On Tue, Dec 16, 2003 at 02:34:36PM -0500, Ray Denenberg wrote:
> Since we  agree that cql and srw will have independent versions, how will a
> client convey the cql version?

I would have thought either a version of SRW says the version of CQL
being used, or if you wrap CQL queries in an element, that element
somehow identifies the version of CQL. I think its better (simpler) for
a version of SRW to identify the version of CQL it is using.

> For that matter if cql is used outside of srw, how is the version to be
> indicated?

The environment can define the version of CQL. I don't think a CQL string
should contain version identification.

> And the cql context set -- do we assume that its version is the same as the
> cql version?

I would have said yes - a CQL version defines the context set version.

The alternative is to have the context set independent of CQL and the
environment defines it.
In practice however I think implementations of CQL need to know the context
set - its not abstracted. For example, 'any' and 'all' are being mapped to
OR/AND trees in parsers. So CQL implementations are being built assuming
a particular CQL version and context set version. Abstraction is good, but
you don't want to over engineer things.

My personal opinions anyway - but I don't feel that strongly on these
points.

Alan