribution: world
Subject: Re: Matters in Minix-L
References: <1992Jun6.231057.23959@doc.ic.ac.uk>
Message-ID: <9206081564@tcl.stgt.sub.org>
Organization: tcl network development sites, Herrenberg, Germany
Date: Mon, 8 Jun 92 14:59:19 +0100
Lines: 42
a...@puffin.doc.ic.ac.uk (Tony Travis) writes:
: [...]
: I 'played' with Minix 1.2 -> 1.5.10 but I *used* Minix/386 to run Unix
: on my own machine. After a careful evaluation of 386bsd, I am now
: using Linux instead of Minix/386.
:
: I think hackers like me have migrated to Linux reluctantly because
: Minix enjoys good cross-platform support and represents 'good' design.
: The problem is that Minix is subject to design constraints imposed by
: Andy so that it will fit on minimal PC/XT hardware.
I think that hackers like me stood with Minix for those reasons: IMHO those
aren't constraints but challenges in the basic sense of the word. (This is
no flame! I just want to express that it can do good to a software designer's
evolution when it is required to enhance a system in such a way that the un-
derlying "basic" system is still operable. Some big companies call that "in-
vestment protection", and drive it to a point where normal-minded humans re-
cognize it as being too much; but one musn't blame the principle for excep-
tions.)
On my home system I have donated a big deal of time, arguing, hardware and
money to Minix. Call me a dinosaur, but still I love to see the same system
run on my (extremly poor) XT and my state-of-the-art system.
: [...]
: True but Minix, like Pascal, was designed to *teach* principles of good
: design. I've learnt a lot from Minix, but Andy has made it clear that
: he does not want Minix to evolve into a 'real' Unix and probably
: welcomes the migration of the Free Unix supporters to other groups.
What you say implies that a 'real' Unix does not have a good design :-). What
make a Unix become 'real'?
:
: Tony
: [...]
Regards,
Jens
PS: before the flame wars start all over again, could some of those who would
like to become the 'new Minix gurus' stand up and stop me from feeling all
alone?
--
Jens-U. Mozdzen | jmozd...@tcl.stgt.sub.org
Elsternweg 1 | tcl networks development sites:
W-7033 Herrenberg | "We guarantee fast service -
Germany | no matter how long it takes!"
Newsgroups: comp.os.minix
Path: sparky!uunet!darwin.sura.net!udel!minix
From: "Fred N. van Kempen" <wal...@roxette.uwalt.nl.mugnet.org>
Subject: [LONG] Re: Matters in Minix-L (== comp.os.minix :-)
Message-ID: <1992Jun11.181238.26873@udel.edu>
Originator: m...@nigel.ee.udel.edu
Sender: use...@udel.edu (USENET News Service)
Nntp-Posting-Host: nigel.ee.udel.edu
Reply-To: "Fred N. van Kempen" <wal...@roxette.uwalt.nl.mugnet.org>
Organization: MicroWalt Corporation Software Development
Date: Thu, 11 Jun 1992 01:32:03 GMT
Approved: usenet=re...@louie.udel.EDU
Lines: 154
Humm.
I keep reading (yeah- still unable to post articles and have
them hit the rest of the world- listening, Matthias?) articles
from people who get unhappy with the current situation of
MINIX.
The clearest one I saw was the article by Glen Overby (hi -
still have me in your killfile? :-). It said most of the
things any of us "oldtimers" (my God, I _am_ getting old...)
want to say on the subject. Good work!
> From: jmozd...@tcl.stgt.sub.org (Mozdzen, Jens-Uwe, 7033 Herrenberg)
> Date: Mon, 8 Jun 92 14:59:19 +0100
>
> I think that hackers like me stood with Minix for those reasons: IMHO those
> aren't constraints but challenges in the basic sense of the word. (This is
> no flame! I just want to express that it can do good to a software designer's
> evolution when it is required to enhance a system in such a way that the un-
> derlying "basic" system is still operable. Some big companies call that "in-
> vestment protection", and drive it to a point where normal-minded humans re-
> cognize it as being too much; but one musn't blame the principle for excep-
> tions.)
Eh?
> On my home system I have donated a big deal of time, arguing, hardware and
> money to Minix. Call me a dinosaur, but still I love to see the same system
> run on my (extremly poor) XT and my state-of-the-art system.
Same here, BUT: one cannot choose to support a certain type of system
forever, because that would (obviously) stand in the way of newer
technologies.
