Z39.50 vs. graphical interface summary

[marc at ckm.ucsf.edu]

|Thanks to everyone who responded to my plea for help in deciding
|between a Z39.50 server and software to support graphical
|browsing of our OPAC.  

You've imposed a false dichotomy onto this system.  

You are going to need to choose both an access method to your data and
a presentation method to get those data to your users through an interface.

It is possible to mix either a graphical interface or ascii interface with 
either Z39.50 or your own home-grown protocol.  The problem of user interface
to OPAC-style search/display catalogues is that of state and protocol
differences, namely the statelessness of HTTP.  This either requires that 
protocol translation smarts and state machine exists in a very intelligent 
client or in a gateway.

The win with Z39.50 is that your data would be accessible, assuming you
played BIB-1 correctly, as a peer of many other OPAC's, both as a data source
for a harvest-style search engine as well as directly accessible through a
Z39.50 client or Z39.50-WWW/HTTP gateway.

-marc