Path: utzoo!attcan!uunet!seismo!sundc!pitstop!sun!quintus!ok
From: ok@quintus.UUCP (Richard A. O'Keefe)
Newsgroups: comp.lang.prolog
Subject: Re: Atom-based module systems
Message-ID: <967@cresswell.quintus.UUCP>
Date: 13 May 88 23:35:31 GMT
References: <136@vor.esosun.UUCP> <841@cresswell.quintus.UUCP> <365@aiva.ed.ac.uk> <425@aiva.ed.ac.uk>
Organization: Quintus Computer Systems, Mountain View, CA
Lines: 24

In article <425@aiva.ed.ac.uk>, jeff@aiva.ed.ac.uk (Jeff Dalton) writes:
> The reason I asked is that your recent remarks on the impossibility of
> sharply distinguishing code from data would tend to support "atom-
> based" designs.

It is possible to distinguish code from data *PROVIDED* you add declarations
to the language.  The Mycroft & O'Keefe type checking paper suggests on the
last page having a family of types void, void(T1), ..., void(T1,...,T93), ...
where if :- pred p(<t1>,<t2>,<t3>) then
	p(X,Y,Z) : void	if X:<t1> and Y:<t2> and Z:<t3>
	p(X,Y) : void(<t3>) if X:<t1> and Y:<t2>
	p(X) : void(<t2>,<t3>) if X:<t1>
	p : void(<t1>,<t2>,<t3>)
and so on.  As I recall it, this was Alan Mycroft's idea.
The :- meta_predicate declarations in Quintus Prolog are just such type
declarations, where void(_1,...,_n) is abbreviated to n, and the other
types are abbreviated to +,-,or ?.  Thus the type of maplist/2 is
	:- pred maplist(void(T), list(T)).
but this is abbreviated to
	:- meta_predicate(1, +).

The BSI substandard, however, attempts to draw a distinction between code
and data *solely* by means of syntax, without the aid of any such
declarations.