Affects Version/s: None
Fix Version/s: None
I Propose to drop the usage of square for conformance check as I think that is strongly error prone. I am not sure at all about the worthiness of using this artifact.
For starting with, I think that there are a good number of places where, if used, it is missing::
. Clause 2.2.12 Optional Input Schemas. Ther it is sayd that RefType and SchemaRefs MUST NOT be used....to me this is a requirement that has to be checked if checking conformance: if one of them is present then the element is not conformance.
. Clause 2.13, 2.14: RequestID. Also corresponds to a conformance requirement in a response
In addition to that along lots of places where there are sentences like "This XXX(an element) MUST include a XXX". To me this is certainly a requirement that a conformance testing tool should test, but it is not preceeded with a square.
I propose then to drop the usage of these squares for identifying conformance requirements.
If we do that, then the conformance requirements will be identified by the sentences that contain one of the words defined in RFC 2119.
If the word appears in sentences like "the element XXX MUST have the XXX attribute" or "the XXX attribute MUST have the following value: ...:" they are clearly conformance requirements for the format and the contents of the protocol messages.
In sentences like "For each DocumentHash in InputDocuments the server MUST perform the following steps" there is clearly a requirement on the way the server must behave but not a conformance requirement on the contents of a certain component of the protocol.