See Summary! M-2 discussion only in comp.lang.modula2 THANKS.

26 views
Skip to first unread message

Brad Taplin

unread,
Mar 14, 1992, 2:44:42 AM3/14/92
to
Hello. I included these four newsgroups because I consider them
likely bastions of experience and intelligence and because each
might have a certain passing interest. Forgive the oddity of it.
Please post discussions relevant to M2 to the M2 newsgroup, and
PLEASE do NOT discuss things unrelated to Modula-2 therein. :)

You have seen the movie? Could Modula-2 be used to design HAL?
Remember that this would involve remarkable progress in systems
design, robotics, and realtime application of AI. It would also
certainly involve some fault-tolerance mechanisms and various
data management and communications advances, but those strike
me as considerably-less-daunting challenges to such a project.

Remember that niether I nor many of the readers of this silly
line are necessarily terribly familiar with Modula-2. I have
heard recently that Modula-2's more popular than C in Europe,
that it can provide very nicely for advanced OOP programming,
that it has been used extensively for embedded systems, that
it will run on most any platform, and that its strict typing
and other restraining features are conducive to huge projects
in which "creative coding" as C allows could spell disaster.
Above all, the modularity seems well-suited to the challenge.

Well?

John M. Hughes

unread,
Mar 14, 1992, 5:48:32 PM3/14/92
to
In article <1992Mar14.0...@bronze.ucs.indiana.edu> bta...@silver.ucs.indiana.edu (Brad Taplin) writes:
<intro deleted>

>You have seen the movie? Could Modula-2 be used to design HAL?
>Remember that this would involve remarkable progress in systems
>design, robotics, and realtime application of AI. It would also

Now that's a classic definition of an understatement if I ever
heard one... ;-)

>certainly involve some fault-tolerance mechanisms and various
>data management and communications advances, but those strike
>me as considerably-less-daunting challenges to such a project.
>
>Remember that niether I nor many of the readers of this silly
>line are necessarily terribly familiar with Modula-2. I have
>heard recently that Modula-2's more popular than C in Europe,
>that it can provide very nicely for advanced OOP programming,
>that it has been used extensively for embedded systems, that
>it will run on most any platform, and that its strict typing
>and other restraining features are conducive to huge projects
>in which "creative coding" as C allows could spell disaster.
>Above all, the modularity seems well-suited to the challenge.
>Well?

I have been using Modula-2 for many years now, and I much prefer it over
C for most projects. It has been my experience that projects which would
otherwise be unwieldy in C are much easier to manage with M-2. I would
agree with all of the assertions in the previous paragraph.

For an example of what can be done with M-2 when it comes to OS design
and implementation, check into the Lumos-2 operating system by G. R.
Guenther of Laurentian University.

It is my opinion that Modula-2 has failed to gain a large audience in the
US because it does not readily lend itself to "stream-of-consciousness"
styles of programming. Module hierarchy definitions require some advanced
planning. In C, on the other hand, it is easy to just start writing code,
and then flip to another window on the editor to create an additional
module. It would seem that Modula-2 is faithful to the credo of modularity
to the point of occasional annoyance. I have also wondered if Modula-2 has
been snubbed by some programmers because of its obvious family ties to
Pascal.

As to M-2's applicability for AI projects, IMHO it is far more suited
to this type of application than C, if for no other reason than that it is
possible to state certain algorithmic constructs in Modula-2 in a much
more concise form than it is with C (not neccesarily better, just more
concise). However, Modula-2 would not be my first choice for a very large
AI-ish project. Large M-2 applications can become so verbous with module
interdependance declarations as to become difficult to read. I would
probably opt for a special-purpose language designed for the task at hand,
with the syntactic features and strict typing of Modula-2, but with
link-time object integration more along the lines of C (ie. reduce or
eliminate the wordy module import/export decl's of M-2 and let the linker
sort it all out). Perhaps Oberon is more along these lines, but I haven't
played with it enough to know.

