XML s Many Dependents

XML's Many Dependents

XML's great strength, its extensibility, has led to the creation of approximately 500 XML-based languages, standards, tools, and so on. The rapid and decentralized opportunity presented by XML has ignited a stampede of opportunistic groups (trade groups, academic groups, government regulators, and software vendors) to stake a claim in the eMarket. This ferment is a major reason that "outside" people tend to become confused and tend to believe that XML is unstable.

The 500 dialects can be grouped in three main categories:

• Tool vocabularies

• Industry-specific vocabularies

• Generic vocabularies

Tool Vocabularies

Most of the XML specifications that are not industry specific, one way or another, are ways of describing a tool or capability. The use of the tool is described in a specific XML dialect. For example, the extensible stylesheets language (XSL) tool is a way of describing how one XML document can be transformed into another XML document, into an HTML document, or into any other type of document (this is handy for displaying XML).

I call this a tool vocabulary, because somewhere behind the syntax of XSL is a tool that reads the XSL tags and does something to the XML. In that sense it is a tool. There are many tools, largely because each tool does few things. This leads to a modular style of assembly.

Some of the other tools of most interest include the following:

• XPath—XPath is a language for addressing part of an XML document. This seemingly simple ability is amazingly useful. By describing, in the form of a pattern, what tags you are interested in, and contextually where you are interested in them, you are able to withstand many changes to the XML document without requiring any changes to your interface.

• XQuery—There are many query languages built for making interrogation of XML documents or XML databases. This is just one of them.

• SVG—Scalable vector graphics is an XML representation of vectorized drawings, which allows them to be rendered or interrogated.

There are many more, and more are being invented all the time, but these are some of the more important ones. They are representative of the breadth and depth of support that is available to operate on XML documents, which in turn can be expressed in XML.

Industry-Specific Vocabularies

Virtually every industry has built a schema on which to base its future interoperation or internal systems. As we discussed in Chapter 2, each industry has a long history of having invented its own terms. These initiatives represent an attempt to codify and normalize them. To the extent that they are comprehensive and accepted, they will provide a great deal of value to the industry participants. To get broad agreement within an industry, most industries have formed consortia, or committees of sponsors who represent the interests of the industry in general.

In many industries these consortia are led by vertically oriented software companies. A vertical software company is one that specializes in a particular industry. In many ways these vertical software companies are the natural choice for this assignment. They make a living going from company to company within the industry. They have had to capture the semantics of the industry in the schemas of their databases. They have had to resolve some of the semantic mismatches in intercompany integration (the topic of Chapter 12).

However, there is a downside to allowing the software vendors to drive this process: It is in their conscious and unconscious best interest to define the industry vocabulary such that it closely resembles their current (legacy) database schemas. Most of these schemas are based on very old models. The schemas of most of the leading vendors in most verticals were defined 10 to 20 years ago.

Representative consortia-led XML dialects include the following:

• Health language 7 (HL7) is an XML version of an EDI standard that is at least 10 years old, and likely much older. It has been heavily represented by the software vendors.

• ACCORD is the consortia-led XML dialect for the insurance industry, primarily property and casualty.

• ChemXL is primarily a markup language to allow for B2B purchasing of chemicals.

• NewsML is a markup language to manage news throughout its life cycle, including production, interchange, and consumer use.

This list just scratches the surface. It seems that every time we turn around there is a new announcement. I'm waiting for the rebuilt carburetor markup language to know that it has finally reached saturation.

One final problem with this proliferation is that it is perpetuating the tower of babble. Each industry is potentially getting further entrenched in its own vocabulary to the detriment of communicating with other industries. As we mentioned in Chapter 2, no industry is an island; all industries have to communicate with others for their basic needs, and many have most of their customers in industries other than their own. To address this, we'll introduce generic vocabularies. Chapters 12 through 15 will then take up three different aspects of this problem.

Generic Vocabularies

Some initiatives cut across all industries. The most important include the following:

• Web Services—We won't go into this here because Chapter 13 is devoted to this new, XML-based way for programs to interact.

• The Semantic Web—Chapter 14 examines this initiative, which adds a layer of shared and discoverable meaning to the data stored on the Web. The Semantic Web includes RDF and the ontology Web language (OWL).

• BPEL4S—This is the latest of many standards to support work flow, both within an organization and across organizations.

• xCBL, cXML, and RozettaNet—These are XML dialects intended to support generic cross-company and cross-industry B2B eCommerce. We will discuss these in Chapter 12 in the context of systems integration.

These initiatives offer the most hope for total interconnection and scale. Each is aimed not at a specific industry and not to support a particular tool function, but is aimed at supporting broad-based interconnection.