Extreme Markup Languages 2006

Volume 9, Issue 79; 17 Aug 2006; last modified 08 Oct 2010

Highlights from my favorite XML conference.

This year's Extreme Markup Languages lived up to all my expectations, as did Extreme 2004 and Extreme 2003, for those keeping score. (I didn't make it to Extreme 2005.)

I started the week by presenting an XSLT 2.0 tutorial (which I think came off pretty well) and then, given that I'd just come out of a face-to-face meeting, spent a fair bit of time struggling to get the slides ready for my late breaking talk about the XProc work.

I thought I might have achieved the latest of all possible late breaking talks, but John Cowan beat me, presenting on Wednesday some ideas about LMNL that he'd worked out with Jeni Tennison and Wendell Piez the day before.

Other highlights and things I'll be reading in the proceedings: Eduardo Gutentag's opening keynote (in Tommie Usdin's absence) Intellectual property policy for the XML geek, Dimitre Novatchev's exploration of higher-order functional programming in XSLT 2.0, Alex Brown's presentation of Frozen streams: an experimental time- and space-efficient implementation for in-memory representation of XML documents using Java which elicited lots of interesting comments, Felix Sasaki's review of the similarities between architectural forms, CSS, and RDF, Adam Souzis's work on RxPath: a mapping of RDF to the XPath Data Model (drat, I meant to chat with Adam a bit, but our paths never crossed), Sam Wilmott's intriguing (if a little pointless, to me) exploration of a non-XML syntax for XSLT, Michael Sperberg-McQueen's discussion of overlapping markup and “rabbit/duck“ grammars, Tomasz Müldner's presentation of multi-encryption for securing XML documents, Mario Blažević's Streaming component combinators (work closely related to XProc), and Harry Halpin's work on XML versioning. Also, I participated in a Tag set promulgation panel and Liam Quin led a discussion of Microformats: contaminants or ingredients?

As before, those are just the most interesting of the talks that I got to see, there were others. I was deeply disappointed that the latest round of travel problems forced me to miss Jeni's Datatypes for XML: The Datatyping Library Language (DTLL).

You know, I expect the whole proceedings is worth a read.

The polemics were fun too. Joseph V. Gangemi argued that XML dropped some SGML features that were really important for publishing and we should demand them back. That's not a practical suggestion in this or any alternate reality that I can easily imagine, but he's right that publishers should do something to achieve standardization and vendor support for XML solutions to the problems those features solved. Martin Bryan exhorted us to DiSRuLe and make up our own tag names! (This was really a problem that needed solving, was it?) And Walter Perry suggested that we should drop ACID and take SALT instead.

See you next year! (Yes, there will be an Extreme 2007!) Especially you, Tommie! We all missed you and hope you're feeling better!


I think you may be shortchanging Martin's talk on ISO DSRL!

In fact, it does solve at least one completely pressing problem that prevents some publishing people moving beyond DTDs: it provides mappings for undeclared entities, so that people can author with ISO character references if they want to and produce well-formed XML.

It does this in the context of a general name and value mapping tool, which should be implementable on top of XSLT 2.

—Posted by Rick Jellfife on 18 Aug 2006 @ 12:39 UTC #

Apologies. It wasn't my intent to shortchange Martin. Perhaps the polemic style of the presentation distracted me. I don't recall mention of mapping undeclared entities. I'll take a closer look at DSRL.

—Posted by Norman Walsh on 18 Aug 2006 @ 12:46 UTC #

Yeah, DiSRuLe itself does provide entity remapping, but an implementation won't be able to actually deliver unless it relies on an XML parser that reliably passes information about undefined entities to the application.

In particular, SAX doesn't provide any way at all to cope with undefined entity references in attribute values, since an attribute value is a string, not a structured object.

—Posted by John Cowan on 21 Aug 2006 @ 02:39 UTC #