Hi John...

Keep in mind that COM does'nt even support inheritance.....< bg >

With regard to using Access/Assign - I am on the fence. On one hand, I am totally in Dave F.'s camp. Its automatic, its clean, it works. I do believe however that Access/Assign methods are actually slower than manually created Get/Set Methods. The speed hit though, is negligible I am sure. On the whole, Access/Assign Methods are pretty cool and are the way to go.

Jacobson's argument I think lies in encapsulation - which is where a lot of debate occurs. How much encapsulaton is enough? How much should you rely on coupling between objects. At some point, you need to build something and get it working. You can spend years on the theory - and not build squat. This is why many OO books are dangerous.

You know, OOP is really a religion, not a Comp. Sci. disipline...

There is no one correct solution.



>I was recently rereading Ivar Jacobson's book "Object-Oriented Software Engineering" which is considered a seminal work on OOAD, OOSE and OOP. I came across a statement in which he writes "objects should not directly manipulate the properties of other objects". Later, in context, he is shown to mean that this applies to both reading and writing properties.
>
>Microsoft's COM theory more-or-less says the same thing: objects expose interfaces but not properties.
>
>So....are we reversing central tenets of COM and OOP by using Access and Assign to intercept property calls/puts when there shouldn't be any direct manipulation at all?