Woodrow Yeung

unread,
Mar 15, 1992, 3:12:39 AM3/15/92
to
In article <33...@caslon.cs.arizona.edu> j...@cs.arizona.edu (John M. Hughes) writes:

>For an example of what can be done with M-2 when it comes to OS design
>and implementation, check into the Lumos-2 operating system by G. R.
>Guenther of Laurentian University.

Where can I get information on the Lumos-2 operating system? Is source code
available for a nominal fee?

Thanks

Woodrow Yeung
ye...@reed.edu

John M. Hughes

unread,
Mar 15, 1992, 4:02:55 AM3/15/92
to

You can write to:

Grant R. Guenther
Department of Mathematics and Computer Science
Laurentian University
Ramsey Lake Road
Sudbury, Ontario, Canada P3E 2C6

or try:

gr...@ramsey.cs.laurentien.ca

Grant will send out a printed version of the sources (there used to not be
a charge, but with the rising cost of life in general that may have changed),
all nicely bound. I don't know what the current policy is on machine readable
sources. Also, Lumos-2 was written for a 68000-based system, and there are
some minor deviations from "standard" Modula-2 (the document has details).

Robert W Murphree

unread,
Mar 18, 1992, 2:55:21 PM3/18/92
to
Sorry to hear that C is worse than Modula-2 for large AI projects.
I am writing a simulation program for navigation schemes for a mobile
robot. I have seen several article titles and talked to people on the
Phone at TRC encouraging me to use an object oriented style to write
mobile robot software. I am writing my program in C++. Does anyone
have any comments about writing mobile robot software using an
object oriented style?

Grant R. Guenther

unread,
Mar 17, 1992, 10:25:29 AM3/17/92
to
In <33...@caslon.cs.arizona.edu> j...@cs.arizona.edu (John M. Hughes) writes:
>In article <1992Mar15.0...@reed.edu> ye...@reed.edu (Woodrow Yeung) writes:
>>Where can I get information on the Lumos-2 operating system? Is source code
>>available for a nominal fee?

>gr...@ramsey.cs.laurentien.ca
^
Oops, it's actually gr...@ramsey.cs.laurentian.ca

>Grant will send out a printed version of the sources (there used to not be
>a charge, but with the rising cost of life in general that may have changed),
>all nicely bound. I don't know what the current policy is on machine readable
>sources. Also, Lumos-2 was written for a 68000-based system, and there are
>some minor deviations from "standard" Modula-2 (the document has details).

I'll try to post a summary of the current status of the LUMOS-2 project
tonight. I haven't been following this group for a few months - didn't
know some interest had developed again.

In brief: the good news is that we've just about finished adding TCP/IP into
the system, the bad news is that our machine is now 7 years old and very
near to dying.

----------------------------------------------------------------------
Grant R. Guenther Math & Comp. Sci., Laurentian U., Sudbury, Canada
gr...@ramsey.cs.laurentian.ca
----------------------------------------------------------------------

Josef Moellers

unread,
Mar 19, 1992, 9:07:59 AM3/19/92
to

>Hello. I included these four newsgroups because I consider them
>likely bastions of experience and intelligence and because each
>might have a certain passing interest. Forgive the oddity of it.
>Please post discussions relevant to M2 to the M2 newsgroup, and
>PLEASE do NOT discuss things unrelated to Modula-2 therein. :)

>You have seen the movie? Could Modula-2 be used to design HAL?
>Remember that this would involve remarkable progress in systems
>design, robotics, and realtime application of AI. It would also
>certainly involve some fault-tolerance mechanisms and various
>data management and communications advances, but those strike
>me as considerably-less-daunting challenges to such a project.

In theory ... yes.
It can be proven that, apart from some languages, all languages are
created equal, i.e. a problem that can be coded in one language can also
be coded in any other language, albeit with varying degrees of
difficulties and (probably) performance.