If MINIX would suddenly drop all support for the 808[6,8] and the
real mode of the higher-level processors (so, the smallest system
would then be a 286 in PM), and redo the OS to accomodate this
change, much of the system would be a lot easier to understand.
Also, it would be possible to get things done (like 286 page
swapping) that is hard to do in the current kernel, since you
always have to remember "that darned '86 mode"....
I bet it is the same for oMotoral people: people with an 68030
could well do without the "mmu" kludge of the 68000 called
shadowing...
> [...]
> What you say implies that a 'real' Unix does not have a good design :-). What
> make a Unix become 'real'?
Compatibility, and conformance to standards (ehh.. sometimes this is of
course an oxymoron..)
> PS: before the flame wars start all over again, could some of those who would
> like to become the 'new Minix gurus' stand up and stop me from feeling all
> alone?
I am not a MINIX-guru, and never will become one (I am now a PowerMOS
guru :-), but I think this group needs a firm talk with Andy to see
which direction MINIX _is_ going. From talks in the past, I learned
that Andy _is_not_going_to_ bend MINIX into the "commercial" way, i.e.
making MINIX truly useful like Xenix, Coherent (don't start on me on
this one, please..) and SysV or BSD/386.
As one of the guys that bothered him for years trying to get him to
"add some usefulness" to MINIX, I found out that MINIX is, and will
forever be, a teaching OS. It will never be a production-system,
because that would require a very thorough modification-and-cleanup
of the OS (no, don't start about Advanced MINIX either- that's in
the past).
MINIX is still a good learning tool, although I miss any depth of
discussion on this group lately. What happened when I left? :-)
> From: rd...@nmrdc1.nmrdc.nnmc.navy.mil (LCDR Michael E. Dobson)
> Date: Tue, 9 Jun 1992 16:44:38 GMT
>
> In article <14...@star.cs.vu.nl> a...@cs.vu.nl (Andy Tanenbaum) writes:
Humm. I never got that. Andy, could you mail it to me?
> >On the other hand, Amoeba is almost at the point of the first public release.
> >Although Linux may work well on a 386, I'll bet Amoeba is better if you have
> >50 of them working together :-) I'll announce more here in due course.
Amoeba is a great thing if you have a spare tun of heavy hardware,
and decent networking equipment. I dislike Linux' structure as much as
Andy does, but I think Michael is right:
> But that's apples and oranges Andy. Now I'll agree if you had said
> Linux+NFS or 386BSD+NFS on 50 '386s vs Amoeba, but how many individuals are
> in a position to want to do that?
You cannot compare a single-CPU "hobby" system (sorry, Linus...) to
a multi-CPU distributed operating system like Amoeba. I saw Amoeba
running, and I know some of its authors personally (hi Jack!, Sape!).
It runs like magic on the VUNET network, but it comes to a screeching
"#asm HALT" when run on a single Sun or 386 machine.
Q to Andy: did you ever try to run Amoeba + Bullet + the UNIX emulator
on the same machine? I was once tempted to do so (as you know), but
you told me yourself that it would "barely" fit and/or run...
LINUX is an OS that tries to get the most out of the 386 CPU, and,
hearing many people talking about it, it does a good job of it. The
fact that I dislike its structure has nothing to do with the fact
that LINUX is a (GOOD/NOT GOOD- whatever _you_ think) OS. It tempts
the 386 into doing things the 386 itself didn't know it was capable
of... :-)
Coherent is (currently) a UNIX-for-small-systems, like MINIX. Since
it is a commercial product, it costs, and sources are not included.
However, with Coh 4.0 coming out, many people that are currently
running Xenix or MINIX on their 286's or 386SX'es might have a peep
of what Coh4 can do for them. It might be better than Linux, Minix
and the like, or be more compatible, or whatever. Who knows?
386BSD and BSDI/386 are the ultimate in BSD compatiblity (great
deduction Fred... :-), but it seems that they are true resource
hogs, and 386BSD is (as I hear) quite unstable yet.
This discussion started out with the question why so little traffic
in general, and even more traffic worth-while reading was appearing
on MINIX-L. Many of us know, that several of the article-posting-
and-program-writing people of this group (ehh.. sorry: LIST :-)
have moved towards new projects like LINUX, Coh, MACH and the
like, or simply bought a "standard" version of UNIX.
What is left is a community of newcomers, with an Andy posting
articles that seem to be different in tone than when I was a
newcomer (waaayyyyyy back :-). Many people got angry over the
"Compiler && 1.6/2.0" discussion, since they felt that either
they will have to pay a lot of $$ to get 1.6 running now, _or_
they'll have to wait until the official 2.0 comes out..
