Home > Articles > Programming > Java

XML at JavaOne: Ten Years After

  • Print
  • + Share This
A decade ago, at the first JavaOne, XML was a non-entity. This year, XML was everywhere, and it has come a long way. Frank Coyle looks at the evolution of XML and Java, including the latest developments and announcements.
From the author of

This year's 2005 JavaOne Conference in San Francisco provided an opportunity to celebrate Java's tenth birthday. At the first JavaOne back in 1995, Java had just emerged from Sun's labs as a hot, new object-oriented language with the mantra "write once, run anywhere". That first JavaOne drew over ten thousand attendees flocking to learn all things Java and to figure out how to apply Java to the emerging World Wide Web.

Ten years later, as I roamed the Moscone Center absorbing as much as I could about the current state of Java, I was struck by the fact that a decade ago at that first JavaOne, XML was a non-entity. Even after XML appeared on the radar screen in 1998 as a W3C recommendation, it was several years before the Java community realized its importance and impact.

The first major interconnect between Java and XML occurred in March of 2000 with the release of JAXP, the Java API for XML Parsing. JAXP gave Java developers the ability to process XML documents using either SAX (the Simple API for XML Parsing) or DOM (the Document Object Model) parsers. But parsing only scratched the surface. It wasn't until 2002 when Sun announced JWSDP, the Java Web Services Developer Pack, that Java, XML and Web services were brought together in one multi-component download. JWSDP gave developers a set of APIs and reference implementations that let Java applications engage XML, and create or consume web services from within enterprise Java applications.

JavaOne 2005

This year at JavaOne, XML was everywhere. On the exhibit floor, a company called Justsystem Corporation occupied major exhibit floor real estate, promoting its xfy product for managing and building complete end-to-end XML-based applications with XML. Billed as an application suite, xfy includes a variety of Java components for processing XML vocabularies called XVCDs. An XVCD is an XML description of the data processing steps necessary to build an application. Programs such as xfy are indicative of the growing trend to push more and more of the programming task into declarative XML. This is, of course, no surprise to Java developers who have been wrestling with their web.xml configuration files to deploy servlets and Enterprise Java Beans. XML, with its many support tools, is driving the increasing use of declarative programming. Tools such as xfy continue push the envelope by coupling XML declarations with code generation tools that handle low level coding details. For those interested in learning more about xfy, Justsystems is encouraging beta downloads from their Web site at www.xfytec.com.

Tech View

At the technical sessions, the big Java-XML news was the 1.6 release of the JWSDP. Although the contents of the JWSDP vary from release to release, the key components of JWSDP continue to be:

  • JAXP, the Java API for XML Processing, which supports SAX and DOM parsing as well as the creation of Java objects that efficiently execute XSLT transformations.
  • JAXR. the Java API for XML Registries, allowing Java programmers to create and interact with UDDI, ebXML and SOA registries.
  • JAXM, the Java APIs for XML Messaging, used to send and receive document oriented XML messages using a pure Java API. JAXM implements the Simple Object Access Protocol (SOAP) 1.1 with Attachments messaging
  • JAX-RPC, which lets developers build Web applications and Web services, incorporating XML-based RPC functionality based on the SOAP 1.1 specification.
  • JAXB, the Java Architecture for XML Binding, which provides a convenient way to bind an XML schema to a representation in Java code.

The APIs and reference implementations found in the JWSDP often begin as specifications in the Java Community Process (JCP). As users gain experience using the APIs, feedback helps evolve the technologies. When a technology is deemed mature enough for prime time, it typically gets rolled out into a release of the JWSDP. If the functionality really turns out to be useful, it then finds its way into either the J2SE or J2EE. JAXP, for example, was delivered with JDK 1.4 and JAX-RPC has found its way into J2EE.

JWSDP 1.6

So what's new with JWSDP 1.6? In addition to the APIs listed above, the most recent JWSDP includes:

  • Fast Infoset 1.0
  • Service Registry with full UDDI 3.0 support
  • XML Web Services Security 2.0 early access
  • JAXP 1.3.1
  • XML Digital Signatures 1.0
  • Sun Java Streaming XML Parser 1.0

While each is separate topic unto itself, for the purpose of this article I'll focus on the announcements that struck me as particularly interesting and reflective of what's happening in the Java-XML world. We'll start by looking at Fast InfoSet and the issues driving it and then track some significant additions and revisions to JAXB and JAX-RPC, two of the workhorse APIs of the JWSDP.

XML Challenges

As XML finds its way into the enterprise, developers face two major challenges. The first challenge is size. Because XML is text, it's larger than a comparable binary format, therefore requiring more bandwidth to deliver. This impacts both enterprise and wireless applications that need to transmit XML across networks. The second challenge is processing overhead. Because XML is a mix of content and markup, parsers need to extract content from markup. This, of course, takes time and resources — resulting in a performance hit for using XML.

Strategies for overcoming the size problem include applying standard compression algorithms, such as those supported by programs like WinZip or XML-specific compression techniques such as XMill. But with straight compression, there's no free lunch. If you compress to reduce size, you then need to do additional processing to expand the XML back to its original form.

Enter Fast InfoSet

Fast InfoSet is a binary XML format that tries to give you the free lunch by reducing both size and processing time. Fast InfoSet works by leveraging another XML technology, the XML Information Set, a W3C recommendation that defines what any application can expect to find in a XML document. For example, the information set for any well-formed XML document contains information items for elments, attributes, entities, processing instructions – any piece of an XML document. Each information item has a set of associated named properties. In effect if you have the InfoSet data for an XML document you can use it to generate the original XML document, a DOM tree or the sequence of SAX events you would expect from a SAX parser.

