On Tue, 21 Dec 2004, Mike Taylor wrote:
>> From: Dr Robert Sanderson <[log in to unmask]>

>>> Yep. So two explain documents for a SRW/SRU/Z39.50 server that
>>> features exactly the same resource behind it (Z39.50 supports
>>> CQL+XML), due to non-repeatable serverInfo.

>> Not to put too fine a point on it, but I don't think this will be
>> very common ;) And even if it does become common, it's not
>> outlandish to have one record for Z39.50 and a different one for
>> SRW/U.

> Hmm.  To represent three closely related services, I would expect
> either to need one record or three.  To need two does seem perverse.

The SRU and SRW services are absolutely identical with the only difference
being that one accepts the parameters via SOAP and one via the URL.
There is then a huge jump to Z39.50 which:

* Doesn't use HTTP
* Supports multiple record syntaxes, not just XML
* Supports multiple query languages, not just CQL
* I know of only two implementations out of hundreds that support CQL,
   which is done via a very recent extension to the protocol.
* Has a totally different service definition
* Doesn't support server side result set naming
* Doesn't support a single Explain operation
* Supports multiple databases at the same endpoint
* ... ... ...

If there were any functional difference between SRU and SRW, then I would
agree. But ZeeRex describes the capabilities of the server, not the syntax
in which the server accepts requests.  For that there's WSDL+XML Schema
(or ASN.1 for Z39.50)  Once the parameters of the request hve been parsed,
the behaviour described by the ZeeRex record is identical for SRW and SRU.
Furthermore, the SRW/U specs mandate a single record in the base profile.

If someone were to create a 'new' protocol that had exactly identical
functionality as SRW but ran via a different protocol, then I would have
no problem with a new way to specify what is currently in the protocol
attribute of serverInfo.

For example:
<serverInfo>
  <protocols>
    <protocol identifier="http://www.loc.gov/srw/">
      <title>SRW</title>
      <version>1.1</version>
      <transport>http</transport>
      <method>POST</method>
    </protocol>
  </protocols>
  <host>
  <port>
  <database>
</serverInfo>

But for the mean time, this seems over-engineering when we have a
perfectly usable "SRW/U" item in the @protocol enumeration.

> I am slowly being persuaded towards multiple <serverInfo>s.

To make myself slightly more clear, as the ZeeRex schema maintainer:

You'll pry multiple serverInfos from my cold dead fingers =)

Rob

       ,'/:.          Dr Robert Sanderson ([log in to unmask])
     ,'-/::::.        http://www.o-r-g.org/~azaroth/
   ,'--/::(@)::.      Dept. of Computer Science, Room 805
,'---/::::::::::.    University of Liverpool
____/:::::::::::::.  L5R Shop: http://www.cardsnotwords.com/
I L L U M I N A T I