+1 for me, as long as there are no gotchas that you are unaware of. :)
Fabio Mancinelli wrote: > Dear all, > > this is more a RFC than a vote... > > Currently the REST subsystem is written using Restlet and its API and > there is a layer based on the XWiki Component manager written by me > that is used to declare and configure resources and components for > providing representations in a more dynamic way. > > During the last weekend I spent some time experimenting with the JAX- > RS API and as a test I tried to port the current implementation to > this API. > > I was able to do so without too much effort and the result was also a > drastic reduction in code complexity. > > In fact all the plumbing I wrote is already handled by the JAX-RS API > implementation. Actually the JAX-RS API provides also more powerful > mechanisms wrt the ones I wrote because it takes into account > representations in responses (GET), representations in requests (PUT > and POST) and exception mappings (i.e., the possibility of capturing > and representing all kind of exceptions, checked or unchecked). > > One consequence of these features is that I was able to write methods > to handle requests that use our data model without having to deal with > complicated try/catch blocks for adjusting the response wrt to > exception thrown. Conversions from objects of our data model to > representations and viceversa (text/plain, text/xml, etc.) are handled > by the framework, and exceptions as well. > > The problem was the implementation to use. I tried both Jersey and > Restlet and both of them were fine. > > Jersey misses some features such as authentication handling (I had to > wrote a basic authentication handler by hand) and is not flexible as > Restlet. Restlet, on the contrary, is very powerful and has a very > good support for JAX-RS but it lacks automatic generation of the WADL > description of the application. > > Coming to a conclusion, I would like to switch to JAX-RS + Restlet > because we can have the best of the two worlds. If we need, in fact, > we can still leverage the Restlet API if the JAX-RS is not suitable > for implementing a resource. We will have to give up the WADL support > for the moment, but I don't think this is a big deal with respect to > the gain we have in code maintainability (no more plumbing and easier > resource declaration) and reduced code complexity overall. And WADL > for JAX-RS will be eventually implemented so it's just a question of > time. > > Here it is my +1 > > WDYT? > > Regards, > Fabio > > > > > > > > _______________________________________________ > devs mailing list > [email protected] > http://lists.xwiki.org/mailman/listinfo/devs > _______________________________________________ devs mailing list [email protected] http://lists.xwiki.org/mailman/listinfo/devs
