On 18 mar 2008, at 13.51, Luke Palmer wrote:
On Tue, Mar 18, 2008 at 12:24 PM, iliali16 <[EMAIL PROTECTED]> wrote:
Now the problem comes here:
play (p1 :>: p2) state
|play p1 state == (i1,state1) && play p2 state1 ==
(i2,state2)
= (i1+++i2,state2)
I know that if I manage to do that function the one above with
this sign :>:
do not need to be impelmented since this one will cater for all
the cases.
Can you please help me?
You just need a nice simple let or where clause:
play (p1 :>: p2) state = (i1 +++ i2, state2)
where
(i1,state1) = play p1 state
(i2,state2) = play p2 state1
Or equivalently:
play (p1 :>: p2) state =
let (i1, state1) = play p1 state
(i2, state2) = play p2 state1
in (i1 +++ i2, state2)
And there's nothing lazily recursive about these, just the information
usage is a little more complex. But it could be implemented perfectly
naturally in scheme, for example.
For further exploration: the pattern here where the state is threaded
through different computations, is captured by the module
Control.Monad.State. So if "play" returned an object of a State monad,
such as:
play :: Logo -> State TurtleState Image
Then this case could be implemented as:
play (p1 :>: p2) = do
i1 <- play p1
i2 <- play p2
return (i1 +++ i2)
Pretty, ain't it? A little too pretty if you ask me. Let's make it
uglier and shorter still:
play (p1 :>: p2) = liftM2 (+++) (play p1) (play p2)
Or use Applicative directly:
play (p1 :>: p2) = (+++) <$> play p1 <*> play p2
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
