Return to BSD News archive
Xref: sserve comp.lang.modula3:1742 comp.os.386bsd.questions:2942
Newsgroups: comp.lang.modula3,comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!munnari.oz.au!spool.mu.edu!howland.reston.ans.net!xlink.net!math.fu-berlin.de!news.netmbx.de!Germany.EU.net!mcsun!sunic!psinntp!psinntp!laidbak!tellab5!ruth.wheaton.edu!johnh
From: johnh@david.wheaton.edu (John C. Hayward)
Subject: Modula-3 port to 386bsd alpha testers wanted
Message-ID: <1993Jun8.073722.7820@wheaton.wheaton.edu>
Followup-To: johnh@david.wheaton.edu
Keywords: Modula-3 386bsd
Sender: news@wheaton.wheaton.edu (USENET News System)
Organization: Wheaton College, IL
Date: Tue, 8 Jun 1993 07:37:22 GMT
Lines: 43
Several people have posted queries about a Modula-3 port for 386bsd. I have
been working one and it seems to work and is complete except for the IEEE
floating point parts.
The compiler thrashes badly if you only have 4MB of memory but seems to
perform as reasonably as one would expect if you have 8MB of memory.
(The driver takes about 3MB of virtual memory as does the compiler itself).
The current port assumes the existance of the math co processor for 386.
There were a number of problems related to 386bsd which made the port more
difficult - including.
ar - if the length of a file name is longer than 16 and odd the resultant
library is unusable to ld and ranlib
sh, csh - do not follow the convention for allowing the first line of the shell
script to specify which shell to invoke.
cc, float.h - core dumps if you try to use HUGE as a floating point constant.
cc, dtoa.h - default rounding of floats to int was round rather than trunc.
/* what does your compiler do with i = 1.6;? */
make - makefiles and SEP seem to behave differently in 386bsd and other unix
platforms
The biggest problem was my own falt in trying to get threads working dealing
with setting up a stack for a new thread.
I have run the ~400 tests included in the distribution and aside from extra
cc warning messages, different cc warning and error messages, the IEEE
floating point tests, rounding errors and number of procedures test I
think all the tests performed as expected.
I am looking for a few brave souls willing to try the port out.
Hardware requirements - 386, 486 with co-processor and 8MB of memory.
Might need 100-150MB of disk space to build and play with the system.
runs on 386bsd0.1 (0.2 won't be out until later this summer and I have not
found the patches yet:-{).
The distribution of the compiler (v1.11) I think will take about 4MB.
I am in Malaysia till the end of the summer and from time to time
our connections via internet go down.
Send e-mail to johnh@david.wheaton.edu if interested.
johnh...