Hi Rob, Thanks for the contribution and for exploring better ways to integrate with GWT.
I've updated the issue with this info: http://restlet.tigris.org/issues/show_bug.cgi?id=127 Concerning the support of a Restlet client API for GWT, I'm eagerly waiting for the release of GWT 1.5 (with support for most Java 5 language features) to port/recompile the Restlet API on GWT. Looking at GWT-REST, it seems that our API is significantly richer than the one they have and that developers would benefit from an homogeneous solution for browser and JVM layers. I'm also open to finding ways to merge both efforts. For example, Jon R. Crosby could lead the development of this GWT-Restlet version. Rob, if you think this is a good idea, please feel free to start a discussion with them on this topic. See also this related RFE: "Add Javascript client" http://restlet.tigris.org/issues/show_bug.cgi?id=129 Best regards, Jerome > -----Message d'origine----- > De : Rob Heittman [mailto:[EMAIL PROTECTED] > Envoyé : vendredi 4 janvier 2008 06:13 > À : [email protected] > Objet : GWT in ServerServlet > > > I had a configuration problem in my earlier post; the > previously attached ServerServlet works fine, other than a > couple transient declarations I left out. I attached it to > RFE #127, which I think it makes a big dent in. The good > news from my perspective is that this better approach can > land in a big deliverable we have this weekend ... :-) > > > A note on why this unfortunately ugly composition approach is > necessary: > > Tom McGee's solution (posted here: > http://www.mail-archive.com/[email protected]/msg0262 > 1.html ) only works if you are mounting Restlet under hosted > mode at a particular location (e.g. /api/v1/* in the > example). If you want Restlet to have full control of the > URI space from /*, a solution like this is needed, since the > GWTShellServlet will only work properly if mounted at /*, and > fixing that limitation is not a high priority for the GWT > team. My use case is wanting to run/debug an entire > application -- GWT client and (fairly complicated) > Restlet-based server -- from an Eclipse project. > > > A neat (and sometimes necessary) feature: > > Restlets can tell whether or not they are running under GWT > Hosted Mode by interrogating the "module" configuration parameter: > > public boolean isHostedMode(){ > return > getContext().getParameters().getFirstValue("module")!=null; > } > > P.S. Jerome, I know you will want to rename the "module" > parameter something else, but GWTShellServlet wants it named > "module," so renaming it would require extra care and > monitoring that I didn't sign up for :-) > > > Next steps in meaningful GWT integration: > > Jon R. Crosby's GWT-REST project, > http://code.google.com/p/gwt-rest/ is discussed here as a > GWT-Restlet adapter in response to a post by Justin Stanczak > to the GWT Google Group: > http://groups.google.com/group/Google-Web-Toolkit/browse_threa d/thread/353f76f8662d778c/1fc5e2a3ddb9cad9?#1fc5e2a3ddb9cad9 > > GWT-REST provides a class library for the REST vocabulary > and, as the creator states, is general enough to work with > Restlet. It's early and very heavy on speaking to JSON and > Rails, but nice. From a few posts I read, I don't think he > did it with any reference to prior art in the Restlet high > level API, so the differences may be more accidental than > intentional. > > I'd like to see some dialogue opened with the GWT-REST > project to see if we couldn't somehow converge its high level > REST API with Restlet -- or if not, provide the Restlet-style > high level API in a form that will compile neatly to GWT's > browser-hosted Javascript. I'd hate to outright compete with > what GWT-REST has done, but I don't want to speak two > different but eerily similar Java REST APIs in different > aspects of the same project. > > - Rob > > >
