I've currently got a pattern like the following happening in my code (this is simplified from the actual code which is a large GADT dealing with expressions).
data I = A | B | C | D
type family TC1 (a :: I) :: Constraint where
TC1 'A = ()
TC1 'C = ()
TC1 _ = ('True ~ 'False)
type family TC4 (a :: I) :: Constraint where
TC4 'B = ()
TC4 'C = ()
TC4 'D = ()
TC4 _ = ('True ~ 'False)
data T (t :: I) where
T1 :: (TC1 t) => T t
T2 :: T B
T3 :: T t
T4 :: (TC4 t) => T t
etc
castL :: (Typeable t) => [Maybe t] -> Maybe t
castL l = case catMaybes l of
(x:_) -> Just x
[] -> Nothing
mkT1 :: Typeable a => Maybe (T a)
mkT1 = castL [cast (T1 :: T 'A), cast (T1 :: T 'C)]
A few questions:
- Is there a nicer way to specify a constraint that always fails instead of
('True ~ 'False)? mkT1looks a bit hacky but it does seem to work. Is there a nicer way of doing it?- Is there a better way to structure the whole thing (I understand this is a tricky question as I can't include all the detail of what I'm trying to achieve in a reasonably length post, but I'm basically generating a parse tree).