I've come across the following scenario: register proxy, the proxy makes some asynchronous service calls but while waiting for those calls, the user changes the view in such way that the proxy is not needed anymore so we remove it from the facade. Then the response from the server comes, proxy gets the data and sends notification that it has loaded some data, although it is no longer registered with the facade. Should this behavior be possible? What scenarios would require it?

PS: I know that in an ideal world, the proxy would cancel all the calls the moment it got removed from the facade, however time constraints lead to little monsters like these...

Edit: of course its mysterious because I was not registering the proper commands on the ApplicationFacade. Thats why nothing was happening. Thank you for your time!

Heh, happens to me sometimes. 

The shell has no knowledge of what the module needs registered, although it gets registered under the same facade. That's the point.

I probably should have made clear what I needed. In the application I'm working on, the user logs in and is taken to a screen, based on his roles. It has a main navigator that can take the user to different screens, but not all users can access all the screens. And because the developing of certain screens made the application size pretty big, we decided to turn those screens into modules, and instead of having a ViewStack with all of them, just use a ModuleLoader. The idea was to have some code loaded only when it's needed, not having multiple modules loaded at the same time, in different areas of the application. So for this specific case, having the module fire his own startup command (just like the main application did) worked fine, and fast. Switching to the multicore version felt like an overkill.

I've just managed to take out a part of a project I'm currently working on, and make it compile as a Module, while still using the standard version of pureMVC framework. Reason for this is that we needed to manage our application size. Since some features are available to only some of our clients, and their size was 1/3 of application size, it made sense to just modularize them. And this was possible without switching to a multi-core architecture.
What I've done is make a startup command responsible for registering whatever elements the Module needed, including its cleanup command, and make the module send a notification trough Facade (thus avoiding a reference to ApplicationFacade) for the startup (just like the main application does). The cleanup command was then triggered by a notification sent from the main code base.
What also helped was the fact that all notification name constants were not defined in ApplicationFacade, but in a different class that had the sole purpose of defining them.
So, I can't help wondering if it's really worth it to switch from the standard framework to the multicore one. Since our Modules aren't really stand alone, but add functionality on top of the existing one, I'm really having a hard time seeing any advantages of such a change. Any thoughts on this?