Yesterday and after I posted the sort entry with my views on MEST vs REST, Simon, Aad, Einar, and I watched the live webcast of the ACM Turing Lecture. This year the award was given to Vinton G. Cerf and Robert Kahn for their work on TCP/IP. It was an interesting lecture in the form of a discussion. I was very pleased to find out that my views on identity/addressing/naming and the issues with the Web are very similar to theirs. Robert in particular talked about the need to preserve information, using persistent identities that can span technologies and be valid throughout the years. The latter was one of our arguments against OGSI (and WSRF) in our WS-GAF paper.
Robert Kahn has been working on a the Handle System for few years now. From a quick look, it appears to be a resolution service for content which is associated with a unique name. Looks interesting.
2 responses to “The ACM Turing Lecture”
I read your WS-GAF paper, and tried to find all your auguments agaist WSRF. I cannot see big conflict between your idea in the paper and the current WSRF specifications. EPR + resource ID, to me, is a way to achieve your requirements for the resource identification.
Hey Don, the paper was about OGSI. We thought that the main issues with OGSI were: the use of standards (it was really pushing the envelope), architectural approach (object-oriented in nature).
WSRF fixed the first one and sort of introduced contextualisation on the way through the introduction of the resource ID in the EPR. However, the EPR is still treated as an opaque entity (like an object pointer). In the contextualisation approach, which has the advantage when more than 2 participants are involved in an interaction (look for an article that compares WSRF and WS-Context for more info), identification of state or the state is itself is explicit.
More importantly, however, WSRF still promotes a resource-oriented style for building distributed applications where applications are built around interactions between resources rather than services. As a result, we design applications that talk to resources directly rather than talking to services about resources. That’s the big difference I see.
However, it’s too late to do anything about WSRF now given the number of technologies that have built on top of it.
Please feel free to email me if you have more questions or if you disagree.