- From: Henrik Frystyk Nielsen <henrikn@microsoft.com>
- Date: Wed, 6 Nov 2002 10:15:47 -0800
- To: <noah_mendelsohn@us.ibm.com>, "David Fallside" <fallside@us.ibm.com>
- Cc: <xml-dist-app@w3.org>
Would it be possible to qualify the MUST requirement on part 1 to indicate that one can implement, say, only a SOAP sender? That is, we state the intent in terms of "if you do X then you MUST do all required parts of X"? Maybe something like this: <HFN_revised_proposal> 1.@@ Conformance In order for an implementation to claim conformance with the SOAP 1.2 specification, it MUST implement all mandatory ("MUST") requirements expressed in Part 1 of the SOAP 1.2 specification necessary to implement the functionality needed by that implementation. That is, an implementation designed to be a SOAP sender MUST implement all mandatory requirements to SOAP senders, a SOAP message MUST fulfill all mandatory requirements to SOAP messages and so forth. An implementation MAY implement any number of the Adjuncts specified in Part 2 of the SOAP 1.2 specification. (Note there is no conformance associated with the convention for describing features and bindings [ref]). The implementation of an Adjunct MUST implement all of the mandatory requirements expressed in the specification of the Adjunct to claim conformance with the Adjunct. The conformance of any other technologies allowed within the scope of the SOAP 1.2 specification is out of scope of the SOAP 1.2 specification. It is recommended that such technologies provide their own conformance rules. SOAP 1.2 is designed to enable at least the usage scenarios described in SOAP Version 1.2 Usage Scenarios [ref], and possibly other scenarios. Informal descriptions showing XML representations of concrete SOAP messages used in some common scenarios are provided in SOAP Version 1.2 Part 0: Primer [ref]. </HFN_revised_proposal> Comments? Henrik Frystyk Nielsen mailto:henrikn@microsoft.com >This appears to rule out sender-only implementations, for >example, since >many of the MUSTS relate to receivers, or at least that's how >I parse this >sentence. I still think we need to say something that makes >clear that >when implementing a particular format (SOAP envelope, >encoding, etc) or >aspect of the specification (sender, receiver, etc.) that you >MUST never >fail to do something that the specification says you MUST do. >That's not >the same as saying you MUST have code to do lots of things >that you know >won't come up. For example, a trivial (if not very useful) conforming >responder is one that always faults with an indication that it >does not >understand the request. One step up is a responder that never >understands >or processes any headers, always giving a fault if any >targeted at it are >mU. And so on.
Received on Wednesday, 6 November 2002 13:16:20 UTC