I'd definitely concur with this.

I recently had a teleconference with a group that was trying to understand how the Semantic Web could be used for services discovery - i.e., applying RDF, etc., to areas that people tried originally to cover with UDDI. In trying to lay this out, one of the first things that I did with them was to move the discussion away from discovery of services and process - which XML and the web are both notoriously poor in handling - and instead moving towards the discovery of resources and the importance of RESTful services and approaches in application development.

For some reason, this is an exceptionally hard sell to both programmers and business people, I suspect because both groups are commissioned to think in terms of process flow and workflow. Yet if you can get people to understand the value of modeling resources first then simplifying the interfaces to semantically neutral publishing operations, it's surprising how readily you can get past the political BS.

One thing I did realize recently in RESTful services models, however - sometimes resource coupling is unavoidable. The classic model for this is the notion of posting an XML model to a validating collection (i.e., one that applies a server-centric validation test) still needs to provide some mechanism for communication at a level beyond HTTP numeric codes. It occurred to me that if you were to attach an action (say an XQuery script) to the appropriate REST URL (such as creating a message logging the PUT interactions to a given service that can be downloaded under separate process), this makes it possible to keep the publishing semantics clean while simultaneously letting you handle necessary coupling in a transparent manner.

My belief is that we're only just beginning to understand the best practices model for working effectively with REST based systems.

Kurt Cagle




--
Kurt Cagle
Managing Editor, http://xml.com
O'Reilly
kurt@o...