I am using Client-Activated objects, well-known mode=singlecall, as I don't have to worry about multithreading issues, and I also need my objects to maintain state. They are developed based on OO principles, and do not only serve to execute self contained methods that do not effect state.

I do have to include the entire dll in the client app which allows it to recognize by objects - but it only reads the interface, and I can change the dll on the server as long as the changes do not affect the interface that the clients are using (such as adding parameters to an existing public method that is being used by the client).

I am new to remoting, and I determined the above from what I could find and read, so please let me know if I am on the wrong track!

But I don't see why I can't have parameterized constructors. Without it, I would force UI developers to call an invoke method immediately after they create the empty object - and this leaves opportunity for failure.