When reading examples across the Internet (including the MSDN reference) I have found that code examples are all doing the following type of thing:
public class FooViewModel : BaseViewModel {
public FooViewModel(FooController controller) {
Controller = controller;
}
protected FooController Controller { get; private set; }
public void PerformSuperAction() {
// This just routes action to controller...
Controller.SuperAction();
}
...
}
and then for the view:
public class FooView : BaseView {
...
private void OnSuperButtonClicked() {
ViewModel.PerformSuperAction();
}
}
Why do we not just do the following?
public class FooView : BaseView {
...
private void OnSuperButtonClicked() {
ViewModel.Controller.SuperAction();
// or, even just use a shortcut property:
Controller.SuperAction();
}
}
1
Consider the model being just data. It doesn’t make choices, and it doesn’t know how to render.
- A product can be a model.
- A list of products can be a model.
- A bundle containing the list of products, the cart and the information about the current user is a model.
The last model from the list corresponds to the data you need to render the page, but it shouldn’t know how to render itself, because it’s not the task of the model.
In order to render it, a view is used, given that the controller may pick different views depending on the context. It will use for example a view to render data to an ordinary guest visitor, a different view for an authenticated user, and a third view if the user is an administrator. Same data, different modes of displaying it.
Consider this a blend of two basic MVCs (this assumes you know the basic MVC).
First, deeper MVC pair is:
- M = model
- C = controller
- V = view model
Second, upper MVC pair is:
- M = view model
- C = collapsed / minimal / interspersed (
OnSuperButtonClickedis part of it) - V = view.
With this separation, of course view from one triad cannot access model, nor the controller, from other triad.