On Thu, Mar 6, 2008 at 3:57 PM, ChrisK
Okay, I get the difference.
The "T a" annotation in "val :: T a)"and "val :: T a" does not help choose the "C a" dictionary. But the "val :: a-> T a" and "val (undefined :: a)" allows "a" to successfully choose the "C a" dictionary.
val :: T a fixes "T a" but does not imply "C a". (undefined :: a) fixes "a" and does imply "C a". I now see how the functional dependency works here (which I should have tried to do in the first place -- I should have thought more and relied on the mailing list less).
"class C a b | a -> b" is here "class C a where type T a = b". So only knowing "T a" or "b" does not allow "a" to be determined.
Am I correct in thinking this would have worked if it were an
associated type instead of an associated type synonym?
ie,
class C a where
data T a
val :: T a
--
Dave Menendez