I very much like Carlos Perez’s writing in general, and link to if often; this time, though, I’ll (naturally) have to disagree with his assertion that SOAP is dead. There’s a great analysis from Chris Ferris which concludes:
The real crux of the issue is the inadequacy of the various programming models used for processing XML, whether it be POX or SOAP, and the tooling which developers use to “simplify” the task of cutting code. They are either considered “too complicated for average developers”, or they are too simplistic (or constrained) in their underlying assumptions (most XML->domain object mapping).
Amen to that. I am convinced that if you want to do application-to-application communication using XML over HTTP, you’ll ultimately end up with your own SOAP variant. There’s much to say against XML Schema, and WSDL, and the pseudo-standard WS-* situation; but SOAP, to me, is very much alive and deserves to be.