Minutes from the Tech/Integration Subgroup

1. HTML, SGML and Hytime
   • SGML and HTML - we need a flexible extension mechanism for HTML, and tools for graceful down-conversion. The situation is analogous to LaTeX, which had user-defined macros which led to standard styles being established locally and within comunities, but poor support for interconversion.
   • Presentation vs content. SGML makes the distinction, common HTML usage blurrs it (as do WYSIWYG wordprocessors etc). CERN's proposed style sheets will help.
   • External link bases. This was felt to be a solution rather than a problem, and was not therefore discussed at this stage.
   • HyTime has much it can offer the Web.
2. Dynamic Documents & Scripting Languages
   Searching (not much discussion, but it was felt that as well as a general framework we need standard tools for the particular case of HTML).
   • Continuous media (this is regarded as `doable', but still a research area and demanding on resources).
   • Dynamic documents. Steve Pemberton explained Views, which incorporates a functional language for describing display of dynamic documents (and dependencies between objects). Also saw use of HUSH: the tag in HTML has TCL attribute specifying tcl script which executes in a (customised) browser window.
   • Scripting languages - several different opportunities for scripting languages in the Web architecture were discussed.
   • Client-side CGI - we would like symmetrical CGI (client and server) to facilitate client-side computation. For example, a student selecting a program for execution might result in server-side or client-side computation.
   • Possibility of peer to peer (browser to browser) communication - a `Web-lite'?
   • Need for support of persistent processes behind CGI (eg for databases, searching).
   • What is a legitimate requirement of WWW and what is misuse - eg is it legitimate to use it as a transport for a programming language in order to distribute computation rather than information?
3. Demonstrations
   • Views
   • Hush
   • Linkbases
   • persistence

Summary:

• There is a lack of standards for interoperability, mainly on UNIX. Perhaps we can go via the consortium to influence this?
• Presentation vs structure. We are worried about the expansion of HTML. Flexibility and generality are required. HTML should not become non-permissive.
• We need to integrate with SGML and HyTime but should provide tools to shield users from (for example) having to deal with SGML DTDs. We should contact SGML community directly.
• We would like a general framework for scripting languages.
• We would like a client-side CGI (but we know a heterogeneous solution is difficult).
• We would like support (in the server?) for persistent processes.