by Robert Caillau
The current crop of articles in this issue’s special theme is all about the Web. Or is it? I read the articles with interest, sometimes a frown, sometimes a smile and sometimes a sigh.
The most interesting observation for me was to see how much XML has become a tool of choice for experimenting. This is good news. When we were working on spreading the Web, HTML was clearly a second priority, and it remained in a pitiful state from 1990 to 1993. It proved impossible to get people at large to understand that SGML with its DTDs was a better technology, so we had to wait until there had been enough corruptions and style problems. That made people look for any good proposal, and XML was one. So that was a smile.
But the other thing I observed is that Java is the unchallenged choice for doing the actual horsework: getting some semantics into it all. Somehow semantics remains an untrappable ghost. When I studied programming languages, in the days when computers would still sit up, I always got confused as to where others defined the boundary between syntax and semantics. Notation does not solve the problem. Only execution does. The prime example of how real programmers define real semantics is C, that ugliest of all ugly systems. But as with many monopolies, one of C’s problems is that it works. Java should give us portable semantics. But does it? It’s all in the intractable classes. Yet, there is a lot of good stuff, and certainly programming feels orders of magnitude better in Java than in C (there are of course still those curly brackets). I always believed that ultimately the Web needed a programming language in the Turing machine sense. This was sometimes frustrating to communicate, and many, including friends, went off into the blue yonder of declarative stuff, often so esoteric that no sane person would spend her time on it. But, be it implicitly, we now have a programming language, unfortunately a proprietary one. So that was a sigh.
When I first heard of the Dublin Core metadata, I got enthusiastic and asked someone to prepare a talk on the state of metadata. An interesting discussion ensued, though I was quite annoyed at how people thought it could all be very simple. So at one subsequent meeting I showed them the lid of a pot of mayonnaise as sold by the local supermarket. It had on it the date of fabrication, the sell-by date and the best-before date. How could the Dublin Core people think they could do it with a single date, if just the local grocery store had used three for decades? So that was a frown. But I am greatly relieved to see how much influence RDF has now obtained.
Another idea (there I go again...) that should have been with the Web from the start was vector graphics. Before GIF, we (I) used PostScript for Web drawings. This was certainly 2D vector graphics, but PostScript was to some a word you could not say without conjuring up the devil, lightning and sulphurous smells. Then, to my great disappointment, GIF took over, but look, SVG is on the horizon, it is non-proprietary and we may yet see it replace PostScript. So there is another smile.
RDF, XML and SVG all at some level are re-inventions of wheels that have been turning in many places for a long time. It is necessary to re-invent once in a while, because only that way can we integrate, improve, get the youngsters enthusiastic, and become non-proprietary. Smile! (SMIL?)
I discovered a number of cats. Cats are independent agents over which you have no control. If they like you, they may spend some time sitting on your lap and purr, giving you an impression of cosy well-being. There are several articles on systems that are supposed to do things for you. But I see cats (mind you, I’m owned by a cat myself).
Finally, there are some very interesting reports on techniques for improving efficiency and for applying the Web in social interaction. The range of this collection is wide, if you are an eclectic reader as I am, you will find a lot of stimulating stuff here!
Robert Cailliau – Web Communications & Public Education,
Education & Technology Transfer Division
C E R N, European Organization for Nuclear Research