Comparison of Different Schema Languages

Different schema languages have different strengths and weaknesses. It is sometimes a good idea to create schemas in different languages for the same xml domain. If and when this is a good idea depends on the details of the xml domain and the tools used when working with the associated instance documents.

Document Type Definition (DTD)

The original schema language for xml, it has the following advantages:

• The basic structure is easily understood and authored.
• There are advanced features that make DTD's surprisingly powerful and expressive.
• The DTD language is built-in to XML itself and so it is a necessary technology. One example of this is the need to use DTD when defining ENTITY's such as character escape sequences.

DTD has the following disadvantages:

• The syntax isn't xml so using DTD requires an independent parser.
• There is only very limited support for namespaces.
• There is no support for datatypes beyond strings and enumerations.
• The structure is document-centric, so there is no inherent support for protocols.

XML Schema Definition Language (XSDL)

Created and maintained by the W3C organization, XSDL is perhaps the most widely used schema language today. It has the following advantages:

• The syntax is XML.
• There is extremely good support for datatypes.
• It is namespace aware.
• It treats elements and attributes as similarly as is possible. This permits rich structural constraints on groups of attributes.
• There are a huge number of tools available for working with XSDL.
• It provides a pseudo-OOP approach to defining structure, in that structures can be defined and reused via a referencing syntax.

XSDL has the following disadvantages:

• It is structurally complex, making the raw XML difficult to write and difficult to read.
• The namespace support is limited. Normally an XSDL schema will define structures within a single namespace. When multiple namespaces are involved, the schema author needs to use document linking artifacts.
• The structure is document-centric, so there is no inherent support for protocols.
• The agnostic treatment of elements vs attributes is also a weakness. It can seduce the XML designer into over-use of attributes. There are strong arguments that can be made that document data should be held entirely in element values, with attributes limited to meta-data.

Skeleton Schema Language

The skeleton schema has the following advantages:

• The syntax is XML.
• Schemas are easily written and understood. This is the principal design goal for the skeleton schema language. This is especially true when an XML design is based on an element-centric approach. It is fortunate that the element-centric approach is always the correct approach.
• Very good support for namespaces. Since a schema looks like the documents that it describes, the usage of namespaces in the schema echos the namespace usage in the instance documents.
• There is support for protocols. This is the second most important advantage of the skeleton schema language.
• Can support all datatypes that are also supportable in the XSDL because the same syntax is re-used. Beyond this, there are a set of inherent datatypes that provide very useful extensions to the core XSDL types.
• The support for attributes is limited to the support that XML itself provides. This helps to constrain the XML designer from over-usage of attributes.

The skeleton schema has the following disadvantages:

• There is currently NO tool support for skeleton schemas. This will change over time, but it is extremely unlikely that there will ever be the rich support currently available for XSDL schemas.
• The limited support for attributes means that the syntax for defining attributes is awkward. This is a weakness in that there are a lot of XML domains that have been designed with the attribute-centric approach. It would be nice to communicate these designs to human readers using skeleton schemas, but the result appears ugly and can give an initially poor impression of the skeleton schema approach.