Newsgroups: comp.sys.transputer
From: Geraint.Jones@wolfson.oxford.ac.uk (Geraint Jones)
Subject: Re: OCCAM3 (was :- Revision of occam 2 Reference Manual)
Organization: Bread Pudding with Lashings of Whipped Cream
Date: Mon, 15 Aug 1994 16:47:14 GMT
Message-ID: <14304588851827519940815163818@comlab.oxford.ac.uk>

(  > This reminds me another question. Why is it a implementation restriction
 ) >  if you use a PAR construct inside a FUNCTION???
(  > 
 ) > Kang Hsin Lu
(  
 ) I've always thought that the reason for this was fairly simple: functions
(  should be deterministic.  If you allow PAR inside a function (or indeed
 ) any VALOF) then it may not be deterministic any more and lots of the
(  nice rewrite rules for occam stop working.
 ) 
(  	Bryan Scattergood

You're hard put to it to write a non-deterministic VALOF with PAR but no ALT.
I argued for allowing PAR in VALOFs when the (first) occam2 manual was being
sorted out, but I was sat on by {someone} from Inmos and told not to be so
silly.

I hope that "implementation restriction" is a technical term meaning that it
is seen as being valid occam2, but not supported by that compiler, because
the author could not be bothered to work out how to do the storage allocation.
(Or am I being silly again?)
									g

