Version 1.1 of this report
Minor changes have been made to fix typos - see the changes
file for details.
This document gives a report on the W3C (World Wide Web Consortium) Advisory Committee meeting held at Rutherford Appleton Laboratory, Abingdon on 15-16th January 1997. I attended the meeting in my capacity as the JISC representative on W3C.
W3C (the World Wide Web Consortium) is a body founded to "further the evolution of the World Wide Web while keeping its interoperability". W3C has host institutions at MIT (USA) and Inria, France. Recently Keio University (Japan) became the third host institution. Tim Berners-Lee is the Director of W3C, and Jean-Francois Abramatic is the W3C Chairman.
As of January 1, 1997, W3C had 63 Full Members and 97 Affiliate members. JISC (Joint Information Systems Committee) is an affiliate member of JISC. Brian Kelly, UK Web Focus based at UKOLN, University of Bath is the JISC representative on W3C.
The W3C Advisory Group meeting was held at Rutherford Appleton Laboratory, Abingdon on 15-16th January 1997. A total of 113 names were given on the list of participants, which included 73 W3C representative from member organisations, 4 invited guests (from CERN, DISC, FEDIM and the IETF), 15 from Rutherford Appleton Laboratories and 21 W3C staff members.
The first day of the meeting provided an opportunity for W3C to report on selected W3C achievements from the past six months. The meeting began (at 8.45 am - an early start, especially for the many participants from the USA and other far-flung locations) with a welcome and introduction from Bob Hopgood of RAL. Following Bob's summary of the local arrangements, Jean-François Abramatic gave and overview of the structure of the two-day meeting. Sally Khudairi then gave an overview of W3C. She described how W3C has been referred to as "... the band of lions all willing to lie down with one another to further online standards has come to dominate the development of standards on the Net." W3C members:
Consortium activity areas fall under one of three Technical Domains:
Following these overview presentations, we then had two technical presentations on architectural developments. Jim Getty (from DEC, currently working for W3C) gave a summary of HTTP performance tests. The HTTP/1.1 specification describes why HTTP/1.1 is needed (HTTP/1.0 is not fully cache-aware, provide support for persistent connections or for virtual hosts). Jim described work to check that HTTP/1.1 fulfills its design goals of (a) improvements in network behaviour and (b) improving the perceived speed for the end user. W3C are currently carrying out a variety of tests using a number of servers (Jigsaw, libwww, Apache and Netscape Communicator) running on a variety of platforms (servers running on Linux with clients running on DEC Alphas. Clients running on Windows NT were used for low bandwidth dialup tests.) The tests carried out looked into the effect of pipelining and output buffering, pages with many inline images, caching, and the network environment (high bandwidth, low latency [e.g. a 10 MBit Ethernet LAN]; high bandwidth, high latency [e.g. the WAN between MIT and LBL] and low bandwidth, low latency [PPP over a 28.8 modem]). The conclusions from the tests were positive, indicating that significant gains are made by using pipelining - it uses less than 1/5 packets of HTTP/1.0, and is up to twice as fast as HTTP/1.0. There are also server gains, although client implementations require care and effort. The conclusions were:
The performance tests have not yet finished. Further information is available at the URL http://www.w3.org/pub/WWW/Protocols/HTTP/Performance/.
The second summary of architectural work in the morning session was give by Dan Connolly and covered WEBDAV: Web Distributed Authoring and Versioning. Although the Web was initially conceived as a collaborative authoring environment, it has been implemented as a publishing environment, with small numbers of authors and many readers. WEBDAV is an IETF group which is dealing with the issues of distributed authoring and versioning on the Web. Its aim is to enable distributed web authoring tools to be broadly interoperable and to gain economies of scale from shared technologies. The talk summarised the history of distributed authoring, from Tim Berners-Lee's original browser on the NeXT platform, the failure to implement PUT from the HTTP/1.0 protocol by NCSA and tools such as NaviPress/NaviServer and Vermeer FrontPage (later taken over by Microsoft). At a WWW4 breakout session (see http://www.w3.org/pub/WWW/Collaboration/9512www4/auth-tools) interoperability among distributed web authoring tools, and identified a number of issues including (a) need for a common access control model (different tools current have different file locking mechanisms) (b) the "lost update" problem (c) the need for new HTTP methods, such as BROWSE and MKDIR and (d) strong authentication (such as MD5). In September 1996 the WEBDAV group was set up. There have been a number of meetings. The current work of the group includes producing (a) a Requirements Document (b) a Protocol Specification and (c) a Scenarios Documents. The group has applied for IETF working group status.
