I think we are saying the same thing them, for some reason the point you are making in your response did not resonate in your original post.

IMHO, the problem with OO is that it does not offer a “bi-directional” interface, therefore a class has difficulty to form a contract with its environment. We are in complete agreement that OO can be used within a service implementation.

thanks,

JJ-

I’m not sure I’d go so far to say that OO is no longer valid in networked systems. I think it is incredibly useful for implementing what happens behind the service interface, as a result of an operation being invoked.

However, I would agree that OOAD as the top level design methodology for networked systems has indeed been proven infeasible.

Also, I guess I should have made clear who was saying what in the conversation

I like a lot what you do and write in general but I beg to disagree on many of the arguments that you are making here.

Maybe this post will make it clearer: http://www.ebpml.org/blog/40.htm

Object Orientation is longer valid in “connected systems”. Trying to apply OO principles there is a dead-end, we need to shift gears and apply a “Service Oriented” paradigm, where in particular methods are not necessarily attached to “classes”. Services and Operations are very different from Class and Methods. OO is perfectly good to implement some services (not all, BPEL is a great programming language too, as you pointed you recently).

Cheers,

JJ-