- From: LeVan,Ralph <levan@oclc.org>
- Date: Tue, 23 Apr 2002 15:07:35 -0400
- To: "'Mike Taylor'" <mike@tecc.co.uk>, barbara.shuh@nlc-bnc.ca
- Cc: azaroth@liverpool.ac.uk, Theo.vanVeen@kb.nl, Kevin.Gladwell@bl.uk, www-zig@w3.org
Mike, you're scaring me! > Yup. Not only that, but Rob's observation applies to CQL too, which > after all intends to model the Type-1 query quite closely. You'd just > use CQL qualifiers with the appropriate util, bib2 or dclib > qualifiers, mixed and matched as required. Something like -- > > util:creator,bib2:personal,dclib:agentDetail=Kernighan We've not agreed to any syntax like that and the sense at the ZIG meeting was to not allow it. But, what that would mean in SRW and what I think you think it means are two different things. I think you mean for that to be a combination of attributes from multiple Attribute Sets to generate an access point. We don't do that in SRW. We use indexes which are externally predefined combinations of attributes from any attribute sets. The only possible interpretation of that query in SRW is that you want Kernighan as either a util:creator or a bib2:personal (huh?) or a dclib:agentDetail (huh?). Sort of like author,subject=levan might mean that you want to find levan as either an author or a subject. (But it could be author AND subject which is why we're not going to support it.) Sorry if I misunderstood. Ralph
Received on Tuesday, 23 April 2002 15:07:45 UTC