Further information is available at the URL http://www.ics.uci.edu/~ejw/authoring/
Following the two presentations on Architecture highlights, Vincent Quint in the summary of User Interface highlights gave a report on the HTML ERB (Editorial Review Board). The HTML ERB aims to:
The HTML ERB is chaired by Dan Connolly. The Voting Members are EBT, HP, IBM, Microsoft, NCSA, Netscape, Novell, Softquad, Spyglass and Sun. In addition Adobe, Pathfinder are invited participants and Dave Raggett, Hċkon Lie, Chris Lilley, Bert Bos attend as W3C staff. The group has a Teleconferencing meeting every Tuesday. Meeting were held on June 20-21, 1996, September 9-10, 1996 and December 5-6, 1996. The ERB has made two recommendations: CSS1 (on 17 December 1996) and HTML 3.2 (13 January 1997).
The HTML ERB is currently preparing the next version of HTML, which is known as Cougar. Working Drafts are available on (a) Scripting, (b) Fill-out forms, (c) Frames and (d) Multimedia objects. Other drafts which are available include (a) Web collections, (b) Link types, and (c) Web page scripting services.
The HTML ERB feel that desirable features of HTML include:
Vincent Quint concluded by asking the question "Is HTML stabilising?" Many recent HTML extensions are actually relevant to CSS. HTML could be seen as a small kernel with general purpose elements. We still need (a) A clean interface (API); (b) A specific mechanism for extensibility (like PEP for HTTP) (c) A richer object model such as Maths.
Following the overview of the work of the HTML ERB Chris Lilley gave a report on CSS (Cascading Style Sheets). CSS level 1 became a W3C Recommendation on 17 Dec 1996. It is the first in a family of specifications. A number of implementations are already available (Chris demonstrated stylesheets using Microsoft's Internet Explorer) and more are expected.
Stylesheets describe the presentation of a document, leaving it to HTML to describe the document structure. Stylesheets in general are lightweight and fast-loading. CSS in particular:
CSS consists of selectors, properties and values. For example:
H1 {color: red}
CSS can be inline, as shown above (this is not recommended), included in the HEAD of a document or external to the document(s) and linked using the <LINK REL=STYLESHEET TYPE="text/cc" HREF="http://mydomain.ac.uk/style/org.sty">
The CSS font specification assumes that the font is available on the client machine. A list of alternatives can be provided (Arial, Helvetica, sans-serif states that Arial is the preferred font. If it is not available, then use Helvetica. If that's not available either, then use any sans-serif font.
CSS provides much (all?) of the layout functionality provided using Netscape and Microsoft HTML extensions (even including BLINK!) as well as providing other features which cannot be implemented in HTML (and are typically implemented using inline images, which results in increased network load, and problems in indexing and access with non-graphical browsers). New features include small-caps, background images, big initials and drop-caps. CSS also enables layout features which can be implemented using HTML 3.2 to be done so more elegantly, such as indenting and outdenting text without resource to tables.
Looking ahead, future developments of CSS include:
ACSS will provide spoken rendition of HTML pages, including auditory icons (for elements such as links) and spatial audio (such as stereo effects for elements such as tables). An important aspect of ACSS is that a special "alternative" document will not be required.
Developments with web fonts are needed, due to the limitations with the <FONT FACE> tag which (a) can't do hyphenation, (b) can't do text transformation (c) not searchable (d) not portable, requires the right font and (e) makes multi-script documents difficult to render Web fonts will provide open, cross platform support with mechanisms for protection of intellectual property rights. They will be network friendly.
Following Chris Lilley's report and demonstration of stylesheets, given in the User Interface Highlights section, there were two presentations under Technology and Society highlights.
