BOb,

>Well... I tell you ... it is quite the article.>

Yes it is...Steve does a good job of breaking down different approaches and explaining the pitfalls and benefits of each. I recommend reading it through a few times! Steve also handles the arguments against subclassing well.

What appeals to me about the hook approach is that you can create a hook object whose functionality can be reused anywhere. In contrast, when you place code in a particular subclass (as in the i-layer) you only get reuse in that particular hierarchy. However if you create a hook object, this can be added to many different class hierarchies providing greater reuse. This is what I meant in my previous message regarding more modern OOP techniques favoring composition over inheritance.

However, as you read in the article one pitfall of the hook approach can be the overhead of object messaging...at this point I am deliberating on a variety of approaches and will let you know ASAP what conclusions I arrive at.

Regards,