ZIT Meeting It is hoped that the ZIT will cease to exist by mid 1993. Case Sensitivity We need to come to terms with case-sensitivity. The standard is silent on the subject. Database and element-set names are treated as case sensitive, and in general, we do not assume that targets are case-insensitive. Some implementors would prefer that names be case-insensitive, but it is not clear that a target is doing the client a favor by providing case-insensitivity. Diagnostics We need to gain implementation experience with diagnostics, and then decide if this is a profile issue (i.e. specify information for the Addinfo field) or if a new diagnostic format is needed. User Testing So far, all interoperability testing has been done by the implementors. It is desirable (for real testing) that at some point, the systems undergoing interoperability testing be made available to users. WordList WordList is a structure attribute, and no one seems to be certain what it means, and implementors are interpreting it differently. Record Identifiers Bob Waldstein (AT&T) described the process of "caching" records. When there are two related queries to be performed, the first is performed and the records stored; then record id's are retrieved for the second query. The id's are matched against the stored records from the first search. For those that match, the records need not be retrieve. However, to do this with Z39.50, the target must supply unique record ids. Although this might be handled for bibliographic records (001 field) in general this is a newly expressed requirment and should be referred to the ZIG. Performance DLA has found that they cannot get records out nearly as fast as they had anticipated. Status FCLA - May be up in two months. NOTIS - Two origins, one for a PC and one for the mainframe. The target is in two forms: one for informational databases and the orhter for the mainframe. The first will be ready in two weeks. It will not be available because a commercial database is included. DRA - They are working on a target. RLG - Will be up for testing soon. They have a test version of all their databases (1%). LC - Developing the interface to TCP. Working on origin first; will be ready to send Inits in September. Target later in the year. VTLS - Working on an origin and target. Will be ready soon. Apple - Will be ready within three months. Initially on a Mac, then on an AUX. Mead - Their infrastructure is built, but they will not have anything before the next meeting. AT%T - Will have Explain in about two weeks. Demos Possible demos could be held at: - ALA, June '92, San Francisco - Share, August '92, Atlanta - Lita, September '92, Denver - Educom, October '92, Baltimore - CNI, November '93, D.C. - Cause, November '92, Dallas - ALA, January '93, Denver - Interop, Spring '93, D.C. - ALA, June '93, New Orleans Upcoming ALA. Too soon to do an organized demo, whatever there is will be ad hoc. Share. ? LITA. ? Educom. DRA can provide a facility. Lynch will get exhibitor list, and post it to ZIT list to see who has connections. Plan on seven hour coverage. CNI. ? Cause. Relatively constrained but same strategy as Educom. Only three hours. ALA January. Michaels and Lynch will pursue space at NISO booth. DRA might supply some funds to get a larger booth. There is reluctance to do a meaningful, organized demo at winter ALA if it is at the expense of posible demos elsewhere, because not that many people go. However it could be worthwhile simply as "practice" for a bigger demo. Interop. Try to get on program, then get a few vendors to provide space. ALA '93. Have a flambuoyant demo (look to Lita for space?). TCP/IP Layering The only layering issue was how to do an application level close. Currently, when a system (origin or target) wants to terminate, it does a TCP close. The peer system find out only by trying to do a read to see if a message has come, and gets an error condition. It has no way to find out if the peer application terminated (i.e. was finished) or if there was some lower level malfunction. There is a real need for a system to be able to determine this. With OSI, this is provided by an ACSE abort or release. With the ZIT profile there is no ACSE, so this could be provided only by adding another message to the protocol. It was decided that a simulated ACSE abort would be used. Support would be optional. Element Set Name There is a ZIG proposal to allow an origin to dynamically specify element sets. This would be downward compatible with version 2. There is at least one ZIT member (Berkeley) for whom Z39.50 provides little use without this functionality. They will implement this on an experimental basis for the ZIT project, with the understanding that it is not an approved feature, nor is it stable as proposed.