Transforming XML
The Extensible Markup Language took center stage Wednesday at the ninth International World Wide Web conference in Amsterdam. Specifically, there was much attention focused on XSLT (Extensible Style Language Transformations), the key to transforming XML documents into other types of documents. As XML use becomes more widespread, so will XSLT. Several major XSLT engines are available, including James Clark's XT, the Apache Project's Xalan, and Microsoft's MSXML3. These XSLT engines have one thing in common: They're all interpreters that take an XML document and an XSL style sheet as input and transform the document into the format specified by the style sheet, which might be Adobe Acrobat PDF format or HTML.

In his talk "An XSLT Compiler for the Java Virtual Machine: Enabling Ubiquitous XML Transformations," Sun Microsystems staff engineer Jacek Ambroziak unveiled Sun's XSLT Compiler (XSLTC), which should be available "soon," probably as part of a free bundle of XML tools from Sun. Unlike existing XSLT engines, XSLTC lets users compile style-sheet transformations into Java byte code, which Ambroziak calls "translets." Performing transformations using translets results in transformations that are faster and require less memory than existing approaches.

While translets result in significant speed improvements for server-side transformations (Ambroziak found it to be three to four times faster than Clark's XT engine, arguably the fastest engine available), translets also offer the possibility of client-side transformation of XML into other formats. That's because translets are typically small, perhaps 5 to 100 Kbytes in size, depending on the complexity of the transformation, and they can run on any Java 1.x or 2.x Java Virtual Machine. Are you a budding XSLT programmer? Then Ambroziak recommends as a resource Michael Kay's new "XSLT Programmer's Reference" from Wrox Press (http://www.wrox.com/Consumer/Store/Details.asp?ISBN=1861003129).

Still More XML
Michael Sperberg-McQueen, World Wide Web Consortium Schema Working Group co-chair, discussed the alpha version of the XML Schema validator, which is available online (http://www.w3.org/XML/Schema.html) and can be used to validate schemas that developers may devise. Schemas are a powerful set of standards (not all in place yet) for describing the structure of XML documents. They will likely supplant the current mechanism of using document type definitions. Sperberg-McQueen recommends that any developers working on new XML projects use schemas rather than DTDs.

That Old Killer App Again
Ever since Bellcore invented ADSL (asymmetric digital subscriber line) in 1989 to bring "video on demand" into America's living rooms, industry pundits have been forecasting the imminent arrival of on-demand video services for home users. The current buzzword for this genesis is "convergence" and thus the title of Paul De Bot's keynote at WWW9, "The Convergence of Internet and Television Technologies: How Will it Change the Living Room Experience?" De Bot, VP of business development for Philips Digital Networks, says the solution revolves around the MPEG4 standard, which permits high-quality compressed audio and video at varying rates. One significant feature of MPEG4 is the object-level addressability of the video frame, which he demonstrated with a video puzzle, showing a full-motion video broken up into puzzle pieces the user can drag and drop with a mouse while the video is running.

While I'm skeptical that we'll see much interesting video streaming into America's living rooms anytime soon, De Bot was bullish. But let's face it, that's his job. Not surprisingly, he wouldn't predict when this convergence might occur.

On the portable audio front, De Bot concedes that the MPEG3 standard requires too much bandwidth (128 Kbps for stereo, CD-quality audio) for the current crop of wireless handheld devices, especially cell phones. New audio algorithms are needed to bring stereo, CD-quality audio streams down below the 56-Kbps range so audio can be reliably streamed over next generation, or 2.5g, cell networks. (Current cell networks are considered second generation, or 2g, networks. The emerging packet cell networks are considered 2.5g, or 2-1/2 generation, networks. And 3g, or third-generation networks, with much higher bandwidth are still in the planning stages.)

Tomorrow: Moderating mobility