And, of course, you're right.  I guess I'm just resistant to putting so much
of the complexity of Z39.50 back into SRW.  So let's not do that.  As you
suggest, it's easy to roll a new protocol whenever we want.  Let's try to
carry as much of SRW forward as we can and add what they need.  (I'm not
sure we really understand what they need yet.)  I think CQL will survive
into the new protocol, but maybe without resultSetNames.  (If the intent is
to drive down costs, then that's a good candidate to go.)  Clearly searches
across multiple databases, but not so clearly multiple searches for multiple
databases.  One combined result set or one result set per database?  (My
preference is the latter.)

By the way, we decided long ago that holding connections open was more
expensive than making and breaking them.  So, our clients negotiate a
reconnect capability and we drop the connection after every response.  The
client sends a sessionID with the request that comes over the next connect.

Ralph

> -----Original Message-----
> From: Sebastian Hammer [mailto:[log in to unmask]]
> Sent: Sunday, May 25, 2003 7:16 PM
> To: [log in to unmask]
> Subject: Re: SRW/SRU and Metasearch products
>
>
> At 23:02 24-05-2003 -0400, LeVan,Ralph wrote:
>
> >These sound like serious folks with specific requirements
> and a commitment
> >to serious code.  Make them do real z39.50.
>
> I have total sympathy for this view, but my sense is that if
> we can't do
> better than that, we (ie. the ZiNG/ZIG community) might as
> well just tell
> the commercial content providers to go roll their own web
> service. That's a
> totally valid position, but it seems to me that it begs a more
> philosophical discussion about exactly who we hope will take
> up the SRW
> protocol, if not those groups. We *were* looking for a
> broader audience
> with SRW, right?
>
> The funny thing about SOAP and its integration into modern development
> environments is that it makes it easy as pie to develop customised
> protocols for just about anything, and people seem to do so.
> What I see as
> the major departure of the "metasearchers" is that they have
> no angst about
> dealing with multiple protocols -- they have business models
> and suport
> frameworks in place for handling it, and the users are paying
> for the party
> but they're also, arguably, getting more interoperability and
> functionality
> than we have been able to deliver with Z39.50.
>
> In that context, the business case for implementing SRW (much
> less Z39.50)
> is much weaker than it might have been 10 years ago, when
> network protocols
> were black magic and metasearchers might have been
> technically feasible,
> but they weren't practical business propositions. And it
> makes sense to me
> to at least seek a dialogue with these folks, and see if we
> can meet them
> halfway.
>
> --Sebastian
> --
> Sebastian Hammer, Index Data <http://www.indexdata.dk/>
> Ph: +45 3341 0100, Fax: +45 3341 0101
>