From: Lennie Stovel Z39.50 Implementors Group February 7 - 9, 1996 Twenty-first meeting Gainesville, Florida, USA Hosted by Florida Center for Library Automation (FCLA) Attendees Mark Hinnebusch, FCLA, MARK@MARK.FCLA.UFL.EDU Sean Donelan, DRA, sean@dra.com Clifford Lynch, University of California, Clifford.Lynch@ucop.edu Cecilia M. Preston, cecilia@well.com Klaus Sondergaard, EFILA, ksoe@ds.dk Mohamad Sakr, Sierra Systems, MSakr@SierraSys.com Erik Lorenz Petersen, EWOS, ELP@FL.DK George Blair, Excalibur Technologies, gblair@cq.com David Loy, Knight-Ridder Info., loy@dnt.dialog.com Ralph LeVan, OCLC, rrl@oclc.org Ron Buckley, OCLC, Ron_Buckley@OCLC.ORG Lennie Stovel, RLG, bl.mds@rlg.org Bob Waldstein, Lucent Tech. (AT&T), wald@library.att.com Bradley McLean, SilverPlatter Information, bradm@silverplatter.com Scott Sutcliffe, SilverPlatter Information, scotts@silverplatter.com Margaret St. Pierre, Blue Angel Technologies, saint@bluangel.com Oliver Javanpour, Gov. Telecommunications, Oliver.Javanpour@gtis.gc.ca Kevin Gamiel, CNIDR, KEVIN.GAMIEL@CNIDR.ORG John Kunze, UCSF, jak@ckm.ucsf.edu William E. Moen, University of North Texas, MOEN@lis.unt.edu Mark Needleman, Univ. of Calif., Mark.Needleman@ucop.edu Robert C. Gray, Gaylord CyberOdyssey Group, gray@cyberodyssey.com Eric Ferrin, Penn State University, egf@psu.edu Kevin Thomas, Ovid Technologies, kevint@ovid.com Ralph Orlik, Library of Congress, ORLIK@MAIL.LOC.GOV Dennis Van Dam, West Publishing, dvandam@research.westlaw.com Jeanne Gaster, West Publishing, jgaster@research.westlaw.com Joe Zeeman, CGI Group Inc., zeeman@fox.nstn.ns.ca Fay Turner, National Library of Canada, fay.turner@nlc-bnc.ca Sara Randall, Ameritech Library Services, srandall@als.ameritech.com Bill Jordan, Univ. of Washington, bjordan@u.washington.edu Andy Oates, Geac Computers, a.oates@geac.com Margery Tibbetts, Univ. of California, MARGERY.TIBBETTS@UCOP.EDU Terry Sullivan, FCLA, fcltps@nervm.nerdc.ufl.edu Scott Heyano, Univ. of Wash., slh@cac.washington.edu Edith Gottswinter, Univ. Libr. Karlsruhe, Ger., edith@ubka.uni-karlsruhe.de Yolanda Herd, Lexis-Nexis, yjh@lexis-nexis.com Peter Ryall, Lexis-Nexis, peterr@lexis-nexis.com Robert Bull, Satellites International Ltd., bull@sun001.sil.com Makx Dekkers, Pica, dekkers@pica.nl Thorn Roby, CARL, troby@carl.org Jeff Graubart-Cervone, Ameritech Library Services, graubart@als.ameritech.com Randy Menakes, Ameritech Library Services, rmenakes@als.ameritech.com Julie Baker, Pace University School of Law, jbaker@genesis.law.pace.edu David Williams, Pace Univ. School of Law, dwilliam@lawlib.law.pace.edu Denise A. Troll, Carnegie Mellon University, troll+@andrew.cmu.edu Les Wibberley, Chemical Abstracts Svc., les.wibb@cas.org Denis Lynch, SilverPlatter Information, DenisL@silverplatter.com Ray Denenberg, Library of Congress, RAY@RDEN.LOC.GOV Simon Igielnik, Washington University, St. Louis, SIMON@MEDICINE.WUSTL.EDU Z39.50 Implementors Group Introductions and status reports Mark Hinnebusch, FCLA: running v2 server and client; working on digital libraries and digital collections profile. Ray Denenberg, LC: has contracted with Margaret St. Pierre to develop v3 server for the version 3 testbed and then to implement the digital collections profile. NISO has published Z39.50. Denis Lynch, SilverPlatter: TRW is not in the Z39.50 business any longer; no one is supporting Search Access as a product. Will continue maintaining SNACC; FTP site at TRW is still there but existence not guaranteed. Les Wibberley, CAS: v3 SciFinder product; v3 server in use at three places. Interoperability testing with German Library Project. Denise Troll, CMU and representing Sirsi: waiting for Sirsi and OCLC to talk. Dave Williams, Pace U. Law School: public domain software running; want to load catalog et al. Jeff Graubart-Cervone, Ameritech: many v2 clients and servers out. V3 toolkit; will use for testbed. Has designed an external for OtherTermInfo for Scan Response for authorities (handout). Thorn Roby, CARL: v2 client and server. Using CNIDR ISITE server. Grant from state library for implementing in Colorado will generate more activity. Gateway/client activity with Dialog. Makx Dekkers, Pica: v2 client and server on central system. German Library Project late; testing in December; will start operating in March. Extension: Item Order. ONE project: will start operating in March; Explain in second half of '96; uses Satellites International software. Client and server in local system by end of '96. Rob Bull, Satellites International: European Space Agency project; live last July; link from Rome to Nice; because of this will do work on European Information Network