Your proposal looks good. It should be able to deal with most of the cases.
There might be some tricky things on the road: * How to check if two data types are compatible. For java types, we can check if two classes are the same. For XML types, we need to deal with element as well as types. * How to make a remotable Java interface into WSDL 1.1. I'm not sure if all the remotable interfaces can be mapped into WSDL 1.1 though, for example, JAX-RS annotated interfaces, RMI or EJB interfaces. Thanks, Raymond ________________________________________________________________ Raymond Feng [email protected] Apache Tuscany PMC member and committer: tuscany.apache.org Co-author of Tuscany SCA In Action book: www.tuscanyinaction.com Personal Web Site: www.enjoyjava.com ________________________________________________________________ On Jul 6, 2010, at 12:24 AM, Simon Laws wrote: > Through looking at BWS_2007 I've discovered that our contract matching > tests are not correct. The code as it stands always returns true > rather than matching operation types for remote interfaces. The TODO > in the code says that this is because it's potentially hard to match > the source and target types when the source and target contracts are > described using different IDLs. The Assembly spec states that WSDL1.1 > should be used as the IDL for matching remote interfaces of different > IDLs. > > I'm starting down the road of providing generated interfaces as > required and I proposed the following approach. > > source target match action > ===== ==== ========== > interface.java interface.java match as interface.java > interface.java interface.wsdl convert source to interface.wsdl and match > interface.wsdl interface.java convert target to interface.wsdl and match > interface.wsdl interface.wsdl match as interface.wsdl > > I have some code that works for matching endpoints and endpoint > reference but I now realize that this isn't enough as matching is > applied up and down the implementation hierarchy before endpoints and > endpoint references are generated. So I'm looking at some surgery to > make generated interfaces available within the service and reference > model objects. > > If anyone has previously started down this route and has comments I'd > be glad to hear them. > > Regards > > Simon > > -- > Apache Tuscany committer: tuscany.apache.org > Co-author of a book about Tuscany and SCA: tuscanyinaction.com
