Navigation, services and interoperability

minutes from a session of the World Wide Web Working group
Chair: Victoria Burrill, Minutes: Eddy Boeve


Questions to CERN People

Because Francois Flukiger and Hakon Lie, both from CERN/Geneva, were only at the W4G workshop for one day, the session started with questions for them, especially concerning the development of HTML 3.0 and the new browser for it, developed at CERN: Arena.

Support for chemical formulae in HTML 3.0?
Answer: not yet
Support for in-line other datatypes?
Answer: no, is more a browser problem.
Non-scrollable area's in HTML documents?
Answer: Maybe, the problem has been addressed.
Are there connections with SGML community?
Answer: Yes there is collaboration; SGML and HTML can be combined well.
Support for page description languages?
Answer: Yes, a style sheet model will be included.
Attributes for data model description (REL and LINK tags, for printing document structures)?
Answer: Maybe.
Support for editing remote HTML pages?
Answer: yes, the PUT request supported and already operational in new Arena browser.
PC/MAC ports of Arena?
Answer: Yes, after stabilisation of the UNIX/X version.
Participants then gave keywords to define their interests and and were on that basis divided into two groups. The first group was interested in more technical oriented aspects; the second group in more user oriented (author/reader) aspects.

Integration and searching group

Chair: David DeRoure??, Minutes: David DeRoure??

Hytime; external link info; interoperability; dynamic documents; integration; scripts; extensibility; programming; SGML as protocol
Highlevel search engines; meta-information; automatic indexing
This is to be filled in by whom??

Applications, tools and usability Group

Chair: Victoria Burrill, Minutes: Eddy Boeve

Attendants: Victoria Burrill (RAL); Oreste Signore (CNR-CNUCE);Vincent Quint (INRIA); Corné van Delft (De Digitale Stad), Michael van Eeden (De Digitale Stad); Michael van Wetering (Rotterdam School of Management); Koos van der Hout (Hogeschool van Utrecht); Hans Eriksson (SICS); ????; Eddy Boeve (General Design)

Business and education, information kiosks, video conferencing, publishing, writing process
Graphical browsers, Web visualisation, editors and browsers, developing tools
Usability: Interactivity; design, navigation, interactivity paradigms, user-interfaces, page design
The result of the discussion in this group on usability aspects of the WWW was a kind of ‘wish list’ for both readers and authors on the Web, and where possible some hints for implementing them.

Editors and browsers

What browsers do we use?

Most people in this group used Netscape and Mosaic as their everyday browsers, both for checking the authoring result as for normal reading. Some of them used all available browsers (including MacWeb, Cello, and Lynx) to check the result of writing HTML pages. They did this because they didn’t know on beforehand what kind of browser their users were using.


Because most participants often used different platforms and browsers for accessing the Web, one of the most heard wish was a portable hotlist and history mechanism, preferably in the HTML language (as Netscape already implemented) and a portable structure.

Editing the hotlist

Most participants were not satisfied with the poor editing facilities of hotlists. What they really needed were:

In fact, the hotlist can be seen as one possibele view on the history list: only previously visited items that are of interest for the reader are marked as ‘in the hotlist’. This has another advantage: it’s easy to find the originating page of an entry in the hotlist.


Some participants used the Web only for public information retrieval and were inexperienced users. For that reason they were interested in simplified user-interfaces having only the most essential controls (e.g. navigation buttons), also refered to as ‘kiosk mode’.

Editing HTML pages

There was a strong need amongst authors for support in editing the HTML pages directly at the server side. With a WYSIWIG viewer that integrated editing and browsing.


Most problems related to navigation were not typical Web problems, but ‘normal’ multi-media and hypertext navigation problems and the Web community stronly suggested looking at the results obtained in this area already. Some of the more Web related navigation topics were:

HTML navigation support

A common problem with navigation was the lack of support in HTML for navigation, e.g. construct for specifying ‘where you came from’, in case there are more than one way you can reach that page from the HTML document structure.


A less stronger facility than the hotlist way of marking interesting places on the Web was looked for: ‘warm lists’. Users can then return, within one session, to these places.


Instead of following links, participants would like ‘pop-up’ interaction for glossary type funcionality or other short explanations (comparible with the MS Windows .help mechanism).

Page Design

Style Sheets

All participants desperately needed style sheets (as Hakon Lie addressed in his key-note talk): the separation of logical structure (contents) and the style (markup) of HTML documents.

Logical Styles

Need for more logical styles than available now. Although the introduction of style sheets makes it possible to create new logical structures on the fly, for simple browsers and for automatic conversion and parsing of HTML code, we still need a fixed set of predefined styles.


More markup facilities, e.g. the definition of the line width, in-line hardspaces, etc.


The wish list for images includes, among others:

Most of the images problems are related with limited bandwidth on the internet. A good solution would be a further expansion of the ‘HEADER’ request for HTTP servers: ‘HTTP Ping’. HTTP Ping requests meta- information about the document specified by some URL, including not only size and title, but also abstract, keywords, the document structure, included images and size and quality of these images, included links, other formats this documents is available in, etc. Browsers can implements ‘two-level clicks’ for following links: the first one gives you the meta-information, the second one, if the reader is still interested, gives the whole document or parts of it. The transmission of the meta-information gives estimations on how long the transmission of the document (with or without images) would take.


Annotations were highly appreciated by all participants, preferably in some Usenet News format: stored at some server (specified by some URL) and accessable by all readers on author, annotiation date, etc. Options for what part of the document the annotation is refering to is needed. Of course, annotations are described in HTML format.

Furthermore, constraints on the server side are needed for maintaining such annotation mechanism: When do they expire? How many are stored? This all resembles the ‘annotation server’ idea already implemented in the past.


Some general remarks on problems you can run into when publishing material on the Web: