I think the few extensions that we possibly need do not justify two response schemas (RS-1 and RS-2). Are there examples of what we need in SRU that we certainly don't want in SRW?

The extensions that I like to have are usable in as well SRU and SRW and could be optional anyway. For example returning the requesting parameters in the response to keep the response selfexplainable. 

Theo


>>> [log in to unmask] 06-02-02 13:58 >>>
> -----Original Message-----
> From: LeVan,Ralph [mailto:[log in to unmask]] 
> Sent: 06 February 2002 11:55
> To: [log in to unmask] 
> Subject: Re: issues list
> 
> I don't see how you can kill responseSchema from SRU.  The whole point
of
> SRU was the clients ability to specify a responseSchema whose contents
> were
> greater than an SRW response.  If I want a button on the screen that
says
> "Next Five Records", then there better be the data in the response
that I
> can put behind that button.  Remember, there is no processing of the
> response, simply reformatting from XML to HTML.

At a guess you'd need the query and base URL in the response and then
use the XSLT count function to work out the startRecord for the next
request.

Reread my original proposal since I think you have missed a fundamental
point.

I was not arguing that SRU should return the SRW response but that we
have an extended schema for SRU which includes SRW plus the additional
information for this sort of application. Once we've defined this
extended schema for SRU (lets call this RS-2) do will still need to
offer the ability to extend this even further.

For simplicity can't we just have RS-1 used in SRW and an extension RS-2
used in SRU and leave it there?

Matthew