Re: Add voidM to Control.Monad

On Mon, Jan 16, 2012 at 10:05:25AM -0600, Antoine Latter wrote:

2012/1/16 Jurriën Stutterheim :

...

It is indeed intended for monads that have no Functor instance. While one could argue that every monad should have a Functor instance, this is not the case in practice. If, on the other hand, Monad was defined as

 class Functor m => Monad m

then all would be fine and we would be able to use the current implementation for `void`. A Functor constraint on Monad is an entirely different discussion though.

We would be better off adding Functor instances for the Monads in question.

IIRC, Functor rather than Monad was chosen for void's constraint as all Monads could/should also have Functor instances. The ticket was http://hackage.haskell.org/trac/ghc/ticket/3292 and it links to the discussion. Thanks Ian