W3C Town Hall Meetings at XTech

XTech 2000 Developers Conference was held February 27 - March 2 in San  Jose, California. Hosted by GCA/IDEAlliance, XTech 2000 brought together  experts, developers and technical managers for an intense look at the current  XML technology and applications of the standard.  But the conference did not  end with keynote presentations and technical sessions. During the evenings,  technical working groups were hosted by IDEAlliance.  In addition, a new  feature of XTech, "W3C Town Hall Meetings" were held at 8:00 pm on Tuesday  and Wednesday evenings. At these meetings, representatives from the W3C's  XML Schema and XML Query Working Groups took the stage in two separate  open-mike forums.

XML Schema Working Group

The editors of the XML Schema Working Group began with key points about  the new specification:

• XML Schema meets the needs of two converging strands: database  applications wanting to become enriched by textual information and  classical document applications wanting to become smarter and pull in  other kinds of structured data
• XML syntax has been selected as the basis for the XML Schema  Language so that common tool sets can apply to both instances and  schemas
• We deal with infoset not the character stream, there is a separation of  transfer syntax from conceptual level so that you can schema validate a  DOM

A discussion of the time-line for the development of the specification followed.  The editors stated that their intention is to go to comment end of April/May and have at least 3-4 months comment period. The intent of the long time period is to allow for some validation and implementation.

A general concern that arose several times about whether the specification had created something too complex to be implemented, whether or not there was an 80/20 point that should be considered instead. Michael Sperberg McQueen's response was to say that no one has much experience with mixed DB/text schemas, and we need to have something to start working with to gain that experience.  Also noted was the fact that at a minimum IBM, Oracle, and the Apache Xerces project have been tracking the spec closely and so schema processors should be available in short order. 

Other concerns that were raised included:

• Can guided authoring applications work with a schema in place of a DTD?
• Can we use a schema to specify business rules like "if element A and element B, then not element C, else C required?"
• Isn't string data type is under-defined without a
  character set and a collation sequence?
• Can we express open enumerations (where you have some of the possible values enumerated, but allow extension)?

XML Query

The second town hall meeting was hosted by the XML Query editors.  Paul  Cotton from IBM began with a short history of XML Query.  According to Cotton, 1998 was still the age of "Roll Your Own" query language. The  W3C Query Languages Workshop in December 1998 was the pivotal event.  At the workshop 66 papers were presented and there were 98 attendees from around the world. Many  came away from that workshop with the feeling that there was a need for a working group on XML Query. Early in 1999, Paul Cotton wrote a draft charter, which was approved in August 1999.

The next presenter was Don Chamberlin.  Chamberlin participated in the development of SQL some 25 years ago in the database industry. At that time there was no universally accepted query model or database model. As a result,  each database vendor had a proprietary language.  This meant that  applications were not portable and it was difficult to exchange data. The  Relational model and SQL formed a framework that made applications portable usable across all database vendors. 

Chamberlin described the current situation as a bigger revolution than SQL.  "Not since the library of Alexandria was burned has all the world's information been stored in one place. But the Web could be the location for storing the world's information. Our whole civilization is built on the notion that information is scarce and expensive, but with the Web information is cheap and ubiquitous. All information is everywhere all the time. To make this come true, we need a data model and a query model."

Michael Rys and Jonathan Robie also presented their insights into why a query language for XML is so critical.  Questions/Discussion followed.  One of the major concerns that was raised was the size of the working group.  According to Cotton, the working group is made up of "more than 35 W3C member organizations, about 50 individuals, 30 or more at face to face meetings."  He believes this group to have the right composition to do the work, but whether Cotton can control "fifty mustangs who want to go in their own directions is best answered by someone else!"

The XML Query activity seems to be off to a good start.  Note that a requirements document was posted this month.  Learn more about that document in the "W3C Standards Update" in this issue.