>From: Alberto G. Corona <[email protected]>
>Sent: Tue, February 22, 2011 3:03:56 PM
>
>Recently I had to navigatate trough data structures chained with mutable
>referenes in th STM monad. The problem is that their values are enveloped in
> Either or Maybe results.
...
>to summarize, given:
>
> foo, : a -> m (Maybe b)
> bar : b -> m (Maybe c)
> baz : c -> m (Maybe d)
>
>how to compose foo bar and baz? Or, at least, Are something out there to
>handle
>
>it in the less painful way?.
Control.Monad.Trans.Maybe.MaybeT
from mtl is defined as
newtype MaybeT m a = MaybeT { runMaybeT :: m (Maybe a) }
with (among others), instance Monad m => Monad (MaybeT m)
>I solved the generalized problem (chaining any double monadic combination)
>with a sort of monadic connector that acts as a " double monadic" operator
>>>>>== so that
>
>return. return (x :: a) >>>>== foo >>>== bar >>>== baz
>
>can be possible. Although I don't know if it is the best solution. I wonder
>why
>
>nobody has written about it before:
>
>class (Monad m, Monad n) => Bimonad m n where
> (>>>=) :: n a -> (a -> m(n b)) -> m(n b)
The standard construction uses a distributive law - if
you have an operation
distribute :: n (m a) -> m (n a),
then you can make m around n into a monad by
m (n (m (n a)))
== by fmap distribute =>
m (m (n (n a)))
== by join (for m) =>
m (n (n a))
== by fmap join (n's join) =>
m (n a)
Here, that's a function Maybe (m a) -> m (Maybe a), which can be
distribute (Just ma) = fmap Just ma
distribute Nothing = return Nothing
Brandon
_______________________________________________
Haskell-Cafe mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell-cafe