Apart from that, I am again getting messages like "can you
please send me Advanced MINIX, because 1.6 seems to be a
well-kept secret of The Beta Group". Especially the latter
is the reason for me posting this- I am no longer involved
in MINIX, so I could care less..
Anyway. WHAT IS WRONG WITH YOU OUT THERE !
If you want activity on this group, create it yourselves.
Write and post programs, find problems in MINIX and start
discussions over it, or answer questions from newcomers.
The evolution of MINIX is for 80% the hard work of people
on the Net, with Andy being the great filter to keep things
somewhat organized (which is the problem wih Linux, I think).
If _you_ do not show up with new things for MINIX, how should
things improve at all?
With (love && kisses);
char *name = "Fred N. van Kempen";
char *mail = "wal...@roxette.uwalt.nl.mugnet.org";
exit(OK);
--
Fred N. van Kempen, wal...@roxette.uwalt.nl.mugnet.org
"Love is - what you want it to be. Alannah Myles"
--
Mail System (MMDF)
Newsgroups: comp.os.minix
Path: sparky!uunet!email!hp
From: h...@vmars.tuwien.ac.at (Peter Holzer)
Subject: Amoeba on a single machine (was Re: Matters in Minix-L)
Message-ID: <1992Jun12.123512.9110@email.tuwien.ac.at>
Sender: n...@email.tuwien.ac.at
Nntp-Posting-Host: quasi.vmars.tuwien.ac.at
Organization: Technical University Vienna, Dept. for Realtime Systems, AUSTRIA
References: <1992Jun11.181238.26873@udel.edu>
Date: Fri, 12 Jun 1992 12:35:12 GMT
Lines: 26
"Fred N. van Kempen" <wal...@roxette.uwalt.nl.mugnet.org> writes:
>[Amoeba] runs like magic on the VUNET network, but it comes to a screeching
>"#asm HALT" when run on a single Sun or 386 machine.
>Q to Andy: did you ever try to run Amoeba + Bullet + the UNIX emulator
> on the same machine? I was once tempted to do so (as you know), but
> you told me yourself that it would "barely" fit and/or run...
This has been bothering me since I heard the first time about Amoeba.
Why does Amoeba need a seperate File server, CPU Server and
Workstations? Is it just the size of the system, or is there something
in the architecture, which makes it infeasible to run it on a single
computer?
If it is the former, how much RAM/Disk would I need on my home PC to
get a working system (i.e. one on which I can develop programs, write
articles, etc. I don't want to simulate the Viennese traffic or
schedule 5000 processes onto 200 processors, or whatever you use an
Amoeba cluster for in real life)
--
| _ | Peter J. Holzer | Think of it |
| |_|_) | Technical University Vienna | as evolution |
| | | | Dept. for Real-Time Systems | in action! |
| __/ | h...@vmars.tuwien.ac.at | Tony Rand |
Path: sparky!uunet!mcsun!sun4nl!star.cs.vu.nl!ast
From: a...@cs.vu.nl (Andy Tanenbaum)
Newsgroups: comp.os.minix
Subject: Re: Amoeba on a single machine (was Re: Matters in Minix-L)
Message-ID: <14941@star.cs.vu.nl>
Date: 12 Jun 92 14:30:08 GMT
References: <1992Jun11.181238.26873@udel.edu>
<1992Jun12.123512.9110@email.tuwien.ac.at>
Sender: n...@cs.vu.nl
Organization: Fac. Wiskunde & Informatica, Vrije Universiteit, Amsterdam
Lines: 19
In article <1992Jun12.123512.9...@email.tuwien.ac.at> h...@vmars.tuwien.ac.at
(Peter Holzer) writes:
>This has been bothering me since I heard the first time about Amoeba.
>Why does Amoeba need a seperate File server, CPU Server and
>Workstations? Is it just the size of the system, or is there something
>in the architecture, which makes it infeasible to run it on a single
>computer?
The real reason is that we have never bothered to try it, so I am safer
saying it won't work. If I say it works and it doesn't, everyone gets
mad. If I say it doesn't and it does, nobody gets mad.
Size is definitely an issue. The file server should have a cache of at
least 16M, preferably 32M to work well. It caches whole files
contiguously in memory. If you add in all the other servers, you'd
probably need 64M to make it all fit. Amoeba was designed as the operating
system for a network of powerful, large computers. That is the other
end of the spectrum from MINIX.
Andy Tanenbaum (a...@cs.vu.nl)