Portrait of Edd Dumbill, taken by Giles Turnbull

Subscribe to updates

Feed icon Atom or RSS

or get email updates

What I make

expectnation
a conference management web application


XTech Conference
a European web technology conference

Three cheers for the XML community

Yesterday I visited Cambridge to speak at XML Open about my Description of a Project work (DOAP). It was wonderful to see many friends and colleagues from the XML world, and I regretted that I'd not been able to attend the conference for the full three days.

XML Open was concentrated on the juncture of XML and open standards with open source. An appropriate topic, as XML's success is predicated to a large degree on the availabilty of the basic processing tools as open source.

My friend Dave Beckett gave a talk on Redland, his RDF processing toolkit. He focused not as much on Redland itself, but more on the issues of developing it as open source. Such considerations include which programming languages to target, which licensing choices to make, and so on. Dave's put his slides online.

The closing talk was given by the entertaining Sean McGrath. In a wide ranging talk that certainly left behind more questions than answers, Sean recommended searching Google for "&" as an indicator of human frailty in our ability to process markup (or create markup systems, I suppose).

Sean's talk explored the tension between simplicity and complexity, and the role of elegance. That is, complexity in the sense of elegant higher-level abstractions. XPath is complex in that it abstracts many simpler operations into an elegant expression. The alternative to XPath would be a bunch of one-off manipulations of the DOM.

Some might say, however, that we don't need XPath. Effective XML document manipulation can be achieved by other means: we could just make a GUI to design the expression, and the GUI can spit out the necessary DOM-traversal code. No new language to learn then.

Sean observed that using GUIs in that way was inimical to sustained development, as you ended up with an unstructured pile of low level operations. GUIs that only preserve the higher level notions onscreen are going to lead us into trouble if we were hoping to create a maintainable system.

I'm sure I've not done Sean's talk justice there, but you get the idea. Anyway, I was sitting in the audience listening to this talk and feeling smug. I felt I was in the right here, preferring elegant expressions to "symbol shovelling".

Then it struck me that perhaps I had been guilty, at least by association, of relying on the "tools will solve it" attitude where RDF was concerned. In our keenness to get RDF out there -- and I will never climb down from that position -- I think we ran the risk of forcing visible low-level RDF on people just to prove we were using it. In other words, I had been afraid that the use of higher level expressions might not have furthered the cause.

Sean cited various technologies in the XML world that were good higher level notations. Among these is RELAX NG, and in particular the compact syntax, for describing XML schemas. RELAX NG is one of those technologies that people instantly take to. I'm very glad that Mono has good support for it. (It was also great to see Mono's Atsushi Enomoto at XML Open.)

On the RDF side, notations such as N3 and N3-subset Turtle are helping us create higher level expressions of data. I admit being against N3 early on, thinking it was moving on too far before we'd sorted more basic things out, but I appreciate its value more now. If you've wanted to play a bit with RDF but can't get your head round the RDF/XML syntax, I recommend installing Redland and trying out parsing some of the Turtle examples with the raptor tool.

The book

O'Reilly had a book stall at XML Open, and I was very glad to hear that Mono: A Developer's Notebook was the best-selling book during the conference. Very amusing, for an XML conference!

Again, some people complained to me that they couldn't find the book in high-street bookstores. I've had several conversations with people at O'Reilly about this, and the reason seems to be that book buying decisions for chains such as B&N are generally made on the basis of which books have sold well in the past. Mono doesn't really fall under any existing category very well -- most buyers are unlikely to make the connection with .NET. Furthermore, if your local store hardly stocks any programming books anyway, they're very unlikely to want to get in a book on Mono.

This isn't so great for Mono, though, as it really reduces the possibility of readers browsing through the shelves and discovering Mono that way.

The way that people can help best is to ask for the book in their local stores and get them to order it. If lots of requests show up from stores, then it really helps make a case to the central buyers for ordering it.

blog comments powered by Disqus


You are reading the weblog of Edd Dumbill, writer, programmer, entrepreneur and free software advocate.
Copyright © 2000-2012 Edd Dumbill