Re: Qualifier Combinations in CQL (Was: Bib-2 and the DC-Lib)

Mike Taylor wrote:

> To my mind, if SRW has a future at all, it's as a mechanism for doing
> serious IR.  To do serious IR (this should come as no surprise) you
> need complexity.  That's because IR is complex stuff.  No-one would
> argue that _all_ of the complexity in Z39.50 Classic is truly
> necessary -- some of it, with the benefit of hindsight, was a
> mistake.  But plenty of it is Real Complexity that's needed to address
> Really Complex Problems.  SRW needs to avoid throwing out the baby
> with the bathwater.

I think the SRW idea is to put the baby on the wire, while classic Z39.50
includes  the bathwater too.

Really, I think the idea of prescribing attribute combinations and giving
them identifiers, where the identifiers go on the wire (rather than the
serialized,  decomposed attributes) is a brilliant simplification (thanks to
Ralph) which sacrifices only the capability to specify undefined
combinations, which is what got us in alot of trouble to begin with. Bath
defines attribute combinations (but Bath doesn't take that additional step,
because classic Z39.50 doesn't really allow it).

And there isn't a limit to the number of these "index sets" (as Ralph calls
them), so theoretically nothing is lost.

So I take issue with your view that we're throwing out the lessons of
Z39.50, particulary the important work we did with attribute architecture.

I don't go as far as Ralph does in linking this all to Explain, however. I
think that this index set approach and Explain are orthogonal issues.

--Ray

Received on Thursday, 2 May 2002 17:58:01 UTC