Daniel Dardailler spoke about JEPI, the Joint Electronic Payment Initiative. Daniel described that, in light of the commercialisation of the Internet and the variety of payment protocols already existing, a "meta" protocol was required to select the payment method. JEPI was set up at the end of 1995 by W3C and CommerceNet (see http://www.commerce.net/). The goals of the group are to produce a specification, a technical demonstration, reference code and an entry point into the second phase, integration with products. A slide showing the architecture of the scheme included PEP and UPP functionality in the client and the server, electronic "wallets" on the client machine and electronic banks on the server. PEP (Protocol Extension Protocol) will provide the HTTP extension mechanism, which will be used not only by JEPI, but also other developments such as PICS and security. UPP (Universal Payment Preamble)is a key component of JEPI, which will provide query/request payment alternatives, purchase information (amount, currency, brand, etc) and provide the transition to the next state of JEPI.
The next presentation in the Technology and Society highlights section was given by Philip DesAutels. The Digital Signature (DSig) project was set up to address issues associated with (a) signing active content (e.g. Java applets, ActiveX components), (b) publishing legally binding documents (e.g. price lists, prospectus, etc.) and (c) non-repudiation of documents. There are currently a number of incompatible and incomplete standards for digital signatures. The DSIg Initiative was established in order to provide a comprehensive solution to the basic problem of helping users decide what to trust on the Web. The initiative is based on a number of building blocks: digital signature, identity certificate, packing list and content label technologies.
The DSig project consists of two phases. Phase 1, which is due to be complete by 3rd February 1997, will develop specification for signature labels and a demonstration trust system. The implementation phase will commence on 4th February 1997. There are four teams in the implementation phase (1) Signature Label, (2) Scenario Development, (3) Trust System and (4) DSig Interest Group.
The DSig Project Deliverables will be:
DSig's Signature Labels differ from previous digital signature techniques in two ways:
The Digital Signature will have the following components:
Further information about DSig is available at the URL http://www.w3.org/pub/WWW/Security/DSig/Overview.html
After lunch Paul Resnick gave a report on PICS. PICS (Platform for Internet Content Selection) is a filtering mechanism which was developed in response to the US Communications Decency Act, which sought to restrict publishing of content on the Internet. PICS provides a mechanism for controlling the reception of information, rather than the distribution. This is more effective, more flexible and preserves free speech.
PICS has received much favourable media coverage, as well as government interest in Europe and Australia. A number of software products are already available (including Microsoft's Internet Explorer). There are several companies which provide self-rating services (such as RSACi - see http://www.rsac.org/ and SafeSurf) or third-party ratings (such as Microsystems and NetShepherd).
PICS is an official W3C Working Group. It has already produced two technical specifications which have W3C Recommended status.
At the PICS Working Group meeting held on 13/14 January 1997 the close relationship between the PICS system (for filtering content) and metadata formats, such as Dublin Core, was discussed. It was agreed to set up a group to look into the ways in which PICS could be extended to provide a general metadata format (e.g. through the incorporation of text strings in PICS labels). Another group was set up to look into performance issues.
The next presentation was given by Brian Carpenter and Harold Alvestrand. These two speakers talked about the IETF standardisation process and the ways in which W3C and the IETF could work together.
After coffee Jean-Francois Abramatic described the Process ERB (Editorial Review Board). The members of the ERB are Carl Cargill (Netscape), Wendy Fong (Hewlett-Packard), John Klensin (MCI), Tim Krauskopf (Spyglass), Kari Laihonen (Ericsson), Thomas Reardon (Microsoft), David Singer (IBM), Steve Zilles (Adobe) together with W3C staff Jean-François Abramatic (Chairman), Tim Berners-Lee and Sally Khudairi (Editor).
As W3C is now a large international organisation with a large budget a formal Process document is required to define its working practices. The Process document includes a Background to the W3C, W3C Activities and Documents, the mechanisms for initiating an activity, communications within W3C and with W3C members and relationships with other organisations.
Jean-Francois Abramatic's talk was the last formal presentation of the day. After the presentations there was a Reception hosted by RAL, followed by a Reception at University of Oxford Museum.
The second day was the Official AC Meeting. The day began with the Chairman's report by Jean-François Abramatic.
W3C now has a total membership of 158, with 64 full members and 94 affiliate members. In America there are 29 full members and 53 affiliates; the corresponding totals for Europe are 25/34 and for Asia-Oceania 10/7. Initially the membership duration covered the period from October 1994 until September 1997. Now membership lasts for three years (starting on the first day of the quarter).
