[m-rev.] For review 1/2: Beefed up the pqueue implementation.

[Julien Fischer]

On Tue, Nov 26, 2013 at 4:15 PM, Peter Wang <novalazy at gmail.com> wrote:

> On Sun, 24 Nov 2013 15:42:14 +1100, Paul Bone <paul at bone.id.au> wrote:
> > +
> > +:- pred pqueue.merge2(pqueue(K, V)::in, pqueue(K, V)::in, pqueue(K,
> V)::out)
> > +    is det.
> > +
> > +pqueue.merge2(empty,                   B,     B).
> > +pqueue.merge2(A at pqueue(_, _, _, _, _), empty, A).
> > +pqueue.merge2(pqueue(_, K, V, L, R),   !.PQ at pqueue(_, _, _, _, _),
> !:PQ) :-
> > +    pqueue.merge2(L, !PQ),
> > +    pqueue.merge2(R, !PQ),
> > +    pqueue.insert(K, V, !PQ).
>
> pqueue.merge2(A, B, C) :-
>     (
>         A = empty,
>         C = B
>     ;
>         A = pqueue(_, _, _, _, _),
>         B = empty,
>         C = A
>     ;
>         A = pqueue(_, K, V, L, R),
>         B = pqueue(_, _, _, _, _),
>         merge2(L, B, C0),
>         merge2(R, C0, C1),
>         insert(K, V, C1, C)
>     ).


I suggest s/merge2/merge_2/ or s/merge2/do_merge/.  The library tends to use
names of the form <<name>>N (without an underscore) for non-auxiliary
predicates
that operate on N of something (or have N accumulators).

Cheers,
Julien.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20131127/8d740945/attachment.html>