I'm a bit puzzled by a sentence in the current spec:
conformance can be achieved with other functional profiles also
which is rephrased in #72 as
conformance to any profile that conforms to the Abstract Model fulfills conformance to this specification
The rephrasing in #72 implies that there are actually two different kinds of conformance involved here:
a functional profile conform with the abstract model
an implementation conforms with a functional profile
This makes a lot of sense, but then this should be explicitly defined in Section 2 Conformance.
Once this is explicit, the notion of "conformance to this specification" used in Section 8 becomes too vague to be useful, so Section 8 should probable be rephrased.
This fine-grained notion of conformance would, IMO, be much more useful in terms of interoperability. Currently, a client implementing cnpr:http and a server implementing cnpr:qsa can both claim "conformance with DX-CONNEGP", and yet they will not be interoperable.
I'm a bit puzzled by a sentence in the current spec:
which is rephrased in #72 as
The rephrasing in #72 implies that there are actually two different kinds of conformance involved here:
This makes a lot of sense, but then this should be explicitly defined in Section 2 Conformance.
Once this is explicit, the notion of "conformance to this specification" used in Section 8 becomes too vague to be useful, so Section 8 should probable be rephrased.
This fine-grained notion of conformance would, IMO, be much more useful in terms of interoperability. Currently, a client implementing cnpr:http and a server implementing cnpr:qsa can both claim "conformance with DX-CONNEGP", and yet they will not be interoperable.