Monthly Archives: December 2004

REST as model

Werner likens REST to physics, in that each is a model of some reality, not the reality itself. He writes;

Whether we use the REST model, or another model to be developed that appears to match it closer or from a different perspective, “the web” and other large scale distributed systems will continue to do “their thing”, whatever model we put on it. The distributed, decentralized, bottom-up, autonomous nature of the web, exhibits complex organic interactions, that are not driven by models or laws, just as that Nature is not driven by the laws of Physics.

Well said.

I would just add though, that there’s also a “metamodel” in play here that shapes our models; software architecture (and while Roy’s view is just one of several, the other views aren’t that different). Of course, this too is a model, and so falls under the domain of the same principle. But I suggest that that so long as this metamodel remains useful, most models of the architecture of the best behaved parts of a future Web, will be REST extensions, like ARRESTED or the (bulk of) the Semantic Web.

Greg Goth article

I was fearing this article being published for a while. The interview went terribly as you can probably tell; I was sick, and had about 4 hours of sleep the night before. Plus Greg was asking what I considered to be all the wrong questions, which irked me, and forced me to try to steer to the conversation to where I felt it should be going … with mixed results. I also interpreted his questions as having a strong Web services bias, which also bothered me. But to his credit, the article came off as reasonably well balanced.

Savas picked up on the article, and the praise I had for MEST in it. I did want to say one thing about that though, that I didn’t have a chance to say in the interview. While I like MEST as an architectural style, I believe that the MESTful software that Savas and Jim would write would have considerable issues integrating with the Web and other Internet based apps, largely because they also believe in protocol independence. On the other hand, since their assumed ProcessMessage semantic is practically semantically identical to HTTP POST (and even SMTP DATA), a considerable amount of that software may integrate well purely by accident! 8-) It’s only when they start using – and trying to be independent of – non-ProcessMessage like semantics, such as HTTP GET, PUT, or FTP STOR, that the integration problems will arise.


In case you hadn’t noticed, I’d not been doing much travelling for the past couple of years. In fact, my drive to D.C. last month was the first time I’d left the confines on the Canadian border in about two years (in contrast to the 350K+ miles accumulated in the previous three!). This was due to some issues I was having with the Canadian government, in turn due to a case of mistaken identity; apparently there’s some guy in Canada with my name and birthday(!!) who’s not quite so law-abiding as yours truly!

I thought I was doing the smart thing in 2003 by applying for my citizenship instead of a permanent resident card. But, in retrospect, that turns out not to have been such a good decision. So I finally applied for the card in addition to my citizenship, and I received the card yesterday. Woohoo!