Matthew wrote:
>> http://srw.cheshire3.org/docs/update/
>
> I think we agreed that Update would be a new "service" rather than an
> operation of SRW. Although it might be a service which could not be
> implemented without SRW, it would be independent of SRW in that it is
> not required by SRW. In fact, I'd like to see update in a form that
> would allow it to work with OAI, etc...
>
> However, it was suggested that we press on with the definition of the
> service, and defer the precise discussion of how to position it in
> relation to SRW for a later date.
It does sound cool to articulate a protocol for creating records in an
index/database. The proposal at the previously distributed URL is
sketchy. One of the things I liked about SRW is the U. I like the fact
that the same protocol can be used in a REST-ful as well as SOAP-ful
way. The update proposal suggests using SOAP, specifically for handling
things like authentication. Fine, but that leaves out SRU because the
protocol explicitly states that authentication will be handled in the
SOAP wrapper. This will not make SRW and SRU compatible.
What is the problem update is trying to solve? Creating a new protocol
to handle such things as update sounds fine to me, even useful, but
mixing it with something that is/was designed for search confuses the
issue. The Unix Way has proven to be very effective in the
implementation of digital library applications. One trick ponies that
do their one trick very well. You then build systems by piecing the
parts you need together. I am able to advocate an update
protocol/operation, but not when it is combined with search. They are
separate functions.
My 2¢.
--
Eric Lease Morgan
University Libraries of Notre Dame
|