Andre Pang wrote:
Just because you can encode the OO idioms in Haskell doesn't mean it's particularly straightforward to use them. As your example shows, getting the syntax right for these OOish constructs isn't easy (not to mention verbose), and even so, the type errors you face when you get things wrong are, well, long :).
This is true enough... but it really isn't as dificault as it looks. Once you get used to the style it is really quite easy - and notice how you don't need class definitions, or types for the objects - it is all derived by GHC. This is an advance over current OO languages.
I guess my point is that in theory, Haskell can support OO right now. In practice, it's something that isn't very tasty.
I find it no harder than writing with monads for example... certainly there are some tricky things going on in both... but that doesn't stop people using monads for IO, state etc. Syntactic sugar over the top for instance and implementation definitions is something we are working on (using template-haskell) - so that end of things can certainly be made neater for the user. The big problem I guess is error messages - and that would require some user defined way of throwing compile time errors. Keean.