Jon Udell writes the gating factor is not WSDL syntax. I agree with him there In the process he identifies a specific problem and explores some of the ramifications. I agree with his analysis, up to but not including the conclusion that a deep changes is required in these languages.
Several points worth exploring based on this example:
- Interfaces change. In a local procedure call, you often have the source to the requestor and server, and can change both in synch. Some protocols, like COM, are based on the presumption that interfaces are immutable, and has this resulted in the proliferation of interfaces ending with the digit "2" or the letters "New" or somesuch. A more realistic approach for an internet protocol designed to not only interoperate across space but also time is for the protocol to be extensible (that's the "X" in XML, after all). Named parameter association and namespaces are key parts of that strategy.
- In this example, the client needed additional information to interoperate with the server. Other examples will require other information. This will occur for every language on every platform - both static and dynamic. The inevitable conclusion is that some additional metadata needs to be injected into the process. Different vendors may approach this different ways.
- However vendors do this, it would be helpful if there were an interoperable exchange format for this metadata.
- Having servers, clients, and tools that can directly produce or consume this interoperable metadata format will tend to increase interoperability across vendors, platforms, and programming languages. This is good.
10:08:12 AM 
|
|
