🚀 go-pugleaf

In article <2521@keele.keele.ac.uk> csa09@seq1.keele.ac.uk (P. Singleton) writes:
>
>Chris Mellish told me of a trick attributed, in Prolog folklore, to
>Colmerauer, whereby a subset of N values is denoted by a list of N+1
>variables, each adjacent pair corresponding to one possible element.
>Iff a pair share, that element is not in the subset.
> [...] can anyone cite an original or early
>appearance of this trick?

I believe Chris heard of the trick and the attribution from me.
When I visited Marseille in 1977, Alain Colmerauer gave me as a puzzle
the question of representing subsets of a finite
set so that non-empty intersection is represented by unification.
I spent a long evening working on it, and came up with some encoding,
but not as elegant as what you describe, which is the solution that
Alain showed me the next morning. I don't think the trick was ever described
in print by him.
>I have generalised it to domains which may grow after some
>subsets have already been denoted. [...]
> Dare I claim originality for these in my thesis?
As far as I know, that is new.

Fernando Pereira
AT&T Bell Laboratories, Murray Hill, NJ
pereira@research.att.com