> Date: Fri, 12 Dec 2003 14:38:37 -0500
> From: "Ray Denenberg, Library of Congress" <[log in to unmask]>
>
> Has there been any discussion of versioning for context sets and
> record schemas?

Nope.

> For example the dc index set listed at
>       http://lcweb.loc.gov/z3950/agency/zing/cql/index-sets.html
> (version 1.0 of srw/cql) has identifier
>       http://www.loc.gov/zing/cql/dc-indexes/v1.0/
> implying version 1.0, but does what does that mean -- that it's the
> dc set for version 1.0 of srw/cql, or that it's version 1.0 of the
> dc set?

Very, very definitely the latter.  Proof by example: if we decide we
made a horrible mistake and omitted the "title" element from the DC
context set, then we need to change it and re-release as version (say)
1.1.  We _definitely_ don't want to require all of SRW to bump up a
version number because of that!

Context sets, and schemas, and extraRequestData elements for that
matter, must be allowed to develop at their own page, and to describe
that development with their own version numbers.

> And we don't even know for sure that srw and cql versions will be in
> lock step.

I am pretty sure they won't.

> At http://www.loc.gov/z3950/agency/zing/cql1-1/context-sets.html the
> dc set (as currently drafted) has identifier
> http://www.loc.gov/zing/cql/context-sets/dc/v1.1/ implying version
> 1.1. But it's the same definition.

I think that's a bug.  If the set hasn't changed, there's no need to
change its version number (and its potentially misleading to do so.)

> Note that zthes still indicates 1.0, presumbably because it's the
> 1.0 version of the context set (or it's the context set
> corresponding to version 1.0 of zthes).

The former.  The Zthes profile is at version 0.5.

> I suggest for cql (the cql context set) call it version 1 (v1), same
> for dc (unless we decide to change it, and if so call it version 2),
> and for bath, version 2.

Why all these suggestions?  Version-numbering of context sets is
properly the responsibility of their maintainers, without reference to
what other standards (e.g. SRW) happen to refer to them.

> I'm not suggesting that we not change the identifier, since we're
> abandoning v1.0 compatibility, but we could call it
> http://www.loc.gov/zing/cql/context-sets/dc/v1/ to signify that it's
> still version 1 of the dc set.

Since the DC set is under the LoC namespace, that's up to you as the
LoC representative.  But I think that faceted version numbers (1.0,
1.1, 2.0 etc.) are more usual.

 _/|_    _______________________________________________________________
/o ) \/  Mike Taylor  <[log in to unmask]>  http://www.miketaylor.org.uk
)_v__/\  "All that I know most surely about morality and obligations,
         I owe to football" -- Albert Camus.

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