things, SGML was used in the documentation of the stealth bomber. A typical SGML project results in the creation of a complex HTML-like language. In fact, HTML is an application of SGML. Why didn’t the Web creators just use SGML? SGML is much too heavyweight for popular use over the Internet. Though the Web would be a richer place if everyone used SGML to create their Web sites, no browser vendors were willing to implement SGML.  SGML would  also  dramatically  increase  the  network  burden.  Perhaps  most important, SGML is complex to learn, whereas HTML is very simple. Without a simple markup  language  like  HTML,  the  Web  probably  would  have  never  reached  critical mass. So then, what’s wrong with HTML? While SGML is too complex, HTML is a bit too simple for the demands of contemporary Web applications. HTML is intrinsically tied to how documents should look to their users. Though HTML documents are highly structured, they are structured around presentation. Consider the H1 tag. The text con- tained inside of an H1 tag is usually a headline of some sort, but you don’t really know much more about it than that. About all you know is that the Web page author wants that text to stand out. Along the same lines, consider Web sites that tell you when the Web site was last modified. You can look at a table of contents Web page and it will tell you instantly when the content was last changed. However, there is no way to look at the HTML code and algorithmically determine when it was last updated. Even though HTML documents are highly structured, they aren’t semantically structured. There is no way to look at HTML and interpret it as much more than text data. What would work is the ability to define your own tags somehow; however, HTML doesn’t let you do that. The tag set is created by a standards body. Because all of the browser vendors are expected to implement the latest features of HTML, any additional functionality  needs  to  be  universally  applicable.  This  is  understandable,  of  course. HTML represents the presentation layer and will never be applicable to the structure of your data. What is needed, then, is a way to structure your data separate from HTML. This is the value of XML. Instead of confining yourself to HTML, you are able to cre- ate  your  own  language  for  your  application.  Oracle’s  canonical  schema  that  you’ll learn about in a few pages is an example of this. Oracle developed a way to represent SQL result sets in XML. Once established, XSQL developers can use it to present and manipulate the data. You also have the power to define your own markup language and can do it very quickly. You don’t have to write a parser—XML parsers are widely available. Your burden is simply to think about what you need, create an XML schema, and document it. Once created, other people inside and outside of your organization can use it for data interchange. NOTE   Defining your own tags is great, but an XML document isn’t instantly useful by itself. While an HTML document is immediately usable by millions of Web browsers worldwide, an XML document isn’t. You can create a <LastUpdated> tag, but what application will understand it? Thus, XML is primarily used, in conjunction with XSLT, to create  HTML and other standardized document types, or by a Web services client that knows the XML schema you are using. 14 Chapter 1 271209 Ch01.F  12/9/02  2:00 PM  Page 14