[clean-list] uniqueness question

[John van Groningen]

Edsko de Vries wrote:
>While all of that is true, it still does not explain why all occurrences of a type variable need the same attribute. For instance, there is absolutely no problem with
>
>second :: *a .a -> *a
>second x y = x

I think you are right, but I am not expert in type systems. I also
don't like this restriction.

However, I doubt that removing the same attribute restriction will
improve the type system much. For example, if someone would
define a function 'f' with type

f :: a -> *a | Eq a

he would probably expect that the typechecker will accept:

g x = f (f x)

but because '*a' cannot be coerce to 'a' this is not allowed.

This might be more confusing for a programmer than a simple attribute
restriction.

The main problem with this restriction seems to occur in combination
with overloading. For example, it prevents class definitions like:

class T a where
	g :: a -> *a

Maybe this can be fixed by allowing only instances of types that
can be coerced (something similar is implemented for the Array class).

Kind regards,

John van Groningen