Fast InfoSet approaches the size/performance challenge by creating a compact representation of all the properties defined in the InfoSet. Since Fast InfoSet documents already organize the information that a parser has to work to extract, Fast InfoSet documents are faster to serialize and parse, and often smaller in size, than the equivalent XML documents. Thus, the developers of FI have worked to define a binary format optimized to balance the needs of both document size and processing time. Fast InfoSet documents are of interest in a number of domains from bandwidth- and resource-constrained mobile devices to high-bandwidth high-throughput systems.

However, it's important to note that while Fast InfoSet documents are generally smaller in size and faster to parse and serialize than equivalent XML documents, FI does not guarantee a reduction in size. However, size and performance measures have shown FI effective for many documents especially those with repeated elements and attributes.

When comparing Fast InfoSet against compression technologies such as gzip, it's important to remember that Fast InfoSet optimizes for both size and performance, while gzip optimizes only for size and still requires the overhead of recreating the document before processing.

While the FI project is only in phase 1 right now, longer term the project goal is to develop high performance Fast InfoSet serializers and parsers that can be seamlessly woven into Java-based distributed communication platforms and Web services. To help benchmark parsing performance the folks developing Fast InfoSet provide a free tool called Japex. Like JUnit in the Java world, Japex makes it easy to write XML processing benchmarks. Input to Japex is an XML file describing a test suite and the output is a time stamped report available in XML and/or HTML. Within the Fast InfoSet project, Japex has been used to estimate parsing performance and size of Fast InfoSet documents vs. regular XML documents.

The Fast InfoSet project group is interested in fostering a community of users interested in developing and/or using Fast InfoSet. If you're interested in exploring the possibilities of FI be sure to visit their Web site.

JAXB 2.0 Update

Another interesting Java-XML development announced at JavaOne is the 2.0 release of JAXB. JAXB's contribution to the XML-Java world has been its ability to let programmers work with XML data from within a Java program. The JAXB download comes with a compiler that takes an XML Schema as input and builds a specialized factory class that delivers a JavaBean class with programmatic access to XML data. The Java class that gets generated includes get and set methods for all the elements and attributes of the XML instance document.

JAXB 2.0 adds new functionality in several important areas while maintaining backward compatibility with JAXB 1.0. First, JAXB 2.0 will support the entire W3C XML Schema spec rather than just a subset as did JAXB 1.0. To place JAXB 2.0 in perspective, I recall attending a birds-of-a-feather session on JAXB 1.0 just prior to its release. The Sun developers had done a tremendous amount of work mapping DTDs to Java classes when it became apparent that XML Schema was about to be finalized by the W3C. As Sun was hurrying to get JWSDP out the door, the JAXB development team had to scramble and switch gears from DTDs to XML Schema. Under a serious time crunch, JWSDP shipped without JAXB and soon after, JAXB shipped but with only partial support for the complex XML Schema specification. But now that will change and JAXB 2.0 will support the entire XML Schema specification.

Also new in JAXB 2.0 is the capability to do a reverse mapping from Java objects to XML Schema. This will be a great help to developers who begin with a Java application and then want to export functionality as a web service. Since XML Schemas are important for many web services tools, the ability to create an XML Schema from a Java object will go a long way to helping speed up Java based web service development.

JAX-RPC – what's in a name?

JAX-RPC was also in the news at JavaOne this year, with the announcement of version 2.0 and a name change. The new name will be JAX-WS2, reflecting a shift from Java API for XML-Based RPC to Java API for XML-based Web Services. The intent is to make it clear that the API is not just about making remote procedure calls but about delivering XML and Web Services.

Name changes are not taken lightly and while JAX-RPC has been a key component of the JWSDP since the beginning, the acronym RPC was often confusing to developers who were interested in sending and receiving XML but not explicitly interested in making remote procedure calls. In practice, JAX-RPC can do both. It can (a) trigger a remote procedure call on a remote server and (b) simply deliver an XML document within a SOAP envelope to a server. But the "RPC" in the name led to some confusion on the part of developers about when it was appropriate to use JAX-RPC. Some of the confusion can be traced to SOAP's early days; in 1998, Dave Winer and Don Box of DevelopMentor flew to Redmond to brainstorm with Microsoft developers about the possibilities of sending XML over HTTP. Winer had been working on using XML to implement remote procedure calls by packaging method names and parameters in the payload of an HTTP request. Microsoft was looking at simply sending XML documents in the same payload. Out of those discussions came what eventually turned out to be the SOAP 1.1 standard, which included support for both XML-RPC and the XML document transport model that Microsoft found more suitable for its .NET initiative. After SOAP was released, Winer continued to work on XML-RPC and Sun, looking for a name that reflected its view of XML over HTTP, chose JAX-RPC.

The name change from JAX-RPC to JAX-WS2 is intended to put the focus on XML messaging and to alert JAX-RPC 1.x users that migration to JAX-RPC 2.0 will not be completely seamless, since JAX-WS2 includes several fundamental changes in dependencies and bindings. For Sun developers, the clean break from the "JAX-RPC" past eliminates the need to maintain legacy API classes. The Java API for XML-based Web Services can now proceed unencumbered by prior design constraints.

Java has come a long way since the first JavaOne in 1995. At the time the focus was on writing code that was processor independent. Then along came XML with the opportunity to move data across networks unencumbered by platform or languages. While it took Sun a few years to realize the significance of independent data vs. independent code, the Java-XML connection is strong and in full swing. Stay tuned.

Links:

JWSDP Download: http://java.sun.com/webservices/downloads/webservicespack.html

XML Information set: http://www.w3.org/TR/2004/REC-xml-infoset-20040204

xfy download: http://www.xfytec.com/download/download.html

XMill: http://sourceforge.net/projects/xmill

Fast InfoSet Project: https://fi.dev.java.net/

  • + Share This
  • 🔖 Save To Your Account