Re: DASL / GILS alignment

Folks,

I always do a double-take when folks refer casually to the "baggage" of
Z39.50, and the impossibility of implementing Z39.50 "In its entirety". To
my knowledge, no-one has *ever* implemented all of Z39.50. That isn't the
intention, and it is certainly no requirement. If what you need to do is
search and get records back, then you need to support those two services,
plus a little one to tell the server that's what services you will be
requiring today. Different groups of people have different requirements,
and hence choose different subsets of the protocol to implement. That's
fine. Because of a very simple options negotiation feature, pretty much all
implementations can interwork remarkably smoothly.

And people's requirements change over time. You may think all you want is
basic searching.. but along the way, you might suddenly feel like adding
database updates, index browsing, document ordering, scheduled queries
(current awareness services). Z39.50 has those things waiting for you, neat
and ready, and probably interoperable with anyone else who has developed
that requirement before you - *given a bit of care*.

In terms of searching and retrieval functionality, the "baggage" of Z39.50
is about equivalent to what you find if you pick up a textbook on
Information Retrieval in general. It is not a trivial application if you
want to do it right from the start. Considering that Z39.50 also represents
a consensus amongst a large group of people and interests, it is not a bad
piece of work. It's got a ton of flaws like any other technology - but by
dismissing it without understanding it, you virtually commit yourself to
years and years of mistakes and kludgework.

--Sebastian
--
Sebastian Hammer        <quinn@indexdata.dk>            Index Data ApS
Ph.: +45 3536 3672    <http://www.indexdata.dk>    Fax: +45 3536 0449
                                        ICQ# : 3478927

Received on Thursday, 28 January 1999 04:00:15 UTC