Z39.50: how about this?

[KAREN SCHNEIDER]

So o.k., I'm getting smarter on Z39.50, and folks in my organization have
shown interest in the topic.  The question I kept getting yesterday was,
what needs to be done to  existing databases to make them work with a
Z39.50 server?  In other words, how would we make them compliant?  
(I consider these wonderfully open-minded questions!)

Then I started mulling over the whole meta-info issue, cataloging issue,
indexing issue, etc etc and if you saw a glow in the Northeast last night it
was my poor tired  brain trying to work through all of these issues in a
coherent manner.

Let's take two examples: our current online library system, OLS, and one
of those widgety databases I referred to earlier.

* OLS (a version of which is publicly accessible) is a BASIS database
running on an IBM ES9000.  It's a home-grown product.  It's actually the
best resource for locating holdings of EPA libraries, because some
libraries are too poor to catalog in OCLC but DO contribute their records
to OLS.  I find it an ugly but sincere and useful  tool that is very good at
locating "hard copy" EPA docs held by EPA libraries.  I would imagine this
would be a resource that our phantom Z39.50 server would access, if
this database were Z39.50-compliant, and retrieve records from (along
with other Z39.50-compliant resources, those we purchase, those we
build, those out there in netland).  Can we do this? How would we do
this?  Where are the guidelines?  Who should we talk to?  

*  An EPA Index To Test Methods is a resource created and regularly
maintained by the Region 1 EPA Library in Boston.  It exists in paper and
on diskette; it will soon be on the Region 1 server.  It consists of DB4
files cross-referencing test methods by EPA doc #, NTIS #, etc.  It is one
of many little databases we have running around the EPA that would
probably get better used if it were more accessible.  It also raises
another line of enquiry for me, because I can see where the index to test
methods would be a database running on some database server (not
DB4; maybe Oracle, or maybe even of the simpler tools, such as the
Perlesque  databases discussed in Shishir Gundavaram's _CGI
Programming_ [O'Reilly, 1996]) somewhere at EPA, with a CGI front end. 
then the question is, what would the primary link be?  I am inclined to
think that the database would be cataloged or at least indexed (MARC,
GILS, free-text, whatever) and that the record for the database,
including a hot link to it,  would be one of the items retrieved when my
phantom Z39.50 server retrieves a search for, say, "test methods." 
That, to me, sounds relatively easy (not building the little database, but
creating an Intercat record for it).  So, nu? Am I on the right track here? 

Karen G. Schneider
schneider.karen at epamail.epa.gov
Director, Region 2 Library, NY
Opinions mine alone