> nhc doesn't like the following program ...
>
> > type Option = Maybe
>
> > f :: Option a -> Maybe a
> > f = id
Yes, nhc98's type checker doesn't do kind inference, so the definition
of a type synonym must have the same kind as all its uses.
> The problem disappears if we change the type definition to
>
> > type Option a = Maybe a
Indeed.
Regards,
Malcolm