- From: Walden Mathews <waldenm@optonline.net>
- Date: Wed, 12 Mar 2003 18:42:25 -0500
- To: Mark Baker <distobj@acm.org>, www-ws-arch@w3.org
> "As applied to WSA, SOA uses the first three constraints; that is, it > uses URIs to identify resources[...]" > > I believe that is incorrect. The latter part of that paragraph, and the > following one, attempt to explain at some length about how only "Direct > SOA", and not "Mediated SOA" uses URIs to identify resources. If one > derived style does, and one doesn't, then the style from which they are > both derived does not. Assuming we know something about the method used for deriving one style from another, namely that it's purely about adding more constraints to derive new styles. Although Fielding prefers that method in his REST thesis, we have no right to assume it's the preferred method here. But it would be nice to know. As far as constraints go, I'd like to see them stated clearly as constraints (formally if necessary), and that goes for the SOA language you're commenting on as well as REST; both are weak in this area. "Identification of resources" and "uses URIs to identify resources" are about equally ambiguous. What is it that's not allowed? Where is the constraint? In the case of Fielding and REST, section 5.2 explains the importance of identifying conceptual mappings, not their representations, when we identify things: "REST connectors provide a generic interface for accessing and manipulating the value set of a resource, regardless of how the membership function is defined or the type of software that is handling the request." In this context, "identification of resources" is not constraining the system against having unidentified resources; it is constraining what is identified to being *conceptual*, not *actual*. What this means for SOA is both good news and bad news. The bad news is that the architecture document is in error when it says that "mediated SOA" doesn't use URIs to identify resources. The good news is, yes it does. It does because SOA resources are "services", hence always abstract and dynamic, as distinct from data representations. It's a trivial test, as compared with some REST cases. Of course, I'm assuming that DaveO is adopting REST concepts unadulterated in his analysis of SOA, and again that's an area which could be improved by making it more explicit. Walden
Received on Wednesday, 12 March 2003 18:42:40 UTC