Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!quintus!ok
From: ok@quintus.UUCP (Richard A. O'Keefe)
Newsgroups: comp.lang.prolog
Subject: Re: :- deterministic (not HP  Prolog, but Trilogy)
Message-ID: <906@cresswell.quintus.UUCP>
Date: 27 Apr 88 18:49:23 GMT
References: <900@cresswell.quintus.UUCP> <2217@ubc-cs.UUCP>
Organization: Quintus Computer Systems, Mountain View, CA
Lines: 32
Keywords: Deterministic predicates and also Cuts

In article <2217@ubc-cs.UUCP>, voda@ubc-cs (Paul Voda) writes:
> I do not know how the deterministic predicates are implemented in HP-Prolog,
> but I do know what they are in Trilogy.  Maybe this will help.
Not really.  Trilogy is a logic programming language which makes no
pretence of being Prolog.  (In some ways it resembles ALEPH, except
that it is a lot purer.)  Trilogy is allowed to reject programs as
invalid based on a data flow analysis; Prolog systems are not.
The HP-Prolog manual I have a (partial) copy of does not mention any
restriction on :-deterministic predicates at all, and as it is an
incremental system, I doubt that it is doing much data-flow analysis.

> I have written a paper on the one solution construct which I will
> mail to all interested parties on request. The title is:
> 
>      The Logical Reconstruction of Cuts as One Solution Operators.

I have received a copy of this.  It's a nice paper, but it makes me even
more unhappy with cuts than ever.  There is a question: how often do we
really need the _control_ component to say "commit to the first solution"
rather than "commit to any solution"?  Consider the contrast between

	member(X, [H|T]) <-> X = H | member(X, T)

	member1(X, L) <-> first member(X, L)
-vs-

	member1(X, [H|T]) <-> X = H | X ~= H & member1(X, T)
-or-	member1(X, [H|T]) <-> if X = H then true else member1(X, T)
[if-then-else being sound as in NU Prolog].

It would be really interesting to see some sketches of how people are
using once(Goal), [! Goal !], or (Goal->true) {all the same thing}.