The Guru[ap]s Guide to SQL Server[tm] Stored Procedures, XML, and HTML

HTML's purpose is to format documents. It specifies display elementstitles, headings, fonts, captions, and so on. It's very presentation oriented. It's pretty good at laying out data. It's not good at describing that data or making it generally accessible.

Web site designers have worked around HTML's many shortcomings in some astonishingly novel ways. Still, HTML has serious flaws that make it ill suited for building complex, open information systems. Here are a few of them:

• HTML isn't extensible. Each browser supports a fixed set of tags, and you may not add your own.

• HTML is format-centric. Although it displays data reasonably well, HTML gives data no context. If the format of data a program is accessing via HTML changes, the program will likely break.

• Once generated, HTML is static and cannot be easily refreshed. DHTML and other technologies help alleviate this, but HTML, in its most basic essence, was never intended to serve up live data.

• HTML provides only a single view of data. Because it is display-centric, changing the view of the data is more difficult than it should be. Again, technologies like DHTML help to some extent, but the bottom line is that we need a markup language that knows about its data.

• HTML has little or no semantic structure. There's no facility for representing data by meaning rather than by layout. As I've said, HTML's forte is displaying data, and sometimes it's not even very good at that.

Although SGML doesn't have these faults, its vast flexibility makes it extremely complex, as I've said. DSSSL (Document Style Semantics and Specification Language), the language used to format SGML, is powerful and flexible, but this power comes at a price: It's extremely difficult to use. What we need is a language that's similar to HTML in terms of ease of use, but that features the flexibility of SGML.