Re: [search-ws] propose a 'resultSetId' parameter for SRU 2.0

["Ray Denenberg, Library of Congress" <rden@loc.gov>]

This was proposed a few months ago, with no response, so I'm reviving it for discussion at Monday's call.

 

--Ray

----- Original Message -----

From: Ray Denenberg, Library of Congress

To: search-ws@lists.oasis-open.org

Sent: Wednesday, January 07, 2009 4:53 PM

Subject: [search-ws] propose a 'resultSetId' parameter for SRU 2.0

I propose for SRU 2.0 a parameter 'resultSetId',  and that we make the query parameter optional.

 

For background, remember the discussion about sorting: we are putting sort back into the protocol for 2.0 - it had been in 1.1 and we  removed it from 1.2 and put it into CQL instead; it had not been in CQL prior to 1.2.  But now we are putting sort back into 1.2 and leaving it it CQL as well. The reasoning for having it in both places is: one might use SRU with a query language other than CQL, one that doesn't have sort; and one might use CQL with a protocol other than SRU, one that doesn't have sort.

 

So, even though CQL allows the specification of an existing result set name, the same reasoning applies.   I think you should be able to name a result set, and a sort specification, both entirely within the protocol, i.e. outside of the query language.

 

(Just to clarify, the parameter would name an existing result set. It would NOT name a new one, the server still names the result set.)

 

 

--Ray