Geert Bevin is founder of RIFE, a full-stack, open-source Java Web application component framework. In this tech talk, given at JavaOne in May 2006, Bevin discusses various features of RIFE - in particular, continuation support and RIFE's templating syntax. Other topics include:

• Continuations' effects on the Servlet model
• The relevancy of RIFE as a benefit for a team
• How RIFE's component model differs from other similar frameworks
• How RIFE/Crud works
• What application types RIFE is best suited for
• Unique features

What do you think of RIFE?

If you want to see what RIFE feels like when you're developing with it or what particular features look like, you can check out the video tutorial series that I published last week: http://rifers.org/blogs/gbevin/2006/10/6/new_series_of_video_tutorials

Hi Geert, Do you can compare your framework with other ones with similar target? For example with: - MDA based frameworks. - OpenXava - Model driven POJO based frameworks: Naked Objects or Roma Framework What do you think about these frameworks? Which are the diffences between RIFE and these frameworks? Which the disadvantage of RIFEs over these frameworks? Thank you Javier Paniza

Honestly Xavier, I can't compare them since it would mean that I stop spending time developing the framework itself. There are so many out there that appeared since we started in 2000 that I only take a look at the most popular ones. However, since you know all about OpenXava, I suppose you're the best to answer that, all you need to do is listen to the techtalk and comment how things compare to your framework. From having a quick look at Roma and Naked, I'd say that RIFE has been designed from the ground up as a full-stack framework, it's not an aggregation of existing ones that people try to get to work together. All the core features (the one you see in the diagram on the homepage at http://rifers.org), are present inside one 2.6MB jar. That's all you need to upgrade, there's nothing to configure. In RIFE, the model-centric approach with a non-invasive pure-Java meta programming facility, is also not the icing on top, it's been developed inside every individual layer from the beginning. RIFE/Crud for example is only a jar of 60KB since the actual work is already done by the core of RIFE itself. This means that even if you're not using any of the auto-generation functionalities, you still get a lot of meta-programming features at your disposal inside your own code. Also, we really took a critical look at most of the 'standard' approaches to do web development (state handling with continuations, templating or components for example) and implemented them in an alternative and, in our opinion, better way. You can still continue to use any of the usual libraries, if you prefer. However, having a unique approach for every single feature gives the framework a lot of value since they all really solve a unique set of problems.

brilliant summary of a brilliant framework; related link is the interview on artima a few weeks ago: http://www.artima.com/lejava/articles/rife.html -- eokyere :)

Quality replica watches.