Thanks to both fast answers. there remain problems with Jakes mail for me. This:
When you define datatypes, you are essentially defining a type-level constructors on the left hand side and (value-level) constructors on the right hand side.
is very useful for me. "data" defines TWO constructors, ok. And if i want construction on the type level, then the arguments must obviously be type-"valued", means parameters. From this i conclude, that data ClockTime Integer Integer = ... would never make sense, whatever on the right size. The next isn't understandable for me - i have not the slightest conception of dependently typed languages. Then i arrive at
. Now, let's say we had tried defining ClockTime with parameters as you suggested.
ClockTime' :: Integer -> Integer -> *
Do you see the problem? In order to use the ClockTime type constructor, we would have to use Integer values.
Cannot see any problem here - do we NOT want ClockTime to be initialized by two Integers ? Or is this the main reason for introducing "TOD" - to be able to change it without having to make any changes to code using ClockTime ? To repeat myself - am i right understanding, that this needs a differently named data constuctor ? (I cited "abstract type" from the library reference. Not important for me at the moment, what that means in Haskell.) Thanks for your attention, Joost