Using Silverlight 1.1 means I am able to write .NET code that runs in the web browser. This should make it very easy to implement a clean model-view-presenter (MVP) architecture. I am envisaging a model that acts as a web-service proxy to get data and perform actions on the server. The view would thinly wrap the XAML UI. The presenter would handle all the interactions between the other two. This makes for very unit-testable code.

The one thing I'm not sure about is what level of granularity to use regards presenters and views. In XAML it is possible to define amazing interfaces that have animations and transitions between states. Should these different UI states have distinct view/presenter objects? Would it make sense to have a single XAML canvas that implements a number of different View (OOP) interfaces?

Does anyone have good resources dealing with switching views?