On Wed, 30 Jul 2003, Matthew J. Dovey wrote:
> MD: Yes - also I would suggest that a 1.1 server should avoid sending
> any 1.1 elements unless it recieves one in the request (or other
> confirmation that it is dealing with a 1.1 client)
Then lets put an explicit version parameter in. Without sessions we can't
tell one time a client connects from another, so servers would constantly
need to check off their list of 1.0 elements, 1.1 elements, 1.2 elements
... to calculate which type of response to generate.
At which point we don't need the namespaces.
What about CQL? When we go to 1.1 we'll also want to go to CQL 1.1 with
the non point relation modifiers. If we don't have explicit versioning,
then the query will still say srw1.0 but will use CQL 1.1
With XCQL, would we then need to check the /contents/ of the element to
see which version it is? (Urgh!)
> What about SRU, which doesn't have any convenient place to put this
> information? It's just out of luck?
> MD: I am assuming that any new 1.1 feature (e.g. scan, xpath, record
> encoding etc.) involves at least one new element or parameter in the
> request, so that the presence of such can be used for the server to
> identify whether it should send a 1.1 or 1.0 response - this may need
> more thought?
I don't think that's a valid assumption. ExplainRequest doesn't change
anything for SRU or SRW, but the response is different. The contents of
query and xQuery will change with the additions to CQL.
If we use Ralph's model for XML responses, then recordData will contain
a string in SRW and raw XML in SRU. Not saying that I agree with that (I
prefer recordPacking and recordXML) but it's another case where semantics
change for the same element across versions.
,'/:. Dr Robert Sanderson ([log in to unmask])
,'--/::(@)::. Special Collections and Archives, extension 3142
,'---/::::::::::. Nebmedes: telnet: nebmedes.o-r-g.org 7777
____/:::::::::::::. WWW: http://nebmedes.o-r-g.org:8000/
I L L U M I N A T I