> Date: Mon, 1 Mar 2004 14:32:34 +0100
> From: Theo van Veen <[log in to unmask]>
>
> The question is do we support unpaired response/request messages.
And the answer is no (for SRW, which rests on SOAP, which tells us
what the answer has to be).
> A 1.2 version client may send a request for an operation which is
> unsupported for a 1.2 server but unknown for a 1.1 server. So
> clients have to deal with unpaired response/request messages anyway.
Nope. If a v1.2 client connects to a v1.1 server, then they will
negotiate v1.1 (as the highest supported version common to both) and
the client will not send a newInVersionOnePointTwoRequest. (In a
proper stateful protocol, that would be enforced by init-time
negotiation; in SRW, you discover it using explain.)
In the less interesting case that v1.2 client talks to a v1.2 server
and makes a fooRequest for some operation foo that the server doesn't
support, then that server must send a fooResponse saying "I don't do
foo".)
> Another example could be that a searchRetrieveRequest with the
> "scanOnFail" flag set responds with a scanResponse when a search
> fails.
That would be incorrect.
> The alternative would be to use extraResponseData for the results.
That would be correct.
_/|_ _______________________________________________________________
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ "The sensational story-teller does indeed create uninteresting
characters, and then try to make them interesting by killing
them. But the intellectual novelist yet more sadly wastes
his talents, for he creates interesting characters, and then
does not kill them" -- G.K. Chesterton.
--
Listen to my wife's new CD of kids' music, _Child's Play_, at
http://www.pipedreaming.org.uk/childsplay/
|