http://www.pbell.com/index.cfm A series of occasional musings on architecting, securing, optimizing and generating web based applications. By Peter Bell. en-us Tue, 18 Jun 2013 20:27:29 -0400 Wed, 15 Nov 2006 22:01:00 -0400 BlogCFC http://blogs.law.harvard.edu/tech/rss test@test.com test@test.com http://www.pbell.com/index.cfm/2006/11/15/A-RESTlike-Administrative-API Most of the websites I create require content management, with pages that define a context for multiple dynamic content areas, so on the whole a <a href="http://en.wikipedia.org/wiki/Representational_State_Transfer">REST</a> style interface with unique URIs for individual content items and actions on them aren't of much use. However, for the associated administrative system for managing content items, a REST-like API (not religiously RESTful - see my <a href="http://www.pbell.com/index.cfm/2006/11/15/An-Appropriate-level-of-REST">earlier post</a> for an explanation of why) would solve a bunch of problems. It would make it easy to define the URIs for any given action in a consistent way across applications and would also allow for the testing and calling of ALL model methods using a single, very simple controller. Over time I could also add support for multiple response formats so this could be the basis for everything from lightweight scaffolding to a remote API for Flex applications. The first step was to come up with the necessary URI templates for defining access to both single objects and collections of 0..n objects (comparable to the difference between a DAO and a Gateway). I decided on different templates for the two to make it clear whether you were operating on a single object or a collection . . . [More] REST Wed, 15 Nov 2006 22:01:00 -0400 http://www.pbell.com/index.cfm/2006/11/15/A-RESTlike-Administrative-API http://www.pbell.com/index.cfm/2006/11/15/An-Appropriate-level-of-REST I've been thinking a lot about URLs and URIs recently. <a href="http://en.wikipedia.org/wiki/Representational_State_Transfer">REST</a> is a <strike>religion</strike> approach that focuses on using unique URLs for interacting with content items. REST purists will tell you that you only need <a href="http://www.pbell.com/index.cfm/2006/7/14/Verb-Hunting-Four-Just-Isnt-Enough">four verbs</a> (corresponding to CRU and D) and that to get the caching/performance benefits of REST, all requests must be completely stateless (no cookies, sessions or other user specific identifiers - if you want personalized content you should create one URI per user for that content item). Realists will point out that trying to hack together a real world system with those limitations (especially one that needs more than just GETs) isn’t worth the trouble and that a "REST-like" approach makes more sense. In case you hadn't guessed, I'm a "REST realist". <a href="http://jimonwebgames.com/articles/2006/06/26/dont-say-crud-say-fucd">David Heinemeier Hansson</a> hasn't been able to convince me that you only need four verbs. Anyone else out there using a strict four-verb syntax successfully for a complex system? I get that you could turn additional verbs into "undiscovered objects" but just don't see the point. Any thoughts? REST Wed, 15 Nov 2006 21:48:00 -0400 http://www.pbell.com/index.cfm/2006/11/15/An-Appropriate-level-of-REST