Hi Simon, Thanks for pointing where can I find more info. I've taken a look at the DatabindingRuntimeWireProcessor and the only thing I'm trying to clarify is how tuscany finds out that a String array is actually hiding behind the Object[] argument when invoking the wire and decides to use JSON (even in the case when other databindings are included). Considering that databinding occurs between the service invocation from the binding and the actual service, I believe source contract type, operation, data types are taken from:
wire.invoke(operation, args); I don't see how operation can be different and I'm not aware of a mechanism for determining what types exactly hide under Object[]. I'm currently building a jsonp sample that will include the various ways of calling jsonp services (json objects, jquery, etc.). It will be good for testing and if you consider it good enough, for the samples/ directory too. Florian On Tue, Aug 31, 2010 at 12:25 PM, Simon Laws <simonsl...@googlemail.com>wrote: > Hi Florian > > On Tue, Aug 31, 2010 at 9:59 AM, Florian MOGA <moga....@gmail.com> wrote: > > Hi Ant, > > It seems I've missed Simon's commit and didn't update jsonp binding > sources > > at the time I was checking them out... That was the problem I wanted to > > bring up. I've checked up the new getJSONRequestStringArray() method and > > I've spotted that the typesIndex is never incrementing... Can't estimate > > what's the impact of not having quotes for all string objects in JSON and > if > > this is handled the right way by the databinding layer, maybe Simon can > give > > us more details? > > It may just be an error. Let me take a look. > > > Looking at the code, I find the abstraction very useful and I would like > to > > use this databinding layer for the comet binding too so we can have > > conversions done in one place. All that's required for parameters is to > pass > > the string array as a Object[] type and tuscany will take care of > everything > > else? How is it working actually? Are the objects passed through another > > layer that takes care of conversions before making the service call? > > Yes, where the type of the interface contract at the binding and the > implementation differ the Tuscany runtime (see > DatabindingRuntimeWireProcessor in modules/core-databinding) adds a > DataTrasnformationInterceptor to the wire. On the service side this > interceptor converts the type of the data coming in from the binding > into the data types expected by the implementation. If you look at the > separate databindings you can see the individual transformers, for > example, in databinding-json you can see a JSON2JavaBean transformer > which, well, converts a JSON object into a JavaBean. The same thing > can happen on references too but in the opposite direction of course. > > The configuration and processing of databindings is all a bit > complicated as it's very flexible. It's also complicate as it deals > with wrapping/unwrapping data types as appropriate depending on the > style of WSDL that's used to describe an interface. The best way to > understand it is to: > > - take a look at Raymond's presentation on the subject [1] > - debug through a running example > > > I can also see that the tuscany-databinding-json artifact is used in > > binding-jsonp-runtime only with test scope. Does this mean every project > > using jsonp binding needs to declare tuscany-databinding-json as a > > dependency as well or tuscany will fail json conversion? Wouldn't that > try > > to convert all the objects passed in the service calls made in the > > composite? > > We should make this a runtime dependency > > ...snip > > > [1] > http://www.google.co.uk/url?sa=t&source=web&cd=5&ved=0CCgQFjAE&url=http%3A%2F%2Fsvn.apache.org%2Frepos%2Fasf%2Ftuscany%2Fsandbox%2Fold%2Fdoc%2Ftuscany.databinding.ppt&ei=Kcl8TJ2aEMyNjAeYte3TDg&usg=AFQjCNHXxhplX1P3r6OA8IoJXhWAsM39uA&sig2=1521aoFfCJ9FcWOMeydH_A > > I just found the above link via a Google search. I'm sure it's > referenced from the web site but I can't find it. I'm hoping Raymond > can point us at it. > > Simon > > -- > Apache Tuscany committer: tuscany.apache.org > Co-author of a book about Tuscany and SCA: tuscanyinaction.com >
