On 6/12/11 8:48 AM, Patrick Browne wrote:
On 12/06/2011 10:43, MigMit wrote:
I fail to understand why instantiating a four-argument class with five
arguments seems obvious to you.
class (Surfaces v o, Paths a b (v o)) => Vehicles v o a b where
Obviously I am wrong! But my incorrect thinking is as follows:
Surfaces takes 2 arguments, Paths take 3 arguments (giving 5).
I do not know how to group those 5 arguments to make the required 4 for
Vehicles. The original classes were defined in [1].
Well, the third argument to Paths is defined as being (v o). So, given
the first two arguments to Vehicles, we implicitly know the third
argument to Paths already (as well as the arguments to Surfaces).
It may be more helpful to think of typeclasses as functions which take
type parameters and return proof of the implementation of whatever
functions and associated types are in the class (colloquially refered to
as the type class's "dictionary"). So, given the following:
class (Surfaces v o, Paths a b (v o)) => Vehicles v o a b where...
We're defining a function "Vehicles" which takes four parameters: v, o,
a, and b; and returns a dictionary (Vehicles v o a b), or rather a kind
of proof that such a dictionary could be returned upon request.
Any implementation (aka instance) of this function (aka class) will call
the function "Surfaces" with the arguments v and o in order to get the
(Surfaces v o) dictionary. It'll then call "Paths" with the arguments a,
b, and (v o) to get the (Paths a b (v o)) dictionary. Finally, with
those two prerequisite dictionaries in hand the instance will package up
whatever definitions the instance gives for the members of the typeclass.
--
Live well,
~wren
_______________________________________________
Haskell-Cafe mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell-cafe
