Enrique Rodriguez ha scritto: > Message objects and codecs are generally trivial though they do > represent some work. I could help out with refactoring any messages > and codecs to new conventions and it sounds like Rich is, too. It > could go pretty quickly and it sounds like there are unit tests > available. > > Also, we could start with dnsjava and slowly migrate to whatever our > thinking is about these objects, to prevent getting bogged down in a > bigger project.
+1 > The hard part is in the protocol workflow, be it on the client side or > the server side. Michael mentioned some use-cases for the server-side > and we'll have to keep that in mind but there are techniques for > making things flexible and extensible. A lot of what has been lacking > on Directory in this regard is feedback from users who drive us to add > extension points, remove deps, etc. In which case, it sounds like > getting this going at MINA and with a focus on a client will benefit > everyone. > > Regarding client-side workflow, the use cases for Kerberos (my focus) > are similar to DNS in that it is a means to an end. You acquire > Kerberos credentials before accessing a subsequent service. I am > interested in what patterns emerge for handling multiple, disparate > protocol calls since I am in the middle of multi-protocol client work > on Directory these days. I haven't hit it yet, but now that I think > about it, I'll have the same problems in DNS that everyone on this > thread is having. Cool! I hope we can find a pattern that fits MINA so that it will be easy to integrate multiple protocols using MINA. Stefano
