Print

Print


> Date: Thu, 18 Dec 2003 09:43:52 -0500
> From: Ray Denenberg <[log in to unmask]>
>
> > My preference would be to have a mandated CQL version for every
> > SRW version.
>
> What defines the srw version in use?  Remember, the version
> parameter for srw doesn't say what version is being used, it says
> what version the client wants in the response.

Nope, that's not my understanding of what we agreed.  In discussion in
DC we repeatedly mentioned the way HTTP handles versions as an
analogue.  When an HTTP client (web browser) sends a request, the
"1.1" is not saying "please give me a response in 1.1, it's saying
"this request is 1.1".  And that is what an SRW packet's version
number is saying, too: "I am an SRW v1.1 client (and so my CQL is of
the version that is tied to SRW); do not respond according to a more
recent version of the protocol, I won't understand your response."

So there's no problem here.

> So let's say a client formulates an srw request using version 1.2,
> and says 1.1 in the version parameter.

That client is broken, and should not be spoken to politely :-)

> By "mandated" we mean "only", I assume. Thus (for example) for srw
> 1.1, only cql version 1.1 is ever used (and the fact that they have
> the same version string is coincidental). Thus if there becomes a
> cql 1.2, then 1.3, then 2.0, srw 1.1 would still only ever use cql
> 1.1.  Then let's say we move to srw 1.2 and we say that it goes with
> cql 2.0, same thing, srw 1.1 may only ever be used with cql 1.1 and
> srw 1.2 may only ever be used with cql

Yes, exactly.

 _/|_    _______________________________________________________________
/o ) \/  Mike Taylor  <[log in to unmask]>  http://www.miketaylor.org.uk
)_v__/\  "In spite of everything, I still believe people are really
         good at heart" -- Anne Frank.

--
Listen to my wife's new CD of kids' music, _Child's Play_, at
        http://www.pipedreaming.org.uk/childsplay/