system. API for German Library Project being extended for Item Order (ILL Order and Status PDUs). Participating in ONE. Porting software to PC, Windows, 32-bit; public domain 2/96. Catalogue Interoperability Protocol (CIP) Specification done by Logica; Satellites International will do prototype by 4/96. Peter Ryall, Lexis-Nexis: v3 client and server in Windows 3.1, moving to NT; for internal products. Looking at gateways. Edith Gottswinter, U. Karlsruhe: working in conformance test center for German Library Project. Margery Tibbetts, U. California: v2 client to OCLC and RLIN; v2 server has all databases, delivers MARC and OPAC; full text this spring. Working on graphical client, Web gateway. Medline active because of UCSF client development. Has increased server's buffers to 20K. Michael Thwaites testing v3 encoder/decoder; will put in public domain when stable. Andy Oates, Geac: improving features and performance of products including Windows client. Bill Jordan, U. Washington: incremental improvements; SilverPlatter interoperability; Windows version of Willow just released. Sara Randall, Ameritech: introduces Randy Menakes, Ameritech's official contact for Z39.50, including technical support questions. Fay Turner, NLC: debugging Amicus for external Z39.50 communications; interoperability testing of server coming; clients will have a two-week window for this testing. GeoPAC client tested for ILL and reference last fall. GILS for Canadian use: need to add data elements to core set; Government Information Finder Technology (GIFT) on a central host. Joe Zeeman, now at Software Kinetics, soon to be at CGI: SKL has grant from Canadian government to develop v3 toolkit; deciding whether to proceed. Jean Gaster, West: looking at v3 for internal use. Ralph Orlik, LC: v2 client and server running. SiteSearch server for catalogers; has foreign library records. Web site searches generate 50K searches per month to LC server. Will replace mainframe client with PC client. Kevin Thomas, Ovid: sells v3 client. Adding full text and graphics, including links to articles and back. Web stuff to talk to Z39.50 databases et al. Maintenance release available at port 2213. Eric Ferrin, Penn State U.: v2 client and server in use. Working on v3 client and server. Robert Gray, Gaylord: v2 client and server. WWW OPAC running, in production. Mark Needleman, U. California: gateway mentioned by Margery will be stateful. Brad McLean, SilverPlatter: v3 server with Explain and GRS in beta; needs clients to test; see Web site for address. Testing with U. Washington. Bob Waldstein, Lucent: v3 stable. Non-stable stuff moving to Lucent. Libraries split up into two separate organizations; R&D went to AT&T; they might buy commercial products rather than clone Bob's client and server. Lennie Stovel, RLG: will change RLG's support to provide sample database for free, complementary accounts to access production database for client developers. Bringing Sort up to current definition. Changing material type searching for format integration. Working with Retro Link on cataloging records service using Z39.50. Looking at Web gateways. Ron Buckley, OCLC: server usage has doubled; doing fixes and performance tuning; added ip address for European users. Ralph LeVan, OCLC: WebZ gateway in production since June; commercial services use it. Cecilia Preston: SICI standard (Z39.56) is available for ballot. Clifford Lynch, U. California: gearing up for v3, Web interface. RFC for Z39.50 over TCP needs updating for v3; will draft and post. Interoperability testbed announcement coming. Sean Donelan, DRA: v2 and v3 activity in parallel. V3 Unix server on VMS almost available; v3 client and GRS available. Electronic commerce available; DRA was certified to do credit card associations(?) but facility not being used. Gateway in general release. DRA is Internet service provider. Oliver Javanpour, Government Telecommunications: Canadian infrastructure for government information dissemination. Fulcrum server. Z39.50 adopted; looking at GILS for describing information. X.500 included. Margaret St. Pierre, Blue Angel Technologies: new project for LC: v3 server with Inquery search engine. Public domain GILS server implementation. Bill Moen, U. North Texas: coordinating CIMI profile development. Register of Implementors (Denenberg) Denenberg circulated a copy of the current register for implementors to mark up, or you can send updates directly to him by email. Date of verification or update is included on the Maintenance Agency (MA) web page. Digital Collections profile group report (Denenberg) Denenberg posted an announcement of Draft 5 of the digital collections