I apologise in case this doesn't make any sense. ghc -fglasgow-exts -fallow-undecidable-instances allows constructs which amount to lambda abstraction over types. I've written a small untyped lambda calculus interpreter in the Haskell class/instance sublanguage, just to prove this point. (The terms are Haskell *types*.) Using this interpreter as a basis, and adapting it as necessary, I can now write arbitrary "functions" [1] that take types to types. But of course it's a pain to write them like that. Maybe, after all, it does make sense to have a dedicated lambda-over-types Haskell extension which is convenient to use? [1] - I cannot produce an equivalent of

type Foo a b c = <some-function-over-types> a b c

but I can do an equivalent of this:

fun :: T1 -> T2 -> T3 -> <some-function-over-types> T1 T2 T3

-- anatoli __________________________________________________ Do You Yahoo!? Yahoo! Movies - coverage of the 74th Academy Awards� http://movies.yahoo.com/