The Davenport Group (aka Initiative for On-Line Documentation and Publication) Meeting summary for July 16, 1991 by Dale Dougherty Meeting location: HaL Computer Systems Next Meeting: Proposed Dates are September 18-19, 1991 in Cambridge, MA ******************************************************** CONTENTS: ATTENDEES MEETING SUMMARY ******************************************************** *ATTENDEES There were fifteen people attending the meeting. Dale Dougherty, O'Reilly & Associates Dave Dow, Fulcrum Technologies Bob Stayton, Santa Cruz Operations Eduardo Gutentag, Sun Microsystems Matt Koehler, Sun Microsystems Tom Athanasiou, Sun Microsystems Norbert Kuhnert, Sirius Technologies Michael Cation, Verity Paul Ausick, Sony Microsystems Marcia Allen, HaL Computer Chris Davidson, HaL Computer Dave Murray, Frame Technologies Dana Boudreau, HaL Computer Mark Hamilton, Frame Technologies Ken Kershner, Silicon Graphics * MEETING SUMMARY We began the meeting with presentations by Dave Dow and Dave Murray. ** Dave Dow, Fulcrum Dave Dow of Fulcrum led off by discussing how Fulcrum views the difference between interchange standards and interoperability standards, and why there is more hope that the latter will have impact. There are a lot of different file formats in existence, and there is significant experience with products that use these formats. Fulcrum feels that it must support a variety of file formats as input to their Ful/Text product. They are not in a position to choose a single input format or push a standard. The formats that were listed on an accompanying slide were: HP-tag, ATA-100 (Airline Transport Assoc. standard), MARC, , MIF from Frame, Printerleaf from Interleaf and PostScript from Adobe. These formats range from markup systems to page description languages. He said that he recognized as a trend the hybridization of markup and page description languages in an attempt to make up for the weaknesses of each. He cited other efforts to standardize on a hypermedia file format: o Hytime -- an SGML "application" (info through Graphic Communication Assoc.) o Dexter Group -- through Xerox PARC (group is not active --ed) o Content Data Model -- through Wright Patterson Air Force base. Dave noted the experience of the Air Transport Association (ATA) in trying to find a way to standardize on-line access. They determined that it was not practical to standardize index file formats. They instead developed SFQL (Structured Full-text Query Language) derived from SQL to provide a standard method of access. It has not been officially approved as a standard, and one of the sticking points is that SFQL contains a return-data model that specifies SGML. Some people feel it is not practical to parse SGML on the fly. Dave also mentioned other efforts of interest: WAIS -- Wide Area Information Service CDRDX -- (encrypted ?) an effort of the intelligence community [Esther Dyson's Release 1.0 (4/30/91) discusses these efforts in more detail. 212-758-3434. Also NY Times had a recent article on WAIS by John Markoff. There appears to be a mail alias for WAIS, wais-discussion@think.com, which I believe is moderated by Brewster Kahle of Thinking Machines (brewster@think.com).] Dave put forth a list of interoperability objectives for Fulcrum. From his slide: Unified multi-platform architecture Read/write capability Non-specific data model Address database publishing & tech publishing markets Cover WAN, LAN and stand-alone operation Extensible query for adtech searching Leverage popular standards work: Query syntax - (front-end) Client/server APIs - (back end) He also presented a slide entitled "Software Independent Approach" from a presentation by Neil Shapiro. It offers an architectural view of the different modules that make up an on-line system. It describes a data independent layer that includes the user interface, the data retrieval intelligence, and document presentation. Underneath that is a data dependent layer. The data retrieval intelligence is dependent upon a Full Text Query Engine based on an inverted index and a Keyed Query Engine based on a key database. The document presentation system is dependent upon a Data Management Engine that handles text, graphics, sound and video in various standard and proprietary formats. ** Dave Muuray, Frame Dave began by stating the obvious -- that there is a need for standards. The existence of such standards in the music industry makes CDs published by a variety of companies readable in CD players made by a number of manufacturers. There is no lack of potential standards in on-line publishing. Yet why don't we have one? The answer is that all of these standards are as yet insufficient. He feels that the current standards suffer from myopia -- that their focus is too narrow. He believes that the best role for this group is to serve as an independent critical voice. In other words, rather than proposing another standard, we should work to make existing standards bodies aware of our needs/requirements. That way, we might foster the emergence of an acceptable standard. Without that voice, we might have an insufficient standard imposed upon the industry. He illustrated the importance of this critical role by discussing SGML and PostScript. He cited two major problems with SGML: 1) It is a metalanguage for describing the elements of documents but it does not define semantics. In other words, a program can parse the elements of an SGML document but there is nothing to indicate what those elements mean -- is the element a footnote or a heading? This information must be supplied by convention (through knowledge of a DTD) outside of SGML -- namely that marks a first level heading. 