profile to the Z3950IW list 1/23/96. The profile group met 2/5 and 2/6. Denenberg reviewed the background, related this profile to one that CIMI is developing, and discussed problems that the profile addresses. The profile provides for navigating through hierarchical collections and for encapsulating both digital objects and their descriptions, or pointers to either, in a descriptive record whose schema is documented in the profile. Draft 6 will be out in about 4 weeks; technical details have changed but the model has not, so reviewers can look at the prose sections of draft 5. The final version will come to the ZIG and then the OIW SIGLA for approval; development will continue after approval. LC will maintain the profile for now. Denenberg is trying to coordinate with the group that is developing the Catalogue Interoperability Profile. The relations between the digital collections profile and the GILS profile is murky; GILS has a different model. An implementor could use GILS for description and digital collections for navigation. The attribute sets in the two profiles have different aims and could be used together to accomplish specific goals (with v3). CIMI profile group report (Moen) The CIMI profile is one part of a multi-part project called CHIO, Cultural Heritage Information Online. It will have a schema for objects that are opaque in the digital collections profile. A challenge is dealing with SGML documents of which only sections need to be returned to the user. The profile will come to the ZIG via the listserv for review. The goal is to have a stable draft by the end of April, to implement it over a nine to twelve month period, and to have a final profile in late summer 1997. Ranked List Query group report (Ryall) The specification of the ranked list query (RLQ), type 102, is complete enough to begin experimental implementations. An interoperability group would be useful. Profiling is also needed. Ryall presented an overview covering the uses of RLQ, its features, the differences between RLQ and type 1 queries, etc. Some of the ensuing discussion concerned the possibility of subsuming the definition of type 102 in type 1, which seemed to be a desirable goal. The RLQ specification together with notes from the discussion following the presentation can be found at http://potomac.ncsl.nist.gov/~over/rlq/main.html. Explain Testbed report (Dekkers) Dekkers outlined the goals in a posting to the list in December and defined three levels of complexity for servers. He got expression of interest from six sites and will form a listserv for this activity. Time frame: finalize definitions this month; set up implementations in April and May; start doing things in June; report at October ZIG meeting; stable testbed environment by 1997 at German conformance site. GILS Profile report (LeVan) The GILS subgroup of the OIW SIGLA meets monthly and has accumulated changes, enhancements, and defect reports. It was close to a stable version 2 but decided to hold off when the Canadian requirements became known. The changes in hand were not substantial or protocol-sensitive. The buy-in by agencies has not been great and current implementations are not compliant with the profile. The GILS Advisory Board will plan evaluations of GILS implementations over the next six months. There is no body funded to provide support or to enforce compliance, though the subgroup acts as an information transfer agent for content issues. A problem is that there is no guidance for what level of information is to be made available; agencies have to decide this for themselves and this takes energy and focus. These issues are independent of the protocol issues. ItemOrder activity reports (Wibberley) OCLC has implemented item order. There is an Interlibrary Loan Protocol Implementors Group (IPIG) whose membership overlaps the ZIG; participants will implement two ILL PDUs, which could be re-used in support the Item Order Extended Service. Satellites International is coding Item Order for the German Library Project, using the same two PDUs; will have test programs available that developers can use against their own implementations. Ameritech is doing a request Extended Service. Lucent's server accepts Item Order and interworks with Ameritech, though it does not recognize the EXTERNAL ILL PDUs. Meeting minutes (Denenberg) Denenberg has put the minutes of past ZIG meetings up on the Maintenance Agency's web page, to the approval of the group. He suggested a process for making the minutes official; the group did not go along with his proposal. Directory of databases/servers (Denenberg) Denenberg queried the group about what information