Re: [Haskell-cafe] Closed Classes

Notwithstanding module Y, I don't think you should infer in module X that g::Int (or, rather, Int->Int).

Oops, I meant Int->Int.

Since f is defined in a type class, it should be polymorphic, and so should g. When you apply g to [an expression of a given] type, it will check to see what instances are available, and match only if Int is the type of the variable. But there's no reason to restrict the type of g itself.

Sorry I wasn't clear: I meant to be talking about (an approach for) "closed classes".