undecidable-instances

When writing some code using UndecidableInstances earlier, I ran into something that I found very odd. I managed to unintentionally create some code that typechecks when I believed it should not: {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE MultiParamTypeClasses #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE UndecidableInstances #-} data Foo = Foo class ConvertFoo a b where convertFoo :: a -> b instance (ConvertFoo a Foo, ConvertFoo Foo b) => ConvertFoo a b where convertFoo = convertFoo . (convertFoo :: a -> Foo) evil :: Int -> String evil = convertFoo Specifically, the convertFoo

By the time I first read serious criticism on -XUndecidableInstances , I had already completely accustomed to it, seeing it as merely removal of an annoying restriction Haskell98 has to make compilers easier to implement . In fact I've encountered plenty of applications where undecidable instances were needed, but none where they actually caused any problems related to undecidability. Luke's example is problematic for quite a different reason class Group g where (%) :: g -> g -> g ... instance Num g => Group g where ... – well, this would clearly be overlapped by any proper Group instance, so