Path: utzoo!attcan!uunet!munnari!munnari.oz!lee
From: lee@munnari.oz (Lee Naish)
Newsgroups: comp.lang.prolog
Subject: Re: bagof/setof
Message-ID: <2362@munnari.oz>
Date: 14 Sep 88 00:41:17 GMT
References: <1436@kulcs.kulcs.uucp>
Sender: news@munnari.oz
Reply-To: lee@munmurra.UUCP (Lee Naish)
Organization: University of Melbourne, Comp Sci Dept
Lines: 27

In article <1436@kulcs.kulcs.uucp> bimbart@kulcs.UUCP (Bart Demoen) writes:
>	p(1,a) .
>	p(_,b) .
>	p(2,c) .
>	?- setof(Y,p(X,Y),L) .
>
>	SICSTUSprolog:
>	X = 1 , L = [a,b]
>	X = 2 , L = [b,c]

"Logically", there are three answers (the last one containing two
constraints on X):
	X = 1 , L = [a,b];
	X = 2 , L = [b,c];
	X ~= 1, X ~= 2, L = [b].
Therefore, I think Sicstus gives the most "logical" set of solutions.
In other, less complex cases, setof generally returns what a logical
all solutions predicate returns except for answers which must contain
constraints.  It seems that Sicstus is more consistent than other
systems.  Does anyone know if it is completely consistent?
Perhaps there is another "logical" way of explaining the behaviour of
other systrems.  (NU-Prolog setof behaves the same as C-Prolog, and we
hope most Edinburgh systems, for compatability).

	lee

	lee