[Springframework-developer] Spring command-driven framework

[<sam...@ma...>]

This might be a little too RCP specific, but I'd like to avail myself of the
knowledge here. As you may be aware I've been working on a spring-driven command
framework view a view to integrating it into the RCP project. Based originally
on my attempts to properly implement Spring into XWork, its a fairly simple
framework. You define Command objects in your applicationContext, then bundle
them all up as Proxies using a single BeanNameAutoProxyCreator (outlined here:
http://www.magpiebrain.com/archives/000187.html).
Anyway, I have a few design issues:

1.) Using a ValidatorRegistry and supporting ValidatorInterceptor, I can apply
validation rules to specific commands easily enough. The fact that I'm using a 
BeanNameAutoProxyCreator however opens the developer up to the danger that they
might forget to apply the ValidationInterceptor even if defined. Is the best I
can do to safeguard against this just to document it well?

2.) Ideally, I want to completely decouple GUI interaction from the Command into
an XWork-like Result object. The Command executes, and depending on the return
value the correct Result object gets executed. Again I'm thinking of using an
Interceptor to grab the result and execute it. Does this make sense to you? 

3.) So far, none of what I have done apart from the creation of a
SwingActionFactory (which creates javax.swing.Action classes to invoke Commands)
is actually dependent on the use of a rich client - it would as far as I can see
work equally well for a web interface. This is unsurprising given that much of
the design was inspired by XWork, which forms the basis for WebWork. Whilst the
focus of this project is certainly to support the RCP project, is it worth me
trying to keep the design abstract enough to support the use of web clients?

Anyway, I hope to have an example Swing GUI using the Command framework that
people can download and have a play with quite soon.

sam