- From: Jan Algermissen <jalgermissen@topicmapping.com>
- Date: Tue, 14 Jun 2005 00:43:24 +0200
- To: Paul Downey <paul.downey@whatfettle.com>
- Cc: public-web-http-desc@w3.org, Mark Baker <distobj@acm.org>
On Jun 14, 2005, at 12:33 AM, Paul Downey wrote: > > On 13 Jun 2005, at 23:16, Jan Algermissen wrote: > >>> >>> I go to buy a book from Amazon, fill in their form, but then realise >>> i left my credit card at the office. An hour later, back from work >>> I complete the form. If in the meantime Amazon change the service >>> in an incompatible way, my request will fail. >>> >> >> Incompatible with WHAT???? >> > > If the service suddenly expects the client to send the > credit card number using a different parameter name, > or starts rejecting country codes of "uk" then that would be > an incompatible change and the client would fail, regardless > if it's a form or a generated Python stub. I disagree. Given you leave your browser when it is in some state X of the purchase order app that tells you that the param name is A and that Amazon changes the Web app to require the param name to be B. When you send the 'outdated' form (== perform the outdated state transition) Amazon can have provided for this case and silently translate your state transition to the required new one without you noticing. And it is exactly the decoupling of application and consumer that allows amazon to handle such cases and not loose a purchase. The extreme at the other end would be you having received an executable from amazon some time ago with the application semantics hard coded. You would do the purchase and after you return the purchase would just fail with Amazon telling you to doenload a new executable. Jan > > -- > http://blog.whatfettle.com > > ________________________________________________________________________ ____________________ Jan Algermissen, Consultant & Programmer http://jalgermissen.com Tugboat Consulting, 'Applying Web technology to enterprise IT' http://www.tugboat.de
Received on Monday, 13 June 2005 22:43:35 UTC