The other question is, whether anything like HAL can be programmed at
all. If You read e.g. Hofstaedters (sp?) book "Goedel Escher Bach", he
comes to the conclusion that real intelligence cannot be programmed but
has to be learned. Therefore all You probably need is a "small" kernel
to manage the information and a HUGE information base.

But then ... I might be all wrong.
--
| Josef Moellers | c/o Siemens Nixdorf Informationssysteme AG |
| USA: molle...@sni-usa.com | Abt. STO-XS 113 | Riemekestrasse |
| !USA: molle...@sni.de | Phone: (+49) 5251 835124 | D-4790 Paderborn |

Grant R. Guenther

unread,
Mar 20, 1992, 8:15:44 AM3/20/92
to
In <1992Mar17.1...@ramsey.cs.laurentian.ca>
gr...@ramsey.cs.laurentian.ca (Grant R. Guenther) writes:

>I'll try to post a summary of the current status of the LUMOS-2 project
>tonight.

I had intended to post this two days ago, but we had some network problems,
Sorry.

LUMOS-2:

The Laurentian University Modula-2 Operating System is a conventional
single-processor, multi-programming, multi-user system written in
Modula-2 for use in our instructional sequence in Operating Systems.
The system was originally developed in 1986 and was influenced by
D.E. Comer's XINU books. It was built to run on the Stride-440, a
moderately popular MC68000 box with a simple MMU that was available
at the time. The M2/68K compiler from H. Burkhart's group (then
at ETHZ-IFE) was available in source form and was the starting point
for the project.

Since 1986, many (senior undergraduate) students have worked on the
system - contributing new utilities, and subsystems. LUMOS-2 is
sufficiently robust that all development is done using the system itself.
In this respect it is quite similar to A. Tanenbaum's MINIX system.

Most recently a team of students has been working to incorporate TCP/IP
into the system. With the completion of that work, LUMOS-2 will provide
approximately 120 system calls. The OS load image is about 125K.
We occasionally print out the source files and bind them up with some
other documentation for use a course text and for distribution to
interested readers. This printout is now about 800 pages long and
includes everything except the M2/68K compiler itself.

Our resources for this project are very limited. The Stride-400 is
now obsolete and it is becoming increasingly susceptible to overheating
problems. It must be replaced. From many perspectives and ugly though the
architecture may be, it looks like we should be moving the work into
the 80486 arena.

So, to those of you that were inquiring about LUMOS-2 source code
availability: if there is sufficient interest, I can arrange for
the source tree to be made accessible for anonymous FTP. I will
not be printing hardcopy listings at this time. Send me e-mail
if you would like this set up.

I would appreciate some information. I intend to spend a portion
of my upcoming leave working on an upgrade or replacement for
LUMOS-2. It is a basic requirement of the project that a compiler
exist that can be ported into the new environment. We did have
a Modula-2 cross-compiler for the 386 that ran under VAX/VMS. We
can no longer use it and attempts to port it to any of the existing
32 bit compilers (that we can afford) have met with enormous difficulties.
(Primarily because of major differences in the language.)

If anyone can help us find affordable M2 compiler source for the 386,
so that this project can continue, I'd be grateful.

I have considered switching to Oberon. Am I correct in recalling a rumour
of a 386 port of Oberon becoming available at some point ? Will it
be possible to acquire the sources without sacrificing a first-born
child ?

Jon Guthrie

unread,
Mar 21, 1992, 7:35:23 PM3/21/92
to
In a message on <Mar 19 00:38>, Grant R. Guenther (1:105/42) writes:

>In brief: the good news is that we've just about finished adding
>TCP/IP into the system,

That's funny, I thought _I_ was the only one working on TCP/IP in Modula-2.

--
uucp: uunet!m2xenix!puddle!106!2000.25!Jon.Guthrie
Internet: Jon.G...@p25.f2000.n106.z1.fidonet.org

Reply all
Reply to author
Forward
0 new messages