Traversing graphs

[Tony Rogvall]

25 maj 2005 kl. 15.50 skrev Vlad Dumitrescu:

> Hi all,
>
> I just hit a problem where I clearly see how simple it would be  
> with lazy evaluation. But since we don't have that, I wonder if  
> anyone could point me to some non-lazy algorithms.
>

By working a bit, lazy evaluation can be done... (WARNING do not use  
in real apps ;-)

For example using the macros:

-define(delay(X), fun() -> (X) end).

-define(force(X), force((X))).

force(F) when function(F) -> F();
force(V) -> V.


You can define a lazy algorithms (can of course only be used in this  
context)

zeros() ->
     [0 | ?delay(zeros())].

ones() ->
     [1 | ?delay(ones())].

twos() ->
     [2 | ?delay(ones())].

seq(Start) ->
     [Start | ?delay(seq(Start+1))].


sieve1(L, P) ->
     [N | T1] = ?force(L),
     case N rem P of
         0 -> sieve1(T1,P);
         _ -> sieve([N | ?delay(sieve1(T1,P))])
     end.

sieve(L) ->
     [P | X] = ?force(L),
     [P | ?delay(sieve1(X, P))].

primes() ->
     sieve(seq(2)).

Regards

/Tony



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20050525/de582853/attachment.htm>