On Thu, May 09, 2002 at 04:20:48PM -0400, Ray Denenberg wrote:
> As I try to write something to justify the srw/zeeRex relationship, it
> occurs to me that the only concrete reason offered why the two efforts
> shouldn't be aligned is the index vs. attribute approach. If there are
> other tecnical incompatibilities that I've missed, or things that SRW need
> that ZeeRex doesn't supply, please list them so we can try to address those
> Looking over the recent discussion on indexes vs. attributes -- that
> discussion got sidetracked to a discussion of soap, but it seems that where
> we left off was that zeeRex would indeed include the abstract index names
> that srw needs. If zeeRex also lists explicit attributes, so what? SRW can
> ignore them, or better, zeeRex could define an element set for srw (srw is
> going to ignore other things too, like record syntax, for instance).
> So please tell me, what am I missing?
Having the abstract names in ZeeRex would be good and make it relevant.
What is not clear to me (and I have not thought about it) is what
is the purpose of 'explain' with SRW. For example, do we want to
return an XML document for applications to pull apart? Or do we want
a SOAP based API with various method calls to return simple types
(strings, integers - not XML documents) which describe what is
in the server.
Put another way, a ZeeRex record may be able to describe a SRW server.
But does that mean SRW should use a ZeeRex record as the SRW explain
mechanism? Or should SRW define functions such as
giveMeAListOfIndexNames() returning an array of strings
giveMeIndexDescription(String name) returning description as string
As a programmer, I would prefer a set of function calls. Yes, XML is
cute and can represent complex data models. But a SRW goal to me is
to have a very simple data model exposed to programmers using SRW.
So a ZeeRex describing a SRW site? Sounds fine. ZeeRex as the SRW
explain mechanism? I would like to think about it a little more
before committing one way or the other.