I think about AJAX in this way:
- Model: The server-side where the data is stored and exposed through webservices. In a way this is a model-view-controller within the larger model-view-controller (model = data, view = XML or some other parsable data structure, controller = server-side code that manipulates the data).
- View: XHTML/DOM
- Controller: Javascript/JQuery
Is this a reasonable way to think about AJAX programming?
You too, huh?
Yes. I think this is a very good way to structure edit operations.
However… You probably want to keep to traditional HTML “view” pages for many documents or objects within an application so that you can support search engines if need be, and so that deep links into pages in your application come up with something meaningful.
This way, using both approaches, you have a hybrid, rather than the “un-webby rectangle” problem where you can never bookmark anything, send links to your friends, or have search engines generate [good] links.
8/10: I guess this also means that you should think of the “server” side of your application as a “REST” service, ideally. Also, be sure not to fall into the trap of thinking the client code is in any way, whatsoever, secure (or even “obscure”, for that matter). Make sure the supporting “service calls” back to your web server have a good security infrastructure.