Thank you for continuing to provide feedback. > It just happened to me that if the goal is support for polymorphism (as > suggested by the javadoc on > SimpleRequestProcessor.ServiceLayer#getClientType), then maybe the > getClientType method should take a Class<?> argument that the returned > object should be assignable to
I have thoughts along the same line there. Good call on the bulk property manipulation. > I'll also note that the RequestSecurityProvider is gone without any > replacement (or I missed it), which leads to the question: how about the > use case that Stephen Haberman described in the above-linked Wave? is it > still supported? will it be? (if not in this patch but a future > iteration)) The plan is to make ReflectiveServiceLayer accept a helper object to provide those kinds of domain-level decisions. I'm waiting until all of the low-level questions that RSL needs to answer are settled before extracting a more developer-friendly helper API. There will be an option to totally replace the ServiceLayer as a nod towards completely isolating SimpleRequestProcessor, but that's intended mainly for improving the test infrastructure. -- Bob Vawter Google Web Toolkit Team -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
