Duncan Oleg's recent message to Haskell-Café tackles your real problem, but I thought I'd just say what is going on with your apparent bug below. Here's a much simpler example: class Foo a b where to :: a -> b fro :: b -> a -- foo :: forall b1 b a. (Foo a b, Foo a b1) => b -> b1 foo x = to (fro x) Without the type signature, the program typechecks "fine", and GHCi reports the type in the comment. But if you uncomment the type signature, typechecking fails. Why? It's because the type is really ambiguous. 'fro' converts from b to a, and 'to' converts back... but the 'a' is entirely lost in the process. It's very like the classic (show (read x)) example. So the type foo :: forall b1 b a. (Foo a b, Foo a b1) => b -> b1 is really unusable, because there is nothing to constrain 'a'. When you add the type signature, GHC can't establish the connection between the 'a' needed as the intermediate type in the (to (fro x)) composition and the 'a' in the type signature. You might argue that GHC should reject foo in the first place, and you'd be right, really. But it's hard in general to reject exactly and only the ambiguous cases (given functional dependencies and all that jazz), and I have not worked hard on that. But I grant that it's rather counter-intuitive that adding a type signature that GHC itself deduces goes wrong! You're right to count it a bug. Simon | This was posted to the Haskell list earlier, | http://www.haskell.org//pipermail/haskell/2004-June/014286.html | but just to record what appears to be a typechecker bug. | | I have a function with this type: | buildUArray' :: (Ix i, MArray (array s) a (ST s)) => | (i,i) -> (i -> a) -> ST s (array s i a) | | Then I ask ghci (6.2.1 and 6.3 recent cvs) the type of the following | function: | | buildUArray bounds f = do arr <- buildUArray' bounds f | unsafeFreeze arr | | ghci says: | buildUArray :: forall a i b s array. | (MArray (array s) a (ST s), Ix i, IArray b a) => | (i, i) -> (i -> a) -> ST s (b i a) | | However if I add this annotation back in the script, ghc complains: | | Could not deduce (MArray (array1 s) a (ST s)) | from the context (MArray (array s) a (ST s), Ix i, IArray b a) | arising from use of `unsafeFreeze' at BuildArray.hs:25 | Probable fix: | Add (MArray (array1 s) a (ST s)) | to the type signature(s) for `buildUArray' | Or add an instance declaration for (MArray (array1 s) a (ST s)) | In the result of a 'do' expression: unsafeFreeze arr | In the definition of `buildUArray': | buildUArray bounds f | = do | arr <- buildUArray' bounds f | unsafeFreeze arr