Proposal: Don't require users to use undefined

David Menendez dave at
Fri Oct 29 18:02:55 EDT 2010

On Fri, Oct 29, 2010 at 5:43 PM, Ian Lynagh <igloo at> wrote:
> On Fri, Oct 29, 2010 at 10:41:38PM +0100, Ian Lynagh wrote:
>> On Fri, Oct 29, 2010 at 05:25:15PM -0400, David Menendez wrote:
>> >
>> > >    map ord "foo" == #map Char Int ord "foo"
>> >
>> > I guess this would only work with explicit foralls, otherwise how do
>> > you know which type is the Int and which is the Char?
>> You can do it based on the order they first appear in the type.
> Actually, maybe it's not so simple once MPTCs, FDs and ATs are mixed in.
> But if nothing else, requiring explicit foralls would work, anyway.

There's also alphabetical order.

What about stuff like foo :: Int -> forall b. b -> b? Currently, that
gets treated as foo :: forall b. Int -> b -> b, but maybe that's not
desirable with explicit type arguments. (I guess that's an argument
for having a different syntax for type application.)

Dave Menendez <dave at>

More information about the Libraries mailing list