Here is more of the fake coverage of the XML 2006 conference. Like Groundhog Day, we get to start from day one again and do it differently now that some more papers are up.
Paul Downey W3C XML Schema Patterns for Databinding
The slides are fun. Paul introduces the Barf Dance. This paper is an introduction to the W3C work on patterns. Guess what it uses to specify patterns? XPaths! Guess what is one implementation language they provide these in? Schematron. Hooray: first non-experimental use at W3C. One of the most contentious issues in XSD development was that one mob said “XSD is too big: it needs to be subsetted or modularized or optionalized or profiled” while another mob said “No, validity must always be validity: no subsets”. However, what the “Basic Patterns” work does is, in effect, define the profile that is supported by the broadest range of tools. When a committee keeps a standard fat when people are using it lean, they are just getting in the way: think SGML 1995. Creswell predicts….time is on the side of paths not grammars!!
Chimezie Thomas-Ogbuji The Essence of Declarative, XML-based Web Applications: XForms and XSLT
The thought in my mind after reading this paper is to wonder whether the ability to think in terms of pipelines and transformations with an extreme separation of concerns, as underlies the paper, is over-ambitious to expect of developers, or whether it is the new recursion: in the sense that a good ability to solve problems recursively is treated by some companies as an indicator of the level of a programmer. (Personally, I have made it my life’s work to rid my brain of recursive thinking: it simply does not gel well with text processing under a fixed-stack system like Java. For text processing, the recursive solution is always the wrong solution…unless your system implements the tail recursion optimization of course) Of course, what is interesting about XForms is that it is existing example of people using XPaths to bind elements to types: it is an example of the kind of thing we will increasingly see as people give up XML Schemas Part 1 and other grammars as being otiose. Thomas-Ogbuji has a neat idea of an abstract UI pattern and gives complete XSL listings for the code too.
Panel:Word and Open Office for XML Authoring.
Two presentations here. It is possible to author for XML in Word…yadda yadda. Marc Jacobson’s paper gets interesting on slide 7 where he mentions some of the problems with using Word2003 in some scenarios. “Cryptic error messages, so much so that both publishers allowed users to save with errors and fix later.” They clean up in other tools. As a vendor of other tools (Topologi) let me get on my hobby horse and point out that one reason you have to use cleanup tools like Topologi and the others is exactly that it is really difficult for grammar-based systems to provide meaningful error messages to laypeople. Especially if the error message is in terms of an XML document that the user is not actually seeing. Important point: “MicroSoft has stated that the do not intend to position Word against XML editors like XMetal and Epic.” This is a point I was making over the weekend to Patrick D. the ODF editor; we are both keen on trying to get good factual bases for judging when and where ODF or OOXML is appropriate.
John Parsons of XYWrite makes the point that there is highly-structured documents, moderately-structured and unstructured documents in a kind of pyramid. he important thing is locate “where you are on the pyramid.” The point is well-made, however I think the terms used are wrong: I think it conflates three different aspects of markup that can be treated separately: markup does three things basically—it labels text, it allows annotations of the labels or the text, and it allows the labels and annotations and labeled-data to be constrained in certain ways. Often “highly structured” is a euphemism for documents that are constrained. The labeling and annotation aspects, which can be available in an entirely unconstrained fashion, work on even “unstructured” documents.
The Schematron angle, also, would be that it is not a choice between lots of structure and no structure, but on being able to express the appropriate constraints at the appropriate phases in the lifecycle or workflow, and to be able to capture the constraints in language domain experts are familar with, and to provide diagnostics in terms that users will find helpful.
Stan Kitsis A study on the adoption and usage of XML Schema - its design and results
2000 participants were asked and up to 127 answered. So we shouldn’t treat this as a sample but a survey, mainly of MS users. Lots of interesting material, though some slides are impossible to understand, such as “Level of adoption over time.” Not many DBAs are working with XML Schemas, but those that are are much more likely to consider themselves experts at it than developers or other IT people. Interesting that people are blaming toolsets not XSD; But the fruit does not fall far from the tree. Especially interesting that Schematron and RELAX NG each have about 3.6% penetration, even among these participants, who are essentially all Visual Studio users; even though Visual Studio provides no special support for Schematron. If someone asked me, I would have guessed 0%. It doesn’t surprise me; even in the system integrator I do consulting for regularly, their lead developers are only just now figuring out about where Schematron fits in to their ecosystem. The survey shows more people are using schemas for more things and in more placed: the same thing is happening for Schematron too, as far as I can tell. Schematron tends to have two kinds of early adopters: for a start, the people who say “There has to be a better way than XSD” and then, secondly, the people who say “OK, I see where XSD’s limitations are now, I want to move forward.”
The previous coverage of day one is here