It is rare that software that re-uses an interface description will have fewer constraints than the description.
While I agree with most of the post, I’m not sure about this particular paragraph. I believe once services are re-used, it’s unavoidable that a service contract will offer both more operations (if you have something like that) and data that the re-using application actually needs.
Jacek, I had to ponder this for a while, but I believe I did not misunderstand him. Consider a service implementation that corrects zip codes in addresses. It has constraints for the address only. Can I pass an invoice document to it, and have it correct billing and shipping address for me?
Depending on my infrastructure, I may be able to deploy the existing address-correcting implementation as a service implementation for correcting invoice addresses. It’ll have less restrictions in its implementation than the schema suggests.
Right?
Stefan, I believe you are in agreement with DaveO - he says most services will be more constraining than just the interface description, and you say that the interface description will have more options (fewer constraints) than the service.