[Haskell-cafe] Portability of MonadError

[Henning Thielemann]

On Mon, 5 Jan 2009, Luke Palmer wrote:

> Oh bother!  My new year's resolution: think before I speak.
> 
> While I do think this is the right answer, it is not the right answer in the status quo.  This is
> because ErrorT e m is only a monad when e is an Error, which Either (and most types) are not.  It
> will be the right answer when fail is factored out of Monad into MonadFail (which will happen
> someday hopefully), because then the typechecker will verify that the above composition cannot
> fail.
> 
> But now I can't think of a good answer.  Darn.

In the explicit-exception package I omit 'fail' and allow an exception 
type without constraints.