Proposal: Applicative => Monad: Call for consensus

[Jan Behrens]

On Friday 28 January 2011 15:13:44 Jan Behrens wrote:
> On Sunday 02 January 2011 13:04:30 you wrote:
> > The patches attached to http://hackage.haskell.org/trac/ghc/ticket/4834
> > make Applicative a superclass of Monad. Default definitions are provided
> > for backwards compatibility.
> 
> I looked at the Prelude of the referenced documentation in the ticket:
> http://bifunctor.homelinux.net/~bas/doc/ghc/html/libraries/base-4.4.0.0/Pre
> lude.html
> 
> Would it be useful to include liftM and ap in the Prelude, so that you can
> use it as a default implementation for fmap, when defining Functor
> instances?

I have to correct myself. It should be "liftA" and "ap", as liftA is more 
abstract than liftM and can thus be used as a default fmap implementation for 
BOTH monads and applicative functors.

> Then you can write...
> 
> instance Functor Something where
>   fmap = liftM
> 
> instance Applicative Something where
>   pure x = ...
>   (<*>)  = ap
> 
> instance Monad Something where
>   (>>=) = ...
> 
> This way you don't need to write explicit implementations of fmap and <*>
> for every monad you define.

The example would read then as follows:

instance Functor Something where
  fmap = liftA

instance Applicative Something where
  pure x = ...
  (<*>)  = ap

instance Monad Something where
  (>>=) = ...

> An alternative name for liftM would be fmapMonad...
> 
> 
> There is a similar approach in Data.Traversable:
> 
> fmapDefault :: Traversable t => (a -> b) -> t a -> t b
> 
> 
> However as Functor, Applicative and Monad are part of the Prelude, it might
> be useful, to include a default implementation for fmap in the Prelude.
> 
> 
> Regards
> Jan
> 
> 
> _______________________________________________
> Libraries mailing list
> Libraries at haskell.org
> http://www.haskell.org/mailman/listinfo/libraries