This long-awaited and well-attended concertation day had 35 'teccies' (slang for technical / systems / computer-orientated people) in attendance. A wide range of issues were discussed, probably too many for one day and certainty too many to be covered in this article. One of the main outcomes of the day was the identification of major issues, of interest to many projects, which warrant further discussion. It was an informal day allowing for a lot of attendee input and interaction.
Chris Rusbridge introduced the day before handing over to Brian Kelly who gave a brief presentation on two topics; this helped to pinpoint areas for further discussion later in the day. These presentations,
are available in Powerpoint format from http://www.ukoln.ac.uk/~lisbk/events/elib/concertation/agenda.html Issues arising from these presentations included the use of 'style sheets' in the creation of HTML documents. Style sheets are a means of separating document structure from document appearance. They are of particular importance for long-life documents which will have to adopt to the constantly moving standards of the Internet. Creating HTML documents without style sheets can be compared to writing an entire program using local rather than global variables. The long term maintenance problems can be immense. Style sheets have detailed specifications for font, colour etc. They are very flexible and functionally rich and are now available for Internet Explorer version 3. The incorporation of style sheets should play a role in the setting of standards, else users won't be able to access old documents.
Another topic which sparked interest was the use of frames. Frames defy the principles of using style sheets by grouping the user interface information together with the document content. However, W3C propose the implementation of frames using style sheets. Some people see frames as a complicated means of doing something simple. The use of frames can leave the end-user with little control and web developers shouldn't force the interface on the user. Although the advantages of frames such as the reduction in maintenance were noted, disadvantages were also identified. These included the increased download time due to large memory load and the fact that 'frames' cannot be bookmarked. Frames are an example of a newly introduced formatting tag which is architecturally flawed and contravene the spirit of SGML. In the quest to separate document appearance from HTML structure, developers should use HTTP facilities to enable content negotiation. This could allow the client to choose the document format which the server will provide.
The second presentation led to discussions on the problems of overloading the server when using cgi scripts to deliver dynamic documents to the client. An alternative is the use of mobile code which is downloaded from the server and run on the client computer. Examples of mobile code which were discussed include Sun's Java and more recently Microsoft's ActiveX; however, security poses a problem with both. These topics were given further attention in the 'breakout sessions' later in the day.
Next there was a brief presentation from the DIGIMAP project which focused on the problems encountered when accessing digital map data across the Internet. Ways of defending the network and server against too much traffic while a user downloads a map were discussed. Although developers can push more processing to the client side, they shouldn't overload the client due to hardware limitations. DIGIMAP currently uses cgi scripts written in Perl and Java; the Java has caused some problems with printing the maps.
After these brief presentations which helped to pinpoint certain problem areas, the attendees were split into groups to identify technical issues for discussion in the afternoon session. A wide range of issues were identified (for details see http://www.ukoln.ac.uk/~lisbk/events/elib/concertation/reports/report-back.html ). Of the many suggested problem areas, four which were common to a number of projects were used as a basis for the thematic discussion sessions. Attendees choose which theme was most relevant to their project and joined the respective group. The four groups of approximately ten people each set about debating the following issues:
The aim of these breakout groups was to identify problem areas, suggest possible solutions and what groups such as eLib and UK Web Focus can do to help; each group gave a report back on their findings.
Authentication was regarded as a difficult area, the main problem being the maintenance of a large number of user ids. Issues such as data protection and management of usernames and passwords were also discussed. One recommendation was to look at cookie technology and software such as VeriSign. Cookies work by sending information from the server in the HTTP header of the server request. They can be used to trace the interaction of a user through an entire session, allowing web browsers to gather information about the user of a web server. VeriSign issues and manages digital ids to support a wide range of users; for details see http://www.verisign.com/. There weren't any strong recommendations for solutions to the many problems; perhaps a day devoted to the issue of authentication would be useful.
Systems issues was another group with a lot to discuss and the outcome of the discussion was the identification of various issues which require further attention. Such issues include database integration, electronic and video conferencing and virtual WWW servers. Apache was felt to be a good server.
The interoperability group focused on the need for eLib resources to be accessible via cross-project searching. The advantages of this close interoperability between projects were put forward but it was noted that cross-searching wouldn't be applicable to all projects. If the idea was to be implemented it would have to be based on a standard such as the Dublin Core or Whois++. Another result from this group was the recognition that there is no standard metadata set and one needs to be developed.
The final group was dedicated to document format and user issues, and discussed the pros and cons of using PDF on the Internet, as well as HTML tools. One of the advantages of Adobe's PDF format for documents was the ease of referencing due to accurate page representation. PDF was deemed as the most secure means of transporting documents across the Internet. One disadvantage noted was the problem of printing PDF documents using non-postscript drivers. Some useful HTML tools were also recommended. These included Microsoft's Internet Assistant and RTFTOHTML. Automated website development tools suggested for the Mac were Frontier and Myrmidian. Frontier can also be used for statistical analysis of users. This group also suggested the introduction of a lis-elib-tech mailbase list rather than using the more general lis-elib. There has been a quick response to this suggestion and there is now a list lis-elib-tech, the centre for which can be found at: http://www.mailbase.ac.uk/lists/lis-elib-tech/; this is devoted to solving those endless technical problems. The overlap between various projects was also noted and an elib FAQ to help people overcome common problems effectively was suggested. Elib teccies were encouraged not only to post questions, but also the solutions, once a problem has been solved.
All in all, the day proved very successful and due to the number of issues yet to be discussed in detail, the need to hold more events for elib technical staff was recognised. Perhaps a day focused on each of the breakout group topics could be arranged. Now that there is a teccies list, people will be able to voice their queries and solutions and maintain cross-project interaction. Another recommendation from the day was to look at the use of hypernews and video conferencing for elib technical staff; small trials are being planned so watch lis-elib-tech for more information.