Describing RESTful services using WSDL 2.0

I've spent a few minutes last night reading this "how to" guideline from IBM, I have to admit that for a while I was affraid that somebody was going to come up with this idea but I really thought that we, as a community, were pass this stage of understanding the fundamental concepts of REST. This is REALLY REALLY bad. In my opinion, the first assumption that is wrong is to assume that RESTful services can be described using the same artifacts that we use to describe traditional Services. I agree that WSDL 2.0 is a HUGE improvement over WSDL 1.1 but it is still a RPC (in the broad sense of the term) type language with constructs like interfaces and operations which makes it less than ideal for modeling RESTful services which are based on a completely different domain with artifacts like resources, relationships, URI templates, etc.

The traditional way on which we use WSDL for SOAP-based service has caused more problems than benefits on large enterprise projects and it's definitely a very BAD idea to extrapolate such practices to RESTful services. There are just a few scenarios (like governance, versioning, etc) on which the use a description language for RESTful services provides some practical benefits, for those WADL is a way simpler and better alternative than WSDL.

No Comments