the Maintenance Agency's web page should carry about available Z39.50 implementations, servers, and/or databases. Different people want different types of data, or different views on the same data; different institutions want different information listed about what they support. Keeping the information current is an issue. Another angle was whether listing on the MA page could imply more (or less) than the lister intended. One proposed solution was for the ZIG to set up a site of its own. Hinnebusch will look into the possibilities. Attribute sets (C. Lynch) Clifford Lynch covered in depth the points about attribute sets that he had made in email to the Z3950IW list. He noted key problems (duplication of common attributes in different sets, interoperability due to proliferation of sets, ambiguous semantics) and technical issues, and made suggestions for working assumptions for future activities. The group agreed that the standard should contain only those attribute sets that have a structural role in the protocol (e.g., Explain); that specific attribute sets should be developed by the relevant interest groups, not the ZIG; and that these groups will need specific guidance in the properties of attribute sets. During discussion of the appropriate venue for developing a bibliographic attribute set (to replace bib-1, which has become more than bibliographic) the point was made that international participation was a requirement. Other topics discussed: the possibility of developing core, base or primitive attribute set(s), classes of attribute sets, or both; the possibility or necessity of extending the type 1 query; dynamically extensible sets; SQL; the possible relationship of schemas to attribute sets. Lynch will establish a subgroup of the ZIG to work on the architecture of attribute sets. This work is likely to be a theme for future ZIG meetings. Bib-1 attribute extensions (Denenberg) Extensions for the bib-1 attribute set are proposed to the ZIG; after discussion at a ZIG meeting, they are listed as proposals on the MA web page on Registered Objects; they are adopted at the next ZIG meeting. The group suggested various changes to the use attributes proposed by the German Library Project (with apologies, as these had first been presented two meetings ago). Denenberg will coordinate with Luchner. Zeeman distributed a document listing proposed attributes from the National Library of Canada'a AMICUS project. Zeeman requested that some numbers be set aside within the bib-1 attribute space for local use, that is, locally-defined attributes used with the bib-1 OID; Denenberg agreed. MARC-1 attribute set (Zeeman) The Library of Congress has expressed a need for a "MARC-1" attribute set based on the USMARC format's tags. Zeeman had proposed such an attribute set earlier as an intellectual exercise; it is more detailed than LC's proposal. Others could use such an attribute set; those interested in it will work on it and come back to the ZIG for more discussion. Zeeman suggested that it be based on ISO 2709. Bib-1 diagnostics extensions (Denenberg) Denenberg distributed a list of seven new bib-1 diagnostics that had been proposed earlier; many had been discussed on the Z3950IW list. The group approved them. Init diagnostics (Denenberg) Denenberg distributed a list of fourteen diagnostic messages that could be returned in an Init Response. The group agreed that they should go into the bib-1 diagnostic set. Two of them (those providing for Addinfo) will be added to the diag-1 diagnostic format as well. Possibilities for returning diagnostics in the Init Response were discussed. Upshot: in v2, send in UserInformationField (because OtherInfo not available) as an EXTERNAL that identifies a syntax like OtherInformation; in v3 use OtherInfo. Denenberg to develop syntax; its OID will be in object class 10, User information format definitions. Body parts for ASCII display (Denenberg) The question is: if the server is returning an element in a GRS record that is ASCII (GeneralString) and it wants to attach a Variant with a body part type to it, how does the server indicate that it is for display to the user? Postponed to the next meeting. Well-known databases (Denenberg) Agenda item dropped. Version 2 cludges (Denenberg) Agenda item dropped. Clarifications/Interpretations (Denenberg) Denenberg has begun posting clarifications and interpretations of Z39.50 on the MA web page. These items are issues that arise about which consensus is reached on the Z3950IW list or elsewhere. Denenberg will make a synopsis of each issue and a response that reflects or proposes a consensus. At each meeting, the ZIG can discuss and