Related link: http://www.wired.com/news/business/0,1367,46383,00.html

I’m happy to see that I’m not the only one who finds many conference expositions to be largely buzzword-compliant at the expense of technical information.

Farhad Manjoo’s Wired report from LinuxWorld looks at what happens as geek culture is replaced by corporate marketing:

“This year the marketers have attacked. Now, it’s virtually impossible to walk around LinuxWorld without hearing pitches from company reps extolling the benefits of Linux for ‘mission-critical applications in the enterprise,’ or some such drivel.“

There are show floors filled with technology and people who understand it, and then there are show floors where the only meaningful activity seems to be collecting toys.

Fortunately, Manjoo managed to find some intelligent life on the show floor - Linux hasn’t rolled over yet!

Do you attend conferences seeking the buzzword-compliance required for mission-critical enterprise applications?

Related link: http://internet.conveyor.com/RESTwiki/moin.cgi

Mark Baker announced a Wiki for REST (REpresentational State Transfer), a description of the Web’s current architecture that raises substantial questions about the architecture of Web Services.

Related link: http://aspn.activestate.com/ASPN/Mail/Message/xml-dev/755432

In a discussion of models for Web Service transactions, Walter Perry examines the historical background of the two-phase commit model used in many large-scale information systems and questions whether two-phase commit retains its relevance for XML transactions.

Perry proposes a “souk” model in contrast to the “shop” and “cartel” models, arguing strongly that the gain in flexibility opens new possibilities worth leaving the security of two-phase commit behind. It’s dense reading, but raises questions worth pondering even if you don’t like Perry’s answers.

Is raw XML’s freedom of expression and structure too dangerous to be used in business environments?

Related link: http://linuxtoday.com/news_story.php3?ltsn=2001-08-17-016-20-OP-CY

Eric S. Raymond suggests that we examine flerbage - “I can behave in the confidence that nobody will take my life, my physical property, or my time without my consent” - to determine the merits of different licensing schemes. I’d suggest that flerbage is a distraction from the real issues, which have much more to do with property than with nonsense words.

Although Raymond claims that he is “not prejudicing the discussion by assuming that the software I write is my property”, the argument he makes is set up to protect property from the outset. In order to prevent infringements on life and liberty, Raymond proposes that we protect an understanding of property, rebuilding the Lockean trinity.

After exploring how liberty isn’t substantially affected by current software practices, including open source, he posits a world in which proprietary licenses are outlawed, and the impact on users and developers. First, users:

“As a user, my flerbage doesn’t change. I never wanted to issue software under a proprietary license to begin with, so the new license doesn’t touch me. “

Sure, the life and liberty situation hasn’t changed for users, but that limited viewpoint doesn’t explore how much users gain from the benefits of not having to deal with proprietary licenses. For developers, Raymond describes an infinitely more dire scenario:

“But as a developer, things are very different now. If I walk up to someone and offer them the same proprietary license that I did before the law was passed, police may come to my house to drag me off to jail, or kill me if I resist arrest. My flerbage has seriously decreased. “

For starters, Raymond forgets that lots of developers - those who didn’t want to use a proprietary license to start with - have a major gain outside of his tightly-qualified “flerbage”. They have access to information they didn’t have before, freedom from fees, and the right to change code to meet their needs.

For those poor oppressed developers who deeply want to release code with a proprietary license, the penalties seem distinctly harsh. Death for resisting arrest? Hasn’t Raymond ever heard of civil proceedings, where arrest and death aren’t part of the equation? Hasn’t he noticed that Microsoft executives continue to occupy and enjoy their houses and offices even while the company has been given a rather powerful legal black eye?

The only people presently facing a threat to life and liberty are those trapped in the ugly maze of proprietary intellectual property - notably Dmitri Sklyarov. Some of us writing open source code fear the prospect of patents that could conceivably drive us out of the United States to carry on our work. Involuntary relocation to avoid (potentially bogus) property claims isn’t exactly fun. Coming up with nonsense words and scenarios that merely reinforce the conception of software as property won’t help those of us who find that conception to have painful practical effects.

Is the value of A’s intellectual property intrinsically worth the cost A may impose on B’s freedom?

Related link: http://www.extrememarkup.com/extreme/

Sometimes it’s worth digging deep into complicated problems to see how they can illuminate the kinds of problems we deal with every day. Extreme Markup Languages 2001, which just ended, was a conference which threw an enormous amount of light even on issues about which there can’t be easy agreement.

While the overall conference program suggests a heavy dose of Topic Map, RDF, XSLT, and schema esoterica, there were a number of themes which ran through pretty much all of the presentations and kept participants at this fairly small (about 150 people?) conference talking both on and off the conference floor.

• The usefulness of theory - This conference and its predecessors emerged from a community deeply interested in theories about information and (for some) knowledge. A large number of the presentations began with a foundation in theory but demonstrated immediate practical applications.
• The curse (or blessing) of ambiguity - Different presenters had very different takes on what consituted useful data. Some wanted strong ontologies and enforced prior agreement about meaning, while others wanted to see how far we could get without understanding or agreement about meaning. Some of the best presentations explored the possibilities of multiple and overlapping interpretations of meaning and even content.
• The challenge of granularity - However you feel about the need to hammer things down into precise containers, it remains clear that there is plenty of room for disagreement about what constitutes adequate categorization, markup, or program design.
• Reconciling models - Even compatible models of information may require human intervention to efficiently reconcile their contents. While many of the conference attendees were excited about automated information processing, the importance of people for sorting out some kinds of issues remained clear.
• The diversity of projects - While some presenters were talking about high-level categorizations of constantly-changing business information, others were dealing with issues like markup for every word or letter in a document, and there was discussion at lunch about markup for every stroke in a letter. Because developers came from some very different worlds of requirements, the variety of ideas was vast.

Dealing with all of these issues was exhausting, and I think it’s reasonable to suggest that no grand consensus was reached in the course of the conference. On the other hand, it’s nice to come home from a conference with a new batch of issues to think about, write about, and even implement.

Related link: http://makeashorterlink.com/index.php?U31331A0

This Java programmer is downright jealous of the Perl tools for XML.

At OSCON a few weeks ago, I watched Kip Hampton give an overview of the Perl modules out there. and it left me thinking that Perl has a much richer set of choices. Not that all the choices are necessarily wonderful, but there are plenty of days I’d like more choices.

Perl folks (and Python folks as well) seem to have moved past the event and tree dichotomy to do interesting work with partial trees, events feeding into and out of small trees, and various other handy pruning and growing approaches.

On the Java side, events can feed into trees which can feed back into events, but there isn’t much creative recombination of the two. There’s DOM (tree) and SAX (events) and that’s the world of well-known options Java also has JDOM, but while it’s certainly less unwieldy, the overall approach isn’t that much different from DOM.

Working with SAX Filters is a lot of fun, but sometimes I end up building my own mini-trees. Maybe I’ll refine those sometime, and see if I can’t add some Perl-like diversity to the Java XML toolkit.