A Report From Extreme Markup Languages 2003

August 27, 2003

James Mason

The annual family reunion for connoisseurs of structured documents, "Extreme Markup Languages", gathered again in Montréal, August 4-8. Tommie Usdin (Mulberry Technologies) chaired, assisted by Debbie Lapeyre (Mulberry), Steve Newcomb (Coolheads Consulting), Michael Sperberg-McQueen (W3C), and me.

What brings us back to Montréal every year? It's not just the food, which can indeed be wonderful. What we like most at Extreme is the opportunity for networking, controversy, and intellectual challenge. From Usdin's opening keynote, "It's the Markup, Stupid", to Sperberg-McQueen's "Playing by the Rules", the latest edition of his eagerly awaited annual wrap-up, the focus was on what makes markup work and how we can stretch its limits.

Once again many of the papers were on the nature of markup itself, such as "First Thoughts on Modal Logic for Document Processing" by Allen Renear (University of Illinois) and Sperberg-McQueen's "Logic Grammars and XML Schema". Mathematical approaches to markup languages appeared in several papers: Robert Lyons (Unidex, Inc.), in "The Difficulty of Schema Conformance Problems", demonstrated that finding generalized conformance instances for many common schema languages is NP-hard or even undecidable, and Jean-Yves Vion-Dury (Xerox Research Centre Europe) and Nabil Layaïda (INRIA Rhô;ne-Alpes), in "Containment of XPath Expressions: An Inference and Rewriting Based Approach", presented a proof system, as opposed to the more usual model system, for analyzing a markup language.

William Kent, the author of Data and Reality, gave a stimulating keynote address on an important and controversial question: "What Is Identity?" This theme also appeared in many of the presentations. As has been the case at many recent conferences, Topic Maps and RDF attracted enough papers for a virtual track. The most theoretical paper in the track, Steve Newcomb's "A Semantic Web Integration Methodology", addressed the question of "identity" almost at the level of platonic forms, mentioning neither Topic Maps or RDF technology, though listeners in the know recognized parts of the Topic Maps Reference Model. (A related paper, "Curing the Web's Identity Crisis," by Steve Pepper and Sylvia Schwab, of Ontopia, appears in the September issue of interChange, the newsletter of the International SGML/XML Users' Group.) The other papers in this track were actually among the most application-oriented at the conference, such as "Content Repurposing with Topic Maps: A Real-World Use in Custom Publishing", by Nikita Ogievetsky (Cogitech) and Roger Sperberg (XTMaps), and "Taking Topic Maps to the nth Dimension" by Eric Freese (LexisNexis).

Every year there are a few papers which stand out simply because they demonstrate neat tricks with markup applications. Thomas Passin (Mitretek Systems) showed how he uses topic maps to manage bookmarks across multiple web browsers. David Birnbaum (University of Pittsburgh) presented a technique for correlating medieval manuscript collections. And Ken Holman (Crane Softwrights) delivered a real tour de force: generating XSLT stylesheets using the techniques of "literate programming".

A theme that ran through the whole conference, not just in the presentations but also in the conversations, was that some things seemed to be rushed into standardization before they were fully tested or their consequences understood. In "Typing in Transformations" Jeni Tennison (Jeni Tennison Consulting) looked at the problems introduced into XSLT 2.0 by the requirements for strong typing; her comments were the cause of a frequent lament that ran something like, "If Jeni's having problems, what hope is there for the rest of us?" A frequently discussed subject was the adverse consequences of letting XQuery have undue influence on datatypes, XPath, and other standards that are needed for more than just database operations. The number of papers looking at schema languages -- and proposing alternatives to W3 Schema -- seemed to reflect a general malaise. As an alternative, some even suggested extending DTDs to support datatypes, namespaces, and XSLT.

The preliminary proceedings for Extreme Markup Languages 2003 are online at More information about the conference, along with links to past conferences, is at the IDEAlliance site: