This patch is for the java client. It doesn't change anything in the bridge.
Right now, if you try to use the local argument resolver functionality from a java application (not a browser, but another servlet or something), you can't do it. The client blows up because it can't find a method that takes the specified arguments. At least I couldn't find a way. The only way we could fix it internally was to patch the java client and pass in nulls in place of the "special" classes. Sorry for the confusion! The main thing I wanted to verify was that it would be OK to introduce a dependency on the J2EE API in the client class. Thanks! Brantley William Becker wrote: > Hi, > > I am a little confused. We already have Local Argument Resolvers on > the bridge (see JSONRPCBridge.registerLocalArgResolver() and here > <http://jabsorb.org/Manual#head-9761c7c1691368e722055f5b958c1c80725998bb>). > Does the patch you have made duplicate this effort, or is it different > and, if so, how? Furthermore if it is different, would it be best to > use our extend out existing LocalArgResolver framework so that you can > make resolve any class, instead of just the three you suggest? In this > way no dependencies would be required. > > If the functionality you require is not currently in the code, I don't > see why your patch wouldn't be accepted. > > Cheers, > Will > > On Wed, Jun 25, 2008 at 1:10 PM, Brantley Hobbs <[EMAIL PROTECTED] > <mailto:[EMAIL PROTECTED]>> wrote: > > Hi all! > > My organization is beginning to use JSON as a general protocol for all > of our applications, rather than dealing with the complexity of > full-blown SOAP web services. We recently patched the old > org.codebistro version of the Java client to allow local argument > resolver functionality to work properly on the servlet. I'm > prepared to > submit a patch to jabsorb as well, but I wanted to run it by the > developers first. My patch will introduce a dependency on > HttpSession, > HttpServletRequest and HttpServletResponse in > org.jabsorb.client.Client. > > The patch works by checking the called method parameters. If the > parameter is one of those special types, then it simply removes that > parameter from the arguments array that's passed into invoke(). > > This way you can have an interface specifying your remote object's > methods and simply pass nulls in place of the session, request, > response, etc. > > Would this patch be accepted? > > Thanks, > Brantley Hobbs > University of Georgia > > _______________________________________________ > Jabsorb-dev mailing list > [email protected] <mailto:[email protected]> > http://lists.jabsorb.org/mailman/listinfo/jabsorb-dev > > > ------------------------------------------------------------------------ > > _______________________________________________ > Jabsorb-dev mailing list > [email protected] > http://lists.jabsorb.org/mailman/listinfo/jabsorb-dev > _______________________________________________ Jabsorb-dev mailing list [email protected] http://lists.jabsorb.org/mailman/listinfo/jabsorb-dev
