Menu

Conference Sketch

March 1, 2000

Edd Dumbill

This week, the XML-Deviant is at XTech 2000, the XML developers' conference. You might think that five days of seeing XML-DEVers in the flesh is a frightening prospect. And you wouldn't be altogether wrong.

Opening the main conference, Tim Bray promised delegates the "most intense experience ever of XML." That night, by the time it got to 10pm and after-dinner conversation was still an intense debate about elements versus the abstract idea of elements, I believed him.

XML Schemas Town Hall

What most XML-DEVers had no doubt been waiting for was the XML Schemas "Town Hall" meeting, chaired by Tim Bray. Members of the Schema Working Group were on stage and available for questioning. Though many had envisaged some kind of meta-bloodbath, by the time the meeting came around, feelings were softer towards the Schema Working Group. The most recent working draft of XML Schema no doubt helped this, as it included a helpful primer (the WG not having decided yet whether or not the normative pronunciation was in fact "primmer"). Additionally, Henry Thompson (editor of XML Schema Part I) acquitted himself well in tutorials and various ad-hoc sessions during the conference, earning much respect for the tremendous task that faced the Schema WG.

Ironically, some of the more aggressively anti-Schema faction stayed away from the meeting, concerned that they might get overly contentious had they attended. They needn't have worried. It seemed the sheer magnitude of the task of constructing the XML Schema spec was enough to awe most of the audience. As a result, the majority of the challenges were at a fine-grained level rather than at the fundamental-ideas level.

Dave Hollander, co-chair of the XML Schemas working group, made a tactical opening by anticipating some of the main questions that might be asked. He explained why the activity had taken so long, and also countered the concern that the spec might be overly complex. He did this by inverting the responsibility and handing it to the audience, saying that making Schemas universally understandable and usable was essentially their responsibility.

The 80/20 Point

Tim Bray opened the questioning with his own query, asking that since the Web has been successful by hitting the greatest 80/20 point ever, the perfect tradeoff between implementability and complexity, perhaps the Schema WG might themselves have missed hitting this mark?

He may later have regretted introducing this concept. Michael Sperberg-McQueen, co-chair of the Schema WG, said of course they've made mistakes, and they'll find out in the Candidate Recommendation period what they were. Yes, the Schema group had hit the 80/20.

In fact, cries of "80/20 point!" from the panel were to become a common response for the rest of the session in response to several questions about some more detailed issues. The fact that some felt like the Schema WG had hit the "99/1 point" instead went largely unnoticed.

Speaking for XML-DEV...

Peter Murray-Rust announced himself as "Peter Murray-Rust, speaking for XML-DEV" before he posed his question. He had to wait before getting any further as cries of "There's no microphone big enough!" and "Can you speak in six voices at once?!" came from the stage. Yet in a sense the likeable Professor of Chemistry was speaking for a large part of the XML community, those who write and use open source software. He asked about an issue crucial to his work in academia, XML editors. The emphasis during Candidate Recommendation (the "testing" phase of the specification) would be on parsers, he said, but what about editors? There was currently only one (freely-available) validating editor around now—would editors be part of the testing phase for XML Schema?

The panel didn't seem very impressed by the notion that an editor needed to be free. Dave Hollander asserted that editors were very complex bits of software and companies' business models dictated that such investment must reap rewards. Tim Bray countered, pointing out that operating systems such as Linux were also very complex bits of software, but were available for free! The day was saved when Henry Thompson indicated that he would be releasing a validating editor in several months' time, combining his work on editors and schema validators.

Velvet Gloves

So, by and large, the XML Schemas Working Group had an easy time of it. They had been handled, as one W3C member said, with "velvet gloves." A significant contributing factor in this is the sheer magnitude of their achievement. Questioners can either hand-wavingly say "It's too big" (response: "It needs to be") or pick on very low-level points, several of which the WG acknowledges as requiring more work.

Perhaps this need for additional work on fine-grained issues should cause the most concern. Several times during the panel, WG members said "This is a research issue," which leaves a nagging worry that anyone implementing XML Schemas might in fact be heading into the unknown.

Still, the Schema WG should be applauded for their work, and also for their recent responsiveness. If one can take the near-messianic enthusiasm of Henry Thompson as a guide, XML Schemas should be a great step forward for the world of XML.