[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Reverse engineering WSDL a bad idea
RA Team, I heard a lot of talk this morning about the potential to reuse what industry has developed in terms of SOA-implementation standards, for example WSDL and the WSDL data model (1.x, 2.0, or otherwise) as a possible means to model the Service Description (SD) of the RA. Isn't that an XML Infoset? Frankly, I think that is a bad idea. What industry is lacking is a technology inert reference architecture. We all know SOA is not new and that XML-based Web services have been buzzing for the past few years. In the not too distant past, we had OMG CORBA IDL and Microsoft IDL as an example of an implementation language neutral service description model. Today, XML-based WSDL is arguable the most popular implementation-language neutral way to document service descriptions/service contracts, but who knows what's going to take its place tomorrow. We should just as easily be able to map CORBA IDL or Microsoft IDL SD onto our RA SD as we can a W3C WSDL SD or a Java Interface SD. The RM provides the basic skeletal structure for us and we should start our modeling activities from the RM as a basis and start putting some "meat on the bones." After a first iteration, we should conduct a test to see if, to first order, we can map a set of technology-specific SDs such as WSDL, IDL, Java Interface, etc. onto our SD model for V&V purposes. If we do not do this, there is little likelihood that the RA spec that we do finally come up with will actually get adopted and used by the stakeholder community we are targeting for the RA. Incidentally, we need to start by defining the static model for the SD and not a dynamic model. The dynamic aspect on use of the SD will come later when we model interaction. My two cents... - Jeff E.