XML and Java(TM) Technology Q & A

★ wanayoo — archive 1999 http://java.sun.com/features/1998/12/Q_A.xml.html

XML and Java^TM Technology
An Interview with Dave Brownell
Part One

XML, the eXtensible Markup Language, is the universal syntax for describing and structuring data independent from application logic. This past February, XML 1.0 became a Technical Recommendation of the World Wide Web Consortium (W3C). From this milestone, numerous applications of XML are popping up far and wide - and more often than not they're using Java^TM technology. We asked David Brownell, designer of Sun's Java Project X, for some perspective.

It seems like early on, people were looking at XML as a HTML replacement. But now it has become much more, and is capturing the interest of mainstream business. What's going on here?

Dr. Charles Goldfarb was an inventor of SGML, from which XML evolved. He has an interesting way to put this: a "MOM and POP" scenario.
The document description approach comes from the "Presentation Oriented Publishing", or POP, community. HTML can do such publishing, but it hits a wall with large collections of content. XML was intended to expose the structure of the information, so it can be maintained better. But that's exactly what the "Message Oriented Middleware", or MOM, community of mainstream business applications does with its own data to ensure that the clients and servers can independently evolve. And that community in turn wants technologies to help it leverage the web.
So those two communities are moving towards each other. XML gives them a glimpse of something that will be very valuable: one way to represent data and either publish it or process it. Consider a purchase order; some folk need to print them, but other people need to process them. What a win to have just one representation!

It seems like people are doing a lot these days with XML and Java technology applications. Why do you think this is?

Well, part of this is Jon Bosak's influence. Jon is a Sun employee, and is the chair of the XML coordination group at W3C. He's been key to making XML happen; he founded and chaired the original XML working group. Early in the process he wrote a paper called "XML, Java, and the Future of the Web" which basically suggested that XML was the data that Java technology should be crunching on in Web clients. The developer community thought that was a good idea too!
Another part is that the Java language lets you be so darn productive. The statistics I've heard claim from thirty to one hundred percent overall productivity improvement; the reality I've seen is at the higher end of that scale. You really want that kind of speedup to be on your side, particularly if you're aiming to be early to market. Both XML and Java technology are very web-ready, so those web-aware XML applications are a good fit for Java technology. Plus, Java technology has internationalization support based on Unicode, matching XML.

Is there some special, natural affinity between Java technology and XML?

You bet. Java software is portable code ... XML is portable data.
When you combine code and data in the right ways, you get objects. So the pair of those gives you "portable objects" -- which are the only really effective way to design large scale distributed systems. No other pair of technologies comes close. And the Web is aiming to be such a system, so it needs them.
By themselves, neither of these technologies provides such a solution. To get portable data, you need to pick a representation that is usable in most programming languages ... like XML, which is structured text. To get portable behavior, you need a maintainable program standard that runs on most computer systems ... like Java language class files. Use Java and XML together, and you get both.

What are some of the more notable or interesting applications you're seeing developed?

E-Commerce applications are big. People expect XML to be a "slam dunk" in this area. Electronic Data Interchange (EDI) standards have long needed upgrading to get Web-aware, and XML (with its DTDs) provides an extranet-ready way to do the upgrade. You basically define DTDs for the documents which organizations interchange as part of their business dealings, like purchase orders or shipping notices. Then the organizations can use Web protocols like HTTP/HTTPS to implement application specific messaging. There are systems out there today, using XML this way.
Those DTDs could be standardized (vertical, horizontal) or specific to some business process two companies have. Supply chain integration is one area where this kicks in, also there are home banking and Internet shopping protocols being designed to use XML. You can define a workflow backbone using XML.

"Java software is portable code... XML is portable data."

Sophisticated Web sites are also using XML as part of a "write once, publish many" strategy. If the content is represented as XML, it could be repurposed in many useful formats. Cataloging can be done better, and metadata can drive content syndication such as what the ICE team is driving. The sort of "browser specific format" issue you hear so much about (including device-specific formats for devices that may have little more than a PersonalJava^TM platform for the runtime environment) is just one of many applications in this space.
And of course there's the cross-pollination between the two of those. Many of today's business processes are integrated with (or synonymous with) delivering information from a Web site. The customer browses content and seamlessly transitions to some sort of business interaction, such as shopping or customer service. People are using XML to do that today, even lacking support for XML documents in browsers.

Tell us about Java Project X. What's in it, and what kind of feedback have you gotten?

Java Project X is the codename for a set of core XML-enabling services, using Java^TM technology. It's now available as an early access release, which means among other things that the APIs and functionality are subject to change. The intention is that the functionality here should be sufficient to add XML support to existing applications, and to let new XML-oriented apps be written. We want feedback about whether we got that right; so far, that feedback is very positive, and quite constructive.
The package lets you parse XML documents, work with them in memory, and write them back out; basic stuff. The validating parser is the fastest one currently available, and is faster than non-validating ones from most other sources. People really like the speed, and also like its level of conformance to the XML specification. Plus, we're told that the package is "more accessible" than the others; it's reasonable to learn from. For example, just use the write() method on any node to print it as XML.
We use two common APIs, which have helped folk use this package out by dropping it in to some existing systems. The parser API is the Simple API to XML (SAX) 1.0, widely supported in the Java software/XML developer community. The object model is the Document Object Model (DOM); think of it as a parse tree. DOM is a Recommendation of the World Wide Web Consortium (W3C); that's as close to a "standard" as W3C gets. You can get started in XML very quickly by just parsing an XML document into a DOM tree, or building one and then writing it out as well-formed XML.
More sophisticated users can either bypass DOM and use SAX directly, or customize the DOM tree as it's built. Those "XML Beans" APIs are still being evolved. You can basically say to use a given bean to represent elements of a given type, and define methods and state as needed by your application.
You use those XML beans to associate behaviors with elements. Elements are the basic way you can extend XML; familiar elements from HTML would be TABLE or BLINK, but in XML you can define whatever elements you like. A CUSTOMER bean might be a good place to provide methods to track purchase orders, or to check spending limits. You can assemble a library of beans that support a particular XML vocabulary.

OK, so how do developers get the early access version of Java Project X?

An early access version of Java Project X is available on the Java Developers Connection. JDC membership is free, and you can sign up or login as part of the download process. Enjoy!

See Also