2) SGML is format-neutral and there is a lack of formatting standards. It is not clear when such efforts as FOSI and DSSSL will produce results and there are no target implementations that prove it works. It is also unable to handle text inside of graphics. Dave recognizes some strengths in SGML: vendor independence, rich data content model, etc. He too responds to the promise that it holds out for many who want to work with information in a standard way. He simply doesn't think SGML is designed to deliver on that promise. The biggest strength of SGML, he said, is the number of people interested in it. He talked about PostScript. Adobe has been talking about PostScript as an information interchange language. The major problem with PostScript is that the document's structure is missing. You can't just add that structure through comments. It needs to be designed as part of the language itself. A second problem is the inability to re-format on the fly. You'd like to be able to modify the format and presentation of a document to have it reflect special uses. A page model is not always an appropriate display model for a document. Dave closed by saying that the group had a valuable purpose in offering a critical voice on the development of standards. ** Discussion of Group's Goals and Methods We set off trying to look at the various "by" statements proposed at last months' meeting. These were statements that described how we sought to "enable publishers of documents to deliver documents in a form usable by multiple viewers from multiple vendors." Ken Kirshner raised the point that he felt the statement of purpose did not really reflect the diversity of interests in the group. We weren't just setting out to enable publishers, but indeed hardware and software vendors as well as users. And depending upon who you are, the statement of purpose could be re-written to reflect a much different viewpoint. Several people agreed with Ken and several argued, myself included, that the term "publishers" was widely defined to include hardware and software vendors, who are publishers of documentation and re-publishers of third-party material. Also, the viewpoint of users was represented in the statement because if documents can be read by different viewers then users can choose the viewer they prefer. Dave Murray felt that behind the statement of purpose was the idea of fostering a kind of "ecology". In other words, we had to support a series of relationships with mutual dependencies and benefits to all. He made the analogy that a book publisher is not the only one who benefits from the publication of a book; there is the printer, the distributor, the bookseller, the reader, the author,etc. For on-line publishing to thrive, there needs to be a similar kind of ecology: book publishers, hardware vendors, ISVs, and users. After various discussions of the goals and objectives for the group, we stopped and rather suddenly agreed by unanimous vote to support the following method: by defining baseline functional requirements. By doing so we could offer a critical perspective on present and future obstacles to success in on-line documentation and publication. Later, we agreed that each of us needed to develop these requirements on our own and bring them to the next meeting for discussion. Ken suggested that we develop two sets of requirements: 1) What are the requirements of the viewer user? 2) What are the requirements of the information publisher? This distinction could also be looked at as the difference between a browser and an authoring system. It could also be seen as a set of requirements for a deliverable product and another set for the process that produces the product. We agreed to exchange this information by e-mail between now and the next meeting. You can send your list of requirements to the davenport mail alias (davenport@ora.com). Comments on those requirements and additional information should also be distributed. We can also use e-mail to involve a greater number of people who cannot attend meetings. The purpose of the next meeting in September will be to evaluate and merge the list of requirements and work upon defining a core set. (We will need to spend some time on process issues at that meeting, as promised, before actually getting started.) One of the goals is to establish these goals and then go public with them. Some of the group's future activities might be publishing articles and papers, holding workshops and other forms of getting our message out to the broader community. We also agreed to extend meetings to two days for those coming from out-of-town. Matt Koehler made the point that we need to be willing to put more time into the group to really get it going. Five hours once a month is not enough. We also need to be patient about achieving results. The next meeting was proposed for September in Cambridge, MA. ** The Davenport Mail Alias Just a reminder to use the davenport mail alias (davenport@ora.com). Perhaps the most valuable function of the group is to offer a forum for the exchange of ideas and information. Send mail to the group if you note a significant development or read an important article. There are so many other related efforts ongoing. Help keep us informed.