Path: utzoo!attcan!uunet!mcvax!ukc!etive!lfcs!jha
From: jha@lfcs.ed.ac.uk (Jamie Andrews)
Newsgroups: comp.lang.prolog
Subject: Re: The meaning of "declarative"
Message-ID: <859@etive.ed.ac.uk>
Date: 19 Oct 88 11:06:06 GMT
References: <ALF.88Oct2184429@hathor.sics.se> <818@etive.ed.ac.uk> <590@dcl-csvax.comp.lancs.ac.uk>
Sender: news@etive.ed.ac.uk
Reply-To: jha@lfcs.ed.ac.uk (Jamie Andrews)
Organization: Laboratory for the Foundations of Computer Science, Edinburgh U
Lines: 62


     Hum, I guess I should be happy that I got *some* response,
although I didn't really expect this...

     Perhaps I wasn't clear enough in what I was saying.  I was
not asserting that Prolog is "more declarative" than Lisp or
that you can program everything declaratively in Prolog.  I was
comparing the logic programming paradigm to the functional
programming paradigm, in general.  I was pointing out that the
overall structure of logic programs makes them suitable for
problem domains with lots of logical dependencies, whereas the
overall structure of functional programs makes them suitable
for problem domains involving functions -- seemingly fewer.
Though functions come up naturally quite frequently in programs,
they seldom seem to capture the essence of the whole problem.

     That doesn't mean that functional programming is "bad",
just that people seem to use functional programming languages
for reasons having nothing to do with the utility (or lack
thereof) of the functional programming paradigm.

In article <590@dcl-csvax.comp.lancs.ac.uk> simon@comp.lancs.ac.uk (Simon Brooke) writes:
>Oh, come on now! Very, very few people can even be taught logic. Most of
>those who can will always find it excrutiatingly difficult. It is a most
>unnatural and tedious way of 'deducing information from assumptions'; and
>if you cand find anyone who thinks logically, I wish you'ld introduce them
>to me - it would be a new and fascinating experience.

     These statements are quite bizarre to me.  Formulae are
formalisations of mathematical statements.  Derivations are
formalisations of proofs.  Proof systems are often given direct
philosophical justifications, from introspection and observation
of the way informal proofs in papers work.  If you meant
something like "proof systems are not close to the way we search
for true assertions", I would agree with you, but that's not
what I was saying.

>                   ... Prolog is not an easy language in which to write an
>inference engine....
>                             ... Most Prolog programs which do anything useful
>are procedural; but when a piece of procedural Prolog goes wrong, it is 
>(in my experience) horrible to debug.

     You're talking about things to do with the problems of
control, determinism, ease of doing code with assignment,
etc., in Prolog.  I didn't deny these problems exist, but you
must realise that these are being addressed in more modern 
logic programming languages like Trilogy.  I was talking,
in any case, about the general applicability of the paradigms.

     Your comments about useful Prolog programs also apply to
useful programs in the Langage International Superieur de
Programmation (:-)).  In most of these, the functional structure
is just superficial and the real work is non-declarative.  The
difference is that the functional structure (*alone*) rarely
buys you anything that you couldn't get in imperative programs,
whereas a logical structure (*alone*) often does.

--Jamie.
  jha@lfcs.ed.ac.uk
"Autumn, to me the most congenial of seasons;
 the university, to me the most congenial of lives" --R.Davies