Wednesday, March 17, 2004

Charles Miller comments about the identifiers that are tacked on the end of URL's in most continuatin based web frameworks. The main problem, which I agree with, is it causes problems when sending links to other people. With standard sites like Amazon and Google you can send a link to a friend and they will see pretty much what you were looking at. They'll even be able to tell from the URL what they are going to. In continuation based frameworks you tend to have one URL which a continuation identifier tacked on the end and all URL's look the same. Avi has the answer, you provide the ability to do both. I do the same with my framework. When you register a function to be called when the URL is accessed you can assign an optional path to it. This path gets added to the URL between the base application path and the continuation identifier. If the continuation has expired then the path is examined and the application can continue from the appropriate point. This works with the login process as well. When the request is made, if the session has expired, or something else indicates it may be a different user or machine making the request, the login is requested. A successful login takes the user to the resource they originally requested. This type of thing is actually very easy to do in a continuation based framework. A check on the validity of the session that fails calls a dialog component to show the login screen which returns on successful login. 9:51:44 AM