One more thing. Just to ride my hobby-horse one more time, wouldn't you agree that everything you say in this discussion regarding compatibility also applies to interoperability between two vendor's "languages". V1 could be Vendor One's implementation of the Name Language and V2 could be Vendor Two's extended implementation. The two vendors implementation of contact functionality would be interoperable if they adopted the specifications illustrated in Figure 5. Another way to make the point about compatibility being equivalent to interoperability is just to think of versions across time (first and second) as simply verisons by different vendors (Vendor "First" and Vendor "Second").
-- Nick
Nick, you are absolutely right on the point about compatibility == interoperability. I have gone to considerable lengths to describe languages in terms of sets that are independent of production and consumption, ie V1 and V2 Accept, rather than in terms of V1 Producer and V1 Consumer.
It is very likely that many of the Consumers and Producers will have different definitions of the Language. Taking this to it's fullest suggests using a non-linear versioning, that is foo and bar and baz Accept (note that even X and Y Accept seems to imply linear versioning). Then the discussion would be around the set relationships between foo, baz, and bar and these could be Consumer 1, Producer 1 and Producer 2 or any other combination.
Previous Message
