I won't link to the schema yet. Too many outstanding issues and
inconsistencies. I'll be away next week and when I get back, hopefully you all
will have it all resolved!
Among the issues: I would strongly oppose adding a database name. The database
name in ZNG is analogous to the host in Z39.50 and we don't specify the host in
the request. This was a huge simplification that I hope we don't back out of.
Regarding number of records in response -- I presume that was a clerical error.
The rest of these you all can argue over. --Ray
> I have some problems with this schema.
> There is a DatabaseName in the request. This is unnecessary. The target
> for the request is the database.
> NoOfRecordsInResponse is unnecessary. I never liked it in Z39.50 Classic
> The ResponseSchemaID should not be necessary. You've already determined the
> response schema by the time you encounter the element.
> Similarly, the record schema ID needs to be part of the record.
> ResultSetID needs to be a structured element if we are going to include a
> TimeToLive value.
> StartRecordNo in the response is unnecessary. It is either StartRecordNo
> from the request or 1 if none was specified in the request.
> ResponseDiagnostic needs to be structured with a Z39.50 integer code and
> addinfo. I wouldn't mind adding an optional human diagnostic, but I think
> we need to be sure to add an explicit language indication. The human
> diagnostic probably needs to be repeatable.
> No ZNGRequestID, please.
Library of Congress
[log in to unmask]