Message-Id: <199602060321.VAA08003@library.wustl.edu> Date: Mon, 5 Feb 1996 14:06:39 -0500 From: Rhyno Art <mailto:arhyno@UWINDSOR.CA> Subject: Re: Loose Ends To: Multiple recipients of list WEBCAT-L <mailto:WEBCAT-L@WUVMD.WUSTL.EDU>
> I too envision this method of navigating through catalogs and
> resources in the future. But doesn't it have some implications for
> the "stateless/stateful" discussion? If one opens sessions on
> several Z-servers and never comes back to them as one navigates
> through a series of catalogs and electronic documents, wouldn't this
> create a real resource problem? Granted that Z-servers will have
> timeouts for inactive sessions and Web gateways or clients should
> keep track of each step so that one could go back to an earlier step
> and reactivate a Z39.50 connection, but on the whole I think it
> might be simpler if http to Z39.50 connections in this type of
> environment were stateless.
>
There are advantages to both approaches and a lot may depend on factors like whether the Z39.50 target performs any local caching in terms of how long it takes for more records to appear and whether maintaining "state" information makes a big difference in response time. Using some type of session ID may not be a problem for activities such as saving a search as a bookmark if the ID is appended to the URL and is ignored if the state information has timed out. If Netscape's notion of "persistent cookies" is supported by other browsers, then the gateway may be able to recognize specific clients without needing to add a session ID to a URL or as a hidden field in the form. I think the folks at the University of Dortmund have an interesting model in the SFproxy supplied with SFgate, which maintains a local database of Web pages that can be searched afterwards. Having a gateway work with a local database may allow records to be stored and indexed on the gateway machine for use in successive searches, and opens the door for moving some of the processing for combining searches in the "middle" of the client and Z39.50 server. I would be curious if some of the commercial Web gateway products already take this approach.The other aspect we have found with the Web is that users tend to be far more patient in waiting for a Web page than for our OPAC or any of the other applications we deal with. Whether this is a short-lived phenomenon or is soon subject to change is another question.
art
--- Art Rhyno, Systems Librarian Leddy Library, University of Windsor Internet: mailto:arhyno@server.uwindsor.ca Tel: (519) 253-4232, EXT. 3163 FAX: (519) 973-7076 WWW: <http://www.uwindsor.ca/library/leddy/people/art.html>