A brief thought on WS-RF

My friend Jim from “down under” told me today over messenger that my blog is “more of a social diary than a blog”. So, here’s a thought on WS-RF with which I think he will agree…

I think from now on the focus of the discussion is going to be on the conceptual model. WS-RF promotes a way of thinking in which an explicitly exposed resource (like a disk drive!!! :-), its state, and its identity are tight-coupled with the interface (the Web service) (hmmm… smells like an object… could it be an object? :-). A WS-Addressing instance containing some local-to-the-service information is seen by the WS-RF authors as an opaque structure because it plays the role of an identity AND it is a way to do message correlation. Ok, there is nothing wrong with the use of the underlying WS technology. I see a potential problem with the conceptual model which may encourage tight-coupling, since it promotes a way of thinking for building distributed applications where resources and their state are explicitly exposed outside a service’s boundaries.

The discussions (WS-CAF mailing list, OGSI-WG mailing list) on the use of WS-Context from WS-CAF versus the use of WS-Addressing as the way to do message correlation continue. I even contributed my thoughts after Mark Little’s request. BTW, Mark’s messages to the two lists contain very good arguments. Well said Mark.