This year's technical plenary was another wall-to-wall meeting-fest. I started with two days of XProc meetings. It was our first face-to-face and I think it went really well. Teleconferences, email, and IRC are effective ways to communicate, but some things just work better if you're in the same room arguing about the same drawings on the same white board. I haven't finished crafting what I hope is a synthesis of our discussions, but I think we reached consensus about the conceptual nature of a pipeline, or at least of a stage in a pipeline, and made some real progress on dealing with multiple inputs and outputs and iteration. We might have made progress on conditionals, but I'm not sure. There are some deeper issues about the nature of pipeline processing (pull or push, forward or backward chaining) that aren't resolved and conditionals seem to bear hard on those issues. Personally, I hope that we will eventually agree that some of those issues are implementation and optimization details, but it's not clear we'll be able to. At this point, though, it's not clear we won't either.
Next was the plenary day. If you aren't familiar with how these things work, the plenary week is divided into three sections. Monday and Tuesday are devoted to working group meetings. Wednesday is a full-day, all-hands plenary session with presentations, lightning talks, and panel sessions. On Thursday and Friday, we return to working group meetings. Part of the benefit of the plenary is that there are so many working groups meeting at the same time and place, it's often possible to bring different groups together to discuss common issues.
I think I saw most of the presentations on the plenary day. In no particular order, I recall sessions on data ownership and microformats; one of the “lightning talks” discussed single transferable voting (STV) as a way of achieving consensus; there was also a session on grid computing; and the semweb folks had a couple of slots, highlighted perhaps by Jim Melton's discussion of SQL and SPARQL. (Something of a reprise of his XML 2005 talk.) There were some interesting SVG demos. The XForms, HTML, and accessibility folks discussed a “rich web application backplane” and demoed some interactive web applications developed with more declarative markup. And somewhere in there Steven Pemberton, who's always provocative and entertaining, discussed the Sapir-Whorf hypothesis. We ended the day with a demo of chad, an IRC bot implementing STV voting, and the consensus was that the session on microformats was the favorite.
I've written about microformats before. They demo well and have enough apparent simplicity to be appealing. As one of my colleagues observed, it's like 1994 all over again, except this time instead of making up tag names, we're making up class values. I suppose, over time, we'll have to make up something like namespaces too. Class values are an extension point in the design of HTML, I wonder what the extension point in the microformats space will look like. What's a micro-microformat going to look like? Square brackets? Hyphens? Maybe there's an appropriate Unicode “extension” character?
Anyway, I'm not saying microformats are without value. It's not the approach that I wish we'd been able to take (I'd have been much happier with a story that allowed me to put elements from a foreign namespace in XHTML, using XML's more tractable extensibility points, and then styling those with CSS) but that's just the way it is. While I see significant problems with the microformats strategy, I'm trying not to sound like an ivory tower nay-sayer to bottom-up design.
I spent all day Thursday and Friday morning with the XML Core working group. We had good discussions with some of the C14N participants, some mobile folks, and hashed out a plan for incorporating IRIs into the XML stack. And then maybe going into maintenance mode. On Friday afternoon, I slipped away to join the second TAG meeting of the week (I was chairing during the first meeting and couldn't get away).
All-in-all a good week. I collected a few new name-to-face mappings and refreshed some stale ones. And I took pictures, like everyone else.
See you next time!
“No” claim the microformats crowd. They're only going to standardize a few common things, I think the argument goes, but that's not how these things work. Cat. Bag. Out of. We will wind up with microformats using the same class values in incompatible ways. But maybe there'll always be enough heuristics to disambiguate them.