🚀 go-pugleaf

In article <8707081024.AA21949@cayuga.cs.rochester.edu> Patrik Eschle writes:

>>Does anyone know of a Modula-2 compiler for UNIX System V, and specifically
>>one that will work with an AT&T 3B2/300? --Roger Terrell
>
>There exixsts one, called MODULA-2/68 for 68xxx processors. Systems
>supported are (June 86, 1 year ago): Alto 3068 (Sys V), IRIS Workstations
>(UniPlus), Sun (BSD 4.2), Tektronix 4400 Workstation (Uniflex), VAX Unix
>and VAX VMS 4.2, Philips VME Systems (UniFive).
>
>It costs min. $500. From what i saw from the users guide, the implementation
>is not that good, they did not find a convincing solution for the
>I/O modules - I decided on Oregon Pascal.
>

We have recently this compiler for our Sun 3/50s.  [We bought it via
Real Time Associates, a British firm]  I think our licence set us back
2000 pounds [roughly]

The compiler still has some bugs, and doesn't interface to C terribly
well [the parameters are REVERSED !!].  However, RTA and the Dhajaheri
Brothers seem to respond positively to bug reports.... we started with
version 1.4, and are now on 1.6 !  As yet, it is only a Revision 2
compiler - no MIN and MAX [sob].

As to I/O modules... when AREN'T these a problem ?  Until someone gets their
finger out and defines a nice, easy-to-use, polymorphic, variable-no-parameters,
and, most importantly, STANDARD I/O system, I/O is a perpetual problem.
For what it's Wirth :-) InOut appears to be implemented adequately, although
without ReadAgain.

Bring back writeln [or printf] !  All is forgiven !

Overall, I would say I'm reasonably happy with the compiler... despite a
six hour debugging session trying to work out how passing strings to a C
module was [sometimes!] giving a segmentation fault in a SunTools routine !
I never did fix it... I just fudged it...

Obviously, I can't say what happens on other 68K targets..

	Duncan.

-----------------------------------------------------------------------------
JANET address : dcw@uk.ac.ic.doc| Snail Mail :  Duncan White,
--------------------------------|               Dept of Computing,
  Disclaimer: I speak for me,   |               Imperial College,
  and me alone.....             |               180 Queen's Gate,
  (sometimes, heh heh, even I   |               South Kensington,
  disagree with what I say)     |               London SW7
----------------------------------------------------------------------------
Tel: UK 01-589-5111 x 4982/4991
----------------------------------------------------------------------------

In article <473@ivax.doc.ic.ac.uk> you write:
>In article <8707081024.AA21949@cayuga.cs.rochester.edu> Patrik Eschle writes:
>
>>>Does anyone know of a Modula-2 compiler for UNIX System V, and specifically
>>>one that will work with an AT&T 3B2/300? --Roger Terrell
>>
>>There exixsts one, called MODULA-2/68 for 68xxx processors. Systems

[...]

>
>We have recently this compiler for our Sun 3/50s.  [We bought it via

[...]

>
>The compiler still has some bugs, and doesn't interface to C terribly
>well [the parameters are REVERSED !!].  

[...]

I spent some time using this compiler this spring.
It seemed like a very nice package with some good provisions
for setting up directory paths to be searched for modules 
not found in the current directory and some other nice features.
However to say that it "doesn't interface to C terribly well"
is probably being nice. Check the M2 representation of real numbers
against that for C. As I recall they're quite different. This means
that the real values returned from system calls must be twiddled to
be used by M2. The manual also gave a lot of information about the
"+g" option (or something like that) to allow symbolic debuggers
to be used. However in a small paragraph in the last two or three
pages they explain that dbx doesn't really support *all* of the
M2 data structures so the +g option has been turned off on all UNIX
releases. :-(


	David Huelsbeck
	dph@lanl.gov.arpa
	{ihnp4,cmcl2}!lanl!dph