At XTech 2000 in San Jose this week, Apache Group co-founder Brian Behlendorf walked into a den of standards wonks and preached the message that implementation of real tools was a more important goal than (seemingly endless) debate and reiteration of XML standards. He received a mixed reception.
"Let's not ever write an XML processor again, unless we have to," he told an audience of about 400 XML developers. "Let's improve the ones that are out there."
Behlendorf was there to promote the Apache Group's efforts in the XML arena, which include bringing several of IBM's XML tools into the open source community. (See sidebar for a description of these tools.) The gist of his talk was that creating a reference implementation of each of these tools, administered by an open source development community, would preserve the standard implementations, prevent splintering, and prevent any one vendor from being able to dominate a niche and unduly influence standards.
Apache's XML Suite
Xerxes. XML parsers in Java, C++, and Perl.
Xalen. XSLT transformation engine (formerly IBM's Lotus XSL).
FOP. XSL formatting objects, in Java.
Cocoon. XML-based web publishing, in Java.
For more information on all of these, visit xml.apache.org.
While acknowledging the important role of standards in securing agreement on what a technology should do, Behlendorf said the standards process usually results in pseudocode, allowing developers to interpret it differently.
He prefers developing a reference implementation of the tool that the community can cooperatively develop. "Let's put it to a compiler, and see how it interprets it."
Behlendorf stands on the success of the Apache project, a community effort that has produced the world's most popular server. Some believe that without Apache, Microsoft and Netscape could have dominated the server market, muddying the protocols for HTTP in the same way that they created multiple standards for browsers.
"Just as we played a role in keeping HTTP open, we saw there was a need for us to play a similar role [with XML]," he said.
Along the way, Behlendorf offered his explanation of why community development seems to have worked better than reusing software components, which was a popular idea in the late 1980s and early 1990s. He said component libraries didn't work because they typically took the form of each large software developer "dumping" an enormous library on the community of users. Others would release libraries of similar components. It was up to each developer to stitch together a solution based on an incomplete knowledge of the components in each of these libraries.
For More Information:
The Apache Group's XML project
XML.com: Apache Software Foundation Launches XML Project (Nov. 10, 1999)
"Let's encourage everyone to use the same components" and focus efforts on improving that reference implementation, he said.
But in the end, whatever the efforts of the Apache Group and the vendors Behlendorf works with, the success of a unified XML approach will depend on the several hundred key developers who hash out the standards -- many of whom were there in the room at the Convention Center.
Behlendorf, of course, knows this: "Where it goes, is up to you."
Also at XTech 2000: Eric Krock Introduces XUL
Copyright © 2009 O'Reilly Media, Inc.