XML 2006 started with a reminder of SGML 1996, where eleven brave people suggested a major simplification of SGML intended to make it usable on the Web, which became XML. I think they hoped to take over the Web specifically more than has actually happened, but it’s clearly made its way to the world. Jon Bosak, chair of that group, will be closing the conference with reflections on the past decade.

(Disclosure: I was on the Planning Committee for XML 2006, and was Track Chair for XML and the Web.)

Next up was Oracle architect (with a parking space) Roger Bamford. He started by reaching even further into the past, talking about the venerable 3270 terminals and the database and network support they needed. Web architecture today has changed, but “if you look at the workflow - it’s the same.” XQuery changes some parts of that workflow, and opens up some new possibilities, though I think it may be yet another long while before much of it percolates through.

I introduced the first two sessions of the XML and the Web track. Fabrice Desre first looked at ways to create small XQuery database applications inside of Mozilla, with a nice sidelight on using REX to avoid a lot of potentially gigantic DOM scripting tangles. Next, Robin Hastings visited the Magpie library for RSS processing in PHP, showing how it actually has become simple to grab data from one place and reuse it inother. To that extent, so of the XML magic seems to have percolated down to the real world.

In the afternoon, XForms overfilled the room, with probably 100 people listening to Chimezie Ogbuji explain what he learned in deploying an XForms application at the Cleveland Clinic. XForms has been promising for a long long time, but seems to get a lot of flak from browser vendors uninterested in implementing it. Ogbuji got around the implementation question using the FormsPlayer plugin, and around a lot of other issues by using XSLT to generate his forms. Next up, John Boyer explored the dynamic programming model XForms creates, and a gap in it that needs filling - a brilliant snapshot feature that needs digital signature support to ensure snapshots are trustworthy.

Patrick Chanezon next talked about Google Checkout - a REST API that lets users run a shopping cart through Google, and merchants can then fulfill their orders. The transaction flow - as with all shopping carts - is potentially complicated, and includes pieces for situations like calculating shipping and handling refunds. In a lot of ways this isn’t something I’d think XML was essential for, but it does seem to ease the communications. There was a question about the relationship to existing work - RosettaNet and so forth - that concluded “or are you making this stuff up? What are you foisting on people?” It was asked in a (mostly) friendly way, though…

The afternoon concluded with “Resistance is Futile: You Will Store XML”, by IBM’s Susan Malaika. She was “glad to be at a conference where everyone likes XML, or at least I hope they like XML.” After ten years of XML percolating through the world, it’s time to stop thinking of XML as something that happens only on the edges of systems, but rather as a core way of handling information. She called for an end to normalizing XML documents into tables, which may or may not preserve any sense of the original documents. It was kind of a funny presentation for me, as her position is one I thought made sense when I first got involved with XML, but it’s not an obvious position if you’re not as XML-centric as I’ve tended to be.