Funding for W3C from membership fees totals $3.5M. A number of external contracts have been awarded including an external contract with the European Commission from February 1995 to September 1996 for 1.4 MEcus ($1.8M). A number of new contracts are to be submitted in the near future, including ones for promotion, PICS, I18N, JEPI and Accessibility.
The main expenses for W3C are staff, travel and equipment. A number of new members of staff have joined W3C recently, including Daniel Dardailler, Philip DesAutels, Joseph Reagle and Ralph Swick (Technology & Society domain), Arnaud LeHors and Ramzi Guetari (User Interface) together with Stephan Montigaud.
W3C has developed partnerships with IETF (covering standardisation of Web protocols) and CommerceNet.
Activities over the next six months will include:
The next W3C Advisory Committee meetings will be on June 18-19, 1997 (Tokyo), January 14-15, 1998 (west coast of America) and June 1998 in Europe.
Following the Chairman's Report, Nobuo Saito described the establishment of the third W3C host at Keio University. The East Asian host has 11 full members and 9 Affiliate members. The activities of particular interest to members in East Asia are still being discussed with the members, but are likely to include:
Tim Berners-Lee then gave his presentation entitled Dreams, Documents and Dependencies in which he put into perspective the various W3C activities.
Tim explained that the Web is intended as an empowering tool for the individual as a source of information and a means of expression. The Web will allow us to work as teams, groups and nations more productively and peacefully than before. The goals for the Web are:
Interactivity and Collaboration will require distributed authoring and versioning (the WEBDAV group), notification of changed resources (related to propagation, replication and caching activities) and knowledge representation and argumentation (which may be a development of metadata work).
Social Functionality covers privacy (a briefing document on Privacy and Demographics is being prepared), security, (confidentiality, non-repudiability and authentication are covered by an Interest group and the DSig project), payment schemes (the JEPI projects) and handling of intellectual property rights (a brief is in preparation).
This goal has a need for a HTTP extension in real time and a consistent architecture for this metadata.
From the above a number of common subtasks emerge (A) an architectural cleanup of HTTP, (B) a HTTP extension in realtime (PEP) and (C) a metadata architecture.
The cleanup of HTTP will distinguish between HTTP transaction generic parts applicable to new methods, functionality specifically associated with cachable methods (currently GET), "Entity headers" included in HTTP which describe the attached MIME part and other parts of HTTP such as authentication.
PEP (Protocol Extension Protocol) is important for the evolution of new facilities. It will provide an extension mechanism for HTTP.
A Metadata Architecture will provide a framework for assigning ownership and authorship of metadata, typed links and collections (for web maps, bookmark lists, etc.) It is intended to be a self-describing format with more power than current PICS labels. The metadata architecture may also provide the basis for knowledge representation.
Dan Connolly's review of the Architecture Domain highlighted developments with HTTP, PEP, document structures, addressing schemes, metadata, multimedia, object technology and caching and replication.
The Goals of the Architecture Domain are to present a simple, consistent view of information systems around the globe, automate information exchange and increase reliability and performance by stimulating development, facilitating consensus making and providing Implementation experience.
The Architecture Domain has links with the Technology and Society domain (regarding protocol support for social systems) and the User Interface domain (Distributed Applications, Dynamic Documents) as well as with the IETF and W3C members.
Workshops on Distributed Objects and Mobile Code (a Joint W3C/OMG Workshop), and Real Time Multimedia and the Web, a HTTP implementers' forum and symposia on 'A Comparison of DCOM and CORBA' and 'Distributed Authoring: Present and Future' were held during 1996. Documents on HTTP/1.1, XML and PEP have been produced and the Jigsaw and libwww 5.0a software released.
For HTTP/1.1 the specification reached Proposed Standard milestone in the IETF. Running Code is available (Libwww 5.0 and Jigsaw 1.0) and there has been Promotion and Dissemination in the World Wide Web Journal and in the HTTP Implementer's Forum. A Performance Study is underway which is looking into how much further current technology can take us. Further development work is needed in the areas of Distributed Authoring and Extensibility and Negotiation (PEP). A revised PEP draft is due out in February 1997.
A new IETF working group has been set up to look into addressing schemes. The group will produce a revised URL syntax specification and a process for new URL Schemes.
