InformationWeek

October 4, 1999

Developments:
HTML Gets Xtended

Extensible HTML brings structure to future Web-page formatting

By John Tibbetts and Barbara Bernstein

Tibbetts&Bernstein elcome, XHTML! We've had excellent results using this next-generation page markup language on a recent project, and now the World Wide Web Consortium (www.w3.org) is about to make it official. XHTML 1.0 is the recommended successor to HTML, the Hypertext Markup Language, which has been specifying (often inelegantly) the graphical layout of Web pages for years.

The new leading "X" signifies that HTML has joined a large family of interconnected, extensible, self-aware ("smart") document types. The family in question is the XML family, whose common ancestor is the Extensible Markup Language, which defines a general way to create and embed specialized commands and fancy attributes into any document. XML provides the raw material from which we can create specialized vocabularies of tags, which help organize a document by providing extra information about how various pieces of media should be handled.

Until now, most XML applications have focused on defining application-related tag sets that let Web sites index and categorize their own contents. The XHTML initiative, however, applies XML to page formatting. It rewrites in XML all those familiar HTML tags: Header1, Table, Background Color, etc. XHTML rigorously duplicates the current HTML 4.0 tag set, limitations and all, in order to be compatible with the hundreds of millions of HTML "legacy" Web pages out there. But it adds one crucial feature: extensibility.

Over the years, HTML has been extended haphazardly--stretched and grafted onto as necessary for this browser or that Web site. Moving it to XML brings a structure to this expansion. The XML substructure provides a framework for defining extensions that will be understood by other XML tools, applications, and projects. For example, if your Web page needs to work directly with a database, you can define an XML extension that adds database-access tags to your HTML vocabulary. If you had made these extensions under HTML, they would have been completely proprietary, another private HTML implementation that only your own application could understand. With XHTML, your extensions can be codified and distributed to others.

On our recent XHTML project, we used the fast, stable, and free IBM XML parser to prepare the pages and extensions that we had defined. This saved us the long and complicated effort involved in extending HTML. In addition, the parser's extensive linguistic analysis of the page language let us verify the correctness of the pages at build time. We didn't have to perform exhaustive manual testing to find every missing "table" tag.

We also found a real team-organization benefit to XHTML. We were able to structure our Web-application project as two distinct subprojects. There was a crisp boundary between the work done by the Web team (the page artists) and the object team (the code-pushers). Neither side needed to know anything about the other except how to communicate through the XHTML extensions. The artists could simply indicate something like "FieldData," and the developers would extend a servlet to understand and implement that tag. The increase in productivity was astounding.

E-business hyperinflation is forcing us to push our implementation architectures to their limits. Everybody has recognized that the venerable, static HTML page model is going to have to be extended in all directions, and quickly. If we make these extensions willy-nilly, without standards to guide us, we will soon be back in Babel, after spending the last decade fighting toward portability with open initiatives such as Linux, Java, and TCP/IP.

XHTML shows us how to make the necessary extensions in a way that keeps us in sync with one another, and positions us to move our documents painlessly to new platforms and devices as they come along.

John Tibbetts and Barbara Bernstein are partners in Kinexis, a San Francisco consulting firm. You can visit their Web site at www.kinexis.com.

Back to Columnists
Send Us Your Feedback
Top of the Page