According to Haskell 98 the principal type of f should be: f :: (Eq a) => a -> a -> Bool. GHC does only perform context-reduction using instance-declarations if there are no type-variables in the type. Because the type in this function is [a], ghc doesn't perform context-reduction. Ghc chooses this strategy because of extensions like overlapping instances, there remains more information in the type of a function to select an instance. For some reason GHC also applies this strategy if one turns of the extensions. Grt william kim wrote:

Hi All,

I am confused by the notion of principal type in Haskell with type classes. Consider a simple example:

f x y = [x] == [y]

GHCi yields type f :: (Eq [a]) => a -> a -> Bool.

But according to the paper "Type classes: an exploration of the design space", predicate Eq [a] should be reduced to Eq a. Is this reduction performed here? What should be the principal type of f?

Thanks.

william

_________________________________________________________________ Get an advanced look at the new version of MSN Messenger. http://messenger.msn.com.sg/Beta/Default.aspx

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Hello william, Thursday, June 22, 2006, 1:22:32 PM, you wrote:

GHCi yields type f :: (Eq [a]) => a -> a -> Bool.

But according to the paper "Type classes: an exploration of the design space", predicate Eq [a] should be reduced to Eq a. Is this reduction performed here? What should be the principal type of f?

Ghc, unlike H98, supports instances like this: instance Eq [MyType] where a==b = True so this extension to type inference allows to use such instance even if MyType is not in Eq class -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com