[nem-en] Questions about lazy evaluation

[mei]

Hi,

There is a Haskell code.

>| tarai.hs : Haskell sample
main = print (tarai 20 10 5)

tarai :: Int -> Int -> Int -> Int
tarai x y z = if x <= y then y
                        else tarai (tarai (x-1) y z)
                                   (tarai (y-1) z x)
                                   (tarai (z-1) x y)
|<

Run very fast above code. Because Haskell is lazy evaluation.  Then I
notice that Nemerle support lazy evaluation.  I try to convert it to
Nemerle.

>| tarai.n : Nemerle sample
using System;
using System.Diagnostics;
using Nemerle;
using Nemerle.IO;

def tarai(x : LazyValue[int], y : LazyValue[int], z : LazyValue[int]) :
LazyValue [int] {
    def a : int = x;
    if (a <= y) {
        y
    }
    else {
        tarai (
            tarai (lazy(x-1), y, z),
            tarai (lazy(y-1), z, x),
            tarai (lazy(z-1), x, y))
    }
}

def x = 20;
def y = 10;
def z = 5;

def s = Stopwatch ();

s.Start ();
def a = tarai (lazy(x), lazy(y), lazy(z));
s.Stop ();

printf("tarai(%d, %d, %d) = %d - %ld\n", x, y, z, a, s.ElapsedMilliseconds);

|<

But this is very slow.  Can I write faster code like Haskell ?


Thanks.

-- 
mei <mei at work.email.ne.jp>