> I'm happy with either of the following choices:
>
> * Class constraints on constructors have effect everywhere
> (as in Hugs).
> * Class constraints on constructors are eliminated (call it a
> typo if you must).
I'd be delighted to eliminate them, but we had a long H98 debate
about it (under John's chairmanship) and decided to keep them.
It depends what you mean by "have an effect everywhere". That's what
I thought GHC implemented! The effect you wanted is:
> It may help if I explain my motivation in introducing class
> constraints
> in data declarations. Often, a type will make no sense without the
> constraints; e.g., an association list type Alist a b makes no sense
> unless Eq a holds. The class constraints on data declarations were a
> simple way for the user to ask the compiler to enforce this invariant.
When you *construct* a value, the overloaded constructor's
type does indeed ensure that you can't build it if the invariant
does not hold.
But when you take a constructor *apart*, the invariant must hold
by construction: you couldn't have built the thing you are taking
apart unless invariant held. So enforcing the invariant again is
redundant; and in addition it pollutes the type of selectors.
So doesn't my proposal give you precisely what you want, and no more?
Simon
