Network Computingwww.networkcomputing.com

In addition to those discussed earlier, a number of recommendations are at varying stages of completion, while some, such as the XML packaging recommendation, have been aborted because of a lack of interest.

One recommendation in discussion, the XML query recommendation (www.w3.org/XML/Query.html), seeks to develop a query language for XML such that programmers can perform searches against XML repositories and retrieve the results of these searches. An analogy would be using a select statement in SQL. However, SQL statements are well-defined because the syntax is exactly described and the SQL statements are applied against a database.

In XML, queries can be applied against a multitude of repositories, including human-legible documents; data documents; configuration files; streams of XML data; DOM (Document Object Model) structures; native XML repositories, such as Web servers; or other XML repositories.

Furthermore, the recommendation does not specify a syntax for the XML queries; it merely offers a description of how the language should look. We can't tell yet whether the XML query recommendation will be useful, but we believe that it will be pre-empted as database vendors work to XML-enable their application servers. For example, Microsoft's Internet Information Server (IIS) 5.0 can be configured with special virtual directories that hold XML documents containing SQL code. The SQL code is executed against Microsoft's SQL Server 2000, which in turn can return the results in XML format. An XSL style sheet can then be applied, resulting in a well-formatted output right in your browser.

Not to be outdone, Oracle is shipping its Oracle9 application server with a special version of the popular Apache Web server. This specially configured Web server lets the developer build XML documents that can execute SQL code against the Oracle database server, much as IIS 5.0 does with SQL Server 2000. The output is generated in XML and sent back to your browser. Other vendors and some public-domain databases are following suit, adding varying levels of XML support.

We find these new features to be much more valuable than the XML query recommendation. On the other hand, having a unified standard followed by all vendors would make it easier for developers to create portable software that works with multiple application and database servers (for more on vendor XML support, see "Dishing Up Dynamic Content").

Defining the Language

We all know that XML is a metalanguage used to define other languages by declaring and specifying new markup tags. What is the mechanism employed toward that end?

Initially, XML 1.0 supplied a mechanism for declaring constraints on the use of the markup tags. The mechanism was called DTD (Document Type Definition) and was based on the long-standing SGML (Standard Generalized Markup Language) standard. XML developers soon realized the shortcomings of DTDs in automated processing of XML documents. In addition, the DTD spec did not satisfy all the needs of developers for constraints on markup. You often need constraints that are tighter or looser than those allowed by the DTD.

Of course, those factors alone do not require a whole new method of constraining markup; enhancing the DTD would have sufficed. However, the majority of XML developers wanted to use XML syntax to define the constraints on the markup. This would let them leverage the many XML tools on the market. The DTD did not satisfy that requirement. Thus, a new working group was formed to create a replacement, known as the XML Schema (see www.w3.org/TR/NOTE-xml-schema-req).

As with most standards, XML Schema has several competing efforts to define what the standard should look like. Over the past two years, the working group has considered the following submissions by vendors and interested developers: XML-Data, DCD (Document Control Description), SOX (Schema for Object-Oriented XML) and DDML (Document Definition Markup Language).

As of October, the output of the XML Schema workgroup attained candidate-recommendation status. The resulting recommendation addresses three major areas: structure, data typing and conformance. Structure defines the relationships between different markups in an XML document; data typing introduces to XML the concept of data types, which will be valuable when extracting data from databases or when using XML in EDI (Electronic Data Interchange); and conformance, which, as the name implies, deals with the mechanisms of making sure that a document is well formed and conforms to its associated schema.

A big part of what makes XML documents so versatile in a Web environment is the DOM, a platform-independent and language-neutral interface that exposes various parts of XML as objects that can be accessed by programmers and scripts. The DOM specification has various parts, including the DOM Core specification, the DOM Views specification, the DOM Events specification, the DOM Style specification, and the DOM Traversal and Range specification. These specs stand at what is called Level 2.

Ironically, one of the most important objectives for creating the DOM specification is to provide a standard programming interface, yet the DOM implementations differ widely by browser. Find the list of all recent W3C recommendations, including DOM documents, at www.w3.org/TR/.