> Date: Tue, 23 Dec 2003 10:19:44 -0500
> From: "Ray Denenberg, Library of Congress" <[log in to unmask]>
>
> I want to know how people feel about how we should state what srw does and
> does not require. For example, in the result set section:
> http://www.loc.gov/z3950/agency/zing/srw1-1/result-sets.html
>
> "SRW does not require the support of persistent result sets that may be
> accessed by a client in subsequent requests. It does require the server to
> state whether or not it supports them, ....."
>
> I think the "does not require" part is fine. However I would prefer
> (second sentence) "It does require..." changed to "It does expect...".

I think the current wording is fine. A server that does not state
whether or not it supports persistent result sets is not an SRW
server.

> Because I'm not sure where this "requirement" would (or whether it
> should) be expressed.

Right there in the spec! You just quoted it!

> I would like srw to continue along the lines of publishing a "base
> profile"
> http://www.loc.gov/z3950/agency/zing/srw1-1/base-profile.html
> and avoiding a formal "conformance" section (the base profile will
> serve effectively as conformance in most cases, that's what
> customers will point vendors to, but for some who want to implement
> less than the base profile, we don't have to deal with the
> philosophical question of whether they're doing srw or not). And
> this is basically what we agreed to at the September meeting.
>
> Is everyone comfortable with this?

I don't understand why it's preferable to specifying what is and isn't
conforming behaviour. I can't see the advantage to the world in
having a situation where there are servers out there and no-one's
really sure whether what they're doing is SRW or not.

My personal preference: no conformance section, no base profile, just
the specification itself. If a server does what specification says,
it's an SRW server, if it doesn't it's not. Anything else just
confuses the issue.

 _/|_ _______________________________________________________________
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ "Who needs to worry about a 10^-15 chance of an MD5 collision
         when the chance of the programmer screwing up seems to be
         about fifty-fifty?" -- Mark-Jason Dominus.

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