I would like to encourage you to go ahead with this Ta-Da List.

While it is a poor example in support of a need for an MVC implementation (Steven’s point) it is a good example of a simple MVC implementation (Sho’s point?) that would be educationally valuable to us numerous non-purists.

When done, perhaps you can use it to illustrate the simplicity of adding another use case and through it the need for MVC itself.

For starters, I’m going to take you and Romain up on his excellent suggestion of looking at the work that you have already done to rework FlexStore using Cairngorm.

Second, this is not really a competition or anything (t-shirt aside), and I don’t think it makes sense to come up with predefined metrics. Rather, the metrics are what you, I, and others individually want to express in terms of what we consider to be good architecture. This is just a different way of having the conversation.

For example, if your point is that real world applications have complexities which are well served by having a controller layer and you can think of a way of illustrating that through this example, that would be great.

If you need to make the scenario more complex to illustrate this, that would work as well. For example, in a real world list product, it might grow to let you move list items to other people, track percentage complete and dependencies between tasks, watch for database access collisions, or any number of other complexities. You could describe the more complex app, and sketch out how you would architect a portion of the application.

In any case, I think we all agree that there is no definition of “best architecture”. I just thought it would be helpful to get concrete in order for us to talk about how different architectures get you different benefits.

The Cairngorm Store application, as Romain discussed, is an exemplification I feel of how an MVC application with domain model objects can be reimplemented with Controllers, Commands, Business Delegate, Value Objects and ModelLocators.

If I can find the time, I’m certainly willing to blog as to how we’d address implementing the todo application with Cairngorm, but if your metric of “best architecture” is fewest lines of code or fewest architectural tiers, then you win already.

So what are the criteria for success here ?

We could discuss how we’d implement the above using an MVC microarchitecture (I’d happily discuss how to do this in Cairngorm) but it’d be an academic exercise in understanding the framework, not a justification of where the framework brings in significant benefits. There’d be a leap of faith in accepting how this would scale, but it’d be very easy to argue that we were “making the simple complex”.

When I introduce MVC microarchitecture to development teams, I want to enable them to successfully build applications like online self-service banking applications, not todo lists.

Appreciate your thoughts.