Position Paper for The W3C Processing Model Workshop

Ken Wilner - Engineering Fellow
David Cleary - Principal Software Engineer
The Progress Company
1 Indian Head Plaza
Nashua, NH 03060


The Progress Company Overview

The Progress Company provides application development and deployment tools used for the creation of business applications. These applications can be client-host, client-server, application server, and HTTP/HTML based. Increasingly, XML is becoming a very important technology within our core product line. We currently utilize DOM Level 2 and XML Schema 1.0. We plan on utilizing XMLP, XSLT, and XPath in the very near future.

The XML Tower of Babel

Over the past year, a number of new XML Recommendations have been created by the W3C. At the top of the tower is the XML Schema Recommendation and the Post Schema Validation Infoset (PSVI). XML Schema brings much-needed features to XML for use in business applications. However, it also brings much complexity with the creation of the PSVI. Today, current Recommendations are being updated and new specifications are being developed that can take advantage of the PSVI. However, there is no overriding authority that determines how all these Recommendations will work together. This can possibly lead to serious interoperability problems in the future.

Output of The Processing Model Workshop

The goal of The Processing Model Workshop should be to develop a roadmap of where all these W3C Recommendations fit in regards to processing an XML document. At the center, you have an Infoset that gets extended with an optional XML Schema to create a PSVI. Some Recommendations such as XMLBase and Xinclude are used in creating the Infoset. Other Recommendations such as DOM and XSLT should work against a generated PSVI, so happen after XML Schema validation in the processing model.

The cornerstone of all these Recommendations are the Infoset and PSVI. Assuming that XML Schema (or other Schema languages in the future) is the only way to generate a PSVI, then you can split all the Recommendations into two groups. The first are those that generate an Infoset . These clearly come first when processing an XML document. Next comes Schema validation and generation of the PSVI. Finally, you have the group of Recommendations that access the PSVI and provide information to applications. As long as all current Recommendations and future Recommendations follow this architecture, interoperability problems should be kept to a minimum.