In the area of Structured Document Interchange phase 1 saw the release of the XML draft. Phase 2 will address linking and phase 3 style and semantics.
The area of Structural and Descriptive Metadata is driven by (A) the Sitemap proposal and the needs to support distributed authoring, search services and copyright and payment. There will be technical interactions with PICS, HTML and HTTP.
In the area of Real Time Multimedia the market is ripe for shared technology to prevent fragmentation. Multimedia Document Formats are needed for on-demand access, with parity with CD-ROM title development. The Network Transmission of Real Time Multimedia should be coordinated with RTSP work and with Multicast work (for Replication).
Similarly with Object Technology there is a need to Prevent fragmentation of HTTP from several incompatible tunnelling mechanisms. Investment in Object Technology needs to be preserved when building Web application and there needs to be increased automation for users and developers.
Market forces are evident in driving Caching and Replication (we are seeing new uses of HTTP (pre-fetching) and new protocols (PointCast, Castanet, multicast). W3C have carried out some experimental development (Jigsaw supports ICP). Also the HENSA cache logs are being analysed.
Vincent Quint's review of the User Interface Domain covers all aspects of the Web involving communications between users and computers. This Domain covers HTML, Style Sheets, Graphics, Fonts, Internationalization (I18N) and Amaya.
Achievements in 1996 (and in January 1997) are the Recommendations for PNG - Portable Network Graphics (October 1996), CSS1 - Cascading Style Sheets level 1 (December 1996) and HTML3.2 - HyperText Markup Language (January 1997) the Symposium on Web Internationalization and Multilinguism in Sevilla in November 96) and release of the Amaya software (which provides a testbed to experiment new features of web specifications on the client.
Ongoing activities include HTML, CSS, Maths, internationalisation and fonts.
Much effort is still needed in HTML. The HTML ERB tried to address too much. A coordination group and three new working groups are proposed which will cover HTML, Stylesheets and Object model and scripting. The proposed activities are:
New application areas will cover:
Jim Miller gave a report on the Technology and Society Domain. On-Going activities include:
Upcoming Activities include:
The final part of the day was devoted to questions and dicussions. A large number of questions were asked, which were grouped in the following categories: (1) W3C Process document, (2) Accessibility, (2) Activities and Technical Issues, (4) Staffing Issues and (5) Consortium Charter.
A number of questions were asked about the W3C process document. Many of the issues will be taken on board by the Process Editorial Review Board. The format of future meetings was discussed. It was felt that discussions, rather than presentations from W3C staff, were of particular importance. However the mechanisms for discussions with a large group (over 100) was let unanswered. It was agreed that the Advisory Committee mailing list could be used by members to raise issues and initiate discussions.
A number of people wanted to know how W3C would meet end user's needs, rather than the need of the computer industry, who were well-represented on W3C AC. End user representatives should attend W3C Symposia (such as the one that had been held on Internationalisation). End users would have particular interest in the activities of the Technology and Society Domain.
Members discussed the implications of the funding which was available to further accessibility with disabilities. There was some concern that, although work in this area was desirable, external funding could be used to influence priorities for W3C activities. It was agreed that a briefing document would be produced which W3C members could vote on.
There was a discussion on HTML developments. It was pointed out that there are still ambiguities in HTML 3.2 (e.g. file upload is ambiguous). Some felt that there should be an HTML extension mechanism (similar to PEP for HTTP). Dan Connolly would produce a HTML briefing paper.
There also needs to be a briefing paper on a streaming synchronisation language. It was felt that this should be a low priority for W3C.
There was also a need for a realtime collaboration mechanism. However it was felt that this should be done elsewhere (e.g. ITU or IETF rather than W3c).
A briefing paper on the ramification on a push publishing model, rather than the user pull mechanism.
The purpose of software such as Amaya, Arena, Jigsaw and libwww was unclear. There is an argument that the marketplace could be used to develop software. However it was felt out that W3C had a need to develop software to act as reference code, demonstrators and to provide test suites. For example the Technology and Society Domain would like to demonstrate browser software for the visually impaired. A W3C browser could be used to demonstrate what could be done.
The final question asked Tim Berners-Lee how the hypertext link architecture mentioned in his presentation would be implemented. Tim replied that PICS could be used, as the manifest (or collection) can be regarded as a multi-way link.