approve the resolution of issues that have come up since the last meeting. Denenberg had six such issues for discussion at this meeting. (1) Piggy-Search/Present Support: [if we discussed this, the discussion didn't make it into my notes; we had approved what the response proposes at the last meeting.] (2) Preferred Message Size: response approved; add note that proposing actual message sizes is unwise. (3) Scan: negative value for position-in-response: response approved; clarify that all values should be allowed; explain what negative numbers mean; list relevant diagnostics. (4) Close Status: response approved. (5) Multiple ranges: response approved. (6) TargetInfo key in Explain: the sense of the response was approved though there was various wordsmithing. Init Negotiation parameters (Denenberg) The question is where a negotiation record should go in an Init request or response, in OtherInfo or in UserInformationField. Upshot: use OtherInfo wrapper in UserInformationField. With this change, the need for OtherInfo disappears, but it will not be taken out of the standard as implementations use it. Structured Vocabulary Browse (Kunze) Kunze distributed a new version of his paper defining a mechanism for moving to neighboring term nodes in a structured vocabulary such as a thesaurus. To connect such a vocabulary with a database that uses terms from the vocabulary, use Explain (in a new way). Work is needed on record syntax. Possibly profiling will be needed as well. EWOS (Peterson) The European Workshop on Open Systems is examining itself, restructuring, and looking for funding opportunities. It is moving away from formal profiling. Rank/analyze (D. Lynch) Denis Lynch had proposed a pair of PDUs for retrieving information from the records in a result set according to client-specified criteria, for example, a list of all the unique subject headings in a result set together with their occurrence. This becomes a new result set. This result set is not the same as a vocabulary as discussed in the earlier agenda item. In response to the suggestion that this should be an Extended Service, the group discussed the criteria for making a service into an Extended Service; LeVan offered three: persistence (things that exist outside a specific Z39.50 sessions); non-core functions (services for which there is a business reason but that are not an inherent part of information retrieval); unpopular functions that would not receive wide support [among NISO voting members, for example], or, in other words, functions that were not in the development priorities for version 3. Based on these criteria, the consensus was that this service does not belong in Extended Services. Issues remaining: integration with an earlier similar proposal by Wibberley; the dbname of the constructed result set; the name of the service ("rank", "analyze", and "dedup[licate]" all have problems); which functions should be within the scope of this service; define an Init options bit for this service. Trigger Access Control (D. Lynch) Lynch had proposed a PDU that asks the target to issue an Access Control Request, so that, for example, a user could change an account in the middle of an association. Wibberley had proposed a similar "set user parameter" Extended Service. There was a consensus that this sounded like an extended service. Further discussion on the Z3950IW list was indicated. Scan (Waldstein) Waldstein listed several things that have come up about Scan that lead him to suggest an Additional Scan Info field to carry externally- defined information. He will post a description to the list. Item Order profiles (Zeeman) Zeeman had updated the two profiles for use of the Item Order service, based on the discussion at the last ZIG meeting. After some discussion of specific points, profile 1 was approved. Wibberley asked for a longer review time for profile 2 and his request was granted. Net-friendliness of version 4 (Zeeman) Zeeman had proposed a redefinition of the elements in which using an OID is possible to incorporate the possibility of using other identifying strings, for example, Internet objects such as URLs. This would mean that Z39.50 would not have to have an OID registered for information objects that already have formal names registered by another agency. The consensus was that this is an interesting idea. Its effect on Explain needs consideration. Future meetings Brussels, Belgium, October 2-4, 1996. Hosted by EWOS. Preceded by tutorials on Z39.50 and character sets. The Europeans are organizing the latter; Randall was volunteered to organize the former. Washington, DC, April 7-9, 1997. Hosted by the Library of Congress.