[mercury-users] A suggestion for a tutorial

[Peter Schachte]

On Mon, Jul 03, 2006 at 03:13:35PM +0200, Nicolas Pelletier wrote:
> Hello again,
> 
> as a newbie to Mercury (and not  much experienced in logic programming
> either), I  find myself making  the following error  more often than I
> would like ;-)
> 
> While writing predicates such as this one:
> 
> :- pred square_list(int :: in, int :: in, list(int) :: out) is nondet.
> square_list(N, M, L) :-
>     N > 0 -> member(X, 1 .. M), square_list(N - 1, M, L0), L = [X | L0]
>     ; L = [].
> 
> I often feel the  impulse to "over-factorize" the  code and write this
> erroneous code instead:
> 
> :- pred square_list(int :: in, int :: in, list(int) :: out) is nondet.
> square_list(0, _, []).
> square_list(N, M, L) :-
>     member(X, 1 .. M), square_list(N - 1, M, L0), L = [X | L0].

This is a common error in writing Prolog code, too.  The key point is
that each clause of a predicate must be a true statement about the
predicate you are defining for all values of the variables in the
head.  The second clause is not a true statement when N=0, because it
makes L a non-empty list.  The key is to remember that the second
clause of a predicate may apply even if the first one does.

If you prefer the multiple clause notation, you can fix it by adding a
N>0 goal.  You might write:

    square_list(0, _, []).
    square_list(N, M, [X|L]) :-
        N>0, member(X, 1 .. M), square_list(N - 1, M, L0).

-- 
Peter Schachte              I do not feel obliged to believe that the same
schachte at cs.mu.OZ.AU        God who has endowed us with sense, reason, and
www.cs.mu.oz.au/~schachte/  intellect has intended us to forgo their use.
Phone: +61 3 8344 1338          -- Galileo Galilei 
--------------------------------------------------------------------------
mercury-users mailing list
post:  mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------