See also: IRC log
<dbooth> GlenD: Goals of processor conformance: Allow someone to point to the spec and complain if someone else is non-conformant. Also to have a product stamped "WSDL 2.0 Conformant".
We're discussing adding a way to mark in WSDL the difference between a server requiring a feature and actually engaging the feature.
I.e. A server can require a feature but then not use it.
A client can choose whether or not to engage a non-required feature.
Suggesting adding some guidance (not a marker).
Glen: Hard to do that without adding more confusion.
Umit: Client always wants to recieve messages in an encrypted fashion. Not a WSDL problem.
Glen: Has to be out of band agreement.
DBooth: This is what I wanted to warn about. If there's an optional extension, the client must be able to indicate (in-band or out-of-band) whether to engage that extension.
Marsh: So a client can't tell just from looking at a batch of WSDL whether a required feature will be engaged by the server.
Glen: No, but individual features (e.g. security), can specify how or whether a feature will be engaged by the server, and teh client can rely on that.
... This guidance would be great as a note or a blog, but doesn't seem like it should go into the spec.
... Like best practices and patterns of using TCP.
Everyone likes DBooth's definition of node.