Newsgroups: comp.sys.transputer
From: ifunltd@xs4all.hacktic.nl (ifunltd)
Subject: Re: Revision of occam 2 Reference Manual...
Organization: Hack-Tic, networking for the masses
Date: 8 Aug 1994 06:13:57 GMT
Message-ID: <324if5$dr2@news4all.hacktic.nl>

Andres Kruse (NIKHEF),1c/137,2909 (A.Kruse@nikhef.nl) wrote:
: What about a questionaire to OCCAM2 users about what they would like
: to have implemented in new OCCAM versions ? I guess many people will agree
: that:

: - structures are a MUST !
: - pointers should be there as well. 
: - 'PAR FROM something FOR something' where 'something' are variables
:   which can change at run-time.
: - RETURN from a PROC or FUNCTION... (dunno how to handle RETURNs in
:   a PAR.. maybe the compiler could restrict the use of RETURNs to
:   PROCs which don't have PARs..)

Are you sure you do not want to write in C, when that language has a few
parallel OCCAM-like extensions ? You must know there's one available, and
even in your home town. The Par.C System that I have co-designed ages ago
has a "par" construct with a runtime-determined replicator, like:
par(i=0;i<4;i++)
and of course the C language has structures and pointers,plus return values
from procs. Return values from a par are not nice indeed.

If you want to have a look at that real parallel C compiler, call Paul Visser
at ACE: 020-6646416. The system is now available under the name PACT parallel
C compiler, but that has the same roots. Say hello from JANG to Paul.

The right person to ask about the types of extensions to OCCAM you describe
is probably not on this planet. Except for the first part, (RECORDS are in
OCCAM3), your extensions are contrary to the design goal of OCCAM, and will
never be admitted into the language. And I can see their point.

Good luCk ;-)

JANG -> ifunltd

