Anne Thomas Manes weighs on the REST vs WS-* debate:
Developers should really only need to be concerned with a handful of the specs: SOAP, WSDL, and XML Schema–maybe WS-MetadataExchange. Everything else should be handled transparently by the toolkits. I think we’ll get there.
That’s exactly what got me so excited about WS-* in the early days. Delegate the wiring problems to the platform, let the application deal with the business problem. I figured it will take some of the pain of low-level HTTP calls if we can move up the stack.
Instead, WS-* went the route of inventing even more problems that need to be solved. After all, it only makes sense that if you can solve one small problem with tooling, you can solve them all and by just throwing more tools at the problem. Let’s see what else we can solve …
Real life works quite differently. Every solution has weak links, the more complex it is, the more weak links it has. That’s the inescapable nature of software. But if it relies exclusively on tooling, there’s nothing you can do to route around problems. Worse, you’re now at the mercy of the vendors, and the more problems they have to solve to keep up with the growing stack, the longer the release schedule.
It only takes one weak link to break a chain. You just have to wait for the next major release of the pliers to arrive.
Sam Ruby brilliantly picks one line from that post:
re: “everything is automatically implemented behind the scenes”