Can't you trivially satisfy Eq:
instance Eq b => Eq (a -> b) where
f == g = False
John Atwood
---------------------------------------------------------
William Lee Irwin III wrote:
>
> In analysis, functions are usually taken to be an algebra over the
> reals (or C). This tempted me to define the following:
>
> module FunctionAlgebra where
> instance Num b => Num (a -> b) where
> f + g = \x -> (f x) + (g x)
> f - g = \x -> (f x) - (g x)
> f * g = \x -> (f x) * (g x)
> negate f = negate . f
> abs f = abs . f
> signum f = signum . f
> fromInt i = \x -> fromInt i
> fromInteger i = \x -> fromInteger i
>
> Of course, this fails as class Num requires its members to be of
> observable type (instances of Eq, Show). I think it's something
> interesting which Haskell's type system is apparently capable of
> expressing, even though the prelude types conflict with it. Maybe
> someone will find it interesting enough to include support for it
> in the Prelude. Of course, Open Source means if you think something
> should happen, you can make it happen for yourself. =)
>
> Bill
> P.S.: How does one contribute binaries to the binary distribution
> archive?
>
>
