>Larry,
>
>I no difference between 1) a classlib of interface classes or a prg of DEFINE CLASS commands setting them up and 2) a whole pile of AddProperty() calls. If anything the AddProperty() makes it more obscure because it's likely to be done at the point where the creator object instantiates the interface object and thus makes it much harder for the second object to know what the interface is at design time.

The difference here is the limits I place on myself when using this technique.

I never use this technique when there are a "pile" of properties that I have to create. I use AddProperty() when the custom object isn't used in more than 1 or 2 places in the code. Otherwise it becomes problematic when trying to change things. Also, it becomes counterproductive when there are more than a few of properties. If you only need an array property and 1 or 2 others then this technique is fairly simple.

WRT your comment about the second object not knowing the interface, I don't buy that. You can't program without knowing the interface. If you are talking about looking at it in another class designer as you are programming it in the 2nd object, then that's valid. But it's the same idea as opening up notepad (or Word) and reading the interface specification from your design notes.