Extension point schema
Extensions are the key mechanism that a plug-in uses to add new features to the platform. Extensions cannot be arbitrarily created. They are declared using a clear specification defined by an extension point.
Each extension must conform to the specification of the extension point that it is extending Each extension point defines attributes and expected values that must be declared by an extension. This information is maintained in the platform plug-in registry. Extension point providers query these values from the registry, so it's important to ensure that your plug-in has provided the expected information.
In the most rudimentary form, an extension point declaration is very simple. It defines the id and name of the extension point. Any other information expected by the extension point is specific to that extension point and is documented elsewhere. (See the Platform Extension Point Reference for the platform extension point definitions.)
Reference documentation is useful, but it does not enable any programmatic help for validating the specification of an extension. For this reason, PDE introduces an extension point schema that describes extension points in a format fit for automated processing.
Extension point schema is a valid XML schema as defined by W3C specification. However, the full XML schema specification is very complex and mostly unnecessary for this particular use. For this reason, PDE uses only a subset of XML schema features. Each extension point schema is a valid XML schema, but PDE does not use all the available features.
The benefits of extension point schemas
There are many benefits to describing your extension point using the PDE extension point XML schema:
- Extension point grammar allows elements, attributes, and types to be expressed formally. This information can be used by tools to validate extensions or offer assistance during creation of the extension.
- XML schema provides for documentation annotation that is similar to Javadoc in Java source. This mechanism ties short text for valid elements and attributes to the declaration of these elements and attributes. It is much easier to keep the documentation in sync because removal of an attribute will also remove documentation for the attribute. There is no need to update the reference document.
- Reference documentation can be generated. PDE provides a tool that tracks changes in extension point schemas and updates reference documentation on the fly.
- You can provide additional metadata about the extension point that can be used by tools that process the schema. PDE uses this mechanism to add additional information about elements and attributes. For example, if an attribute is marked as "Java," PDE can provide assistance while setting the value of this attribute by interacting with Java platform features.
Limitations of PDE XML Schema support
PDE uses a small subset of XML schema. Using the full XML schema features set would be an overkill in this particular case. The subset allows almost 1->1 mapping from DTDs to schemas, but without DTD limitations. The following are the main limitations of the PDE extension point schema:
- Only global element declarations are allowed.
- Only local attribute declarations are allowed. Global attributes cannot be declared.
- The following compositors are supported: all, sequence, choice and group.
- There is no global type support. Types must be declared and immediately used.
- Attributes can only have string and boolean types.
- If an attribute is of type string, only the enumeration restriction is supported.
If you write an XML schema using these restrictions, you will notice that the resulting file looks strikingly similar to an equivalent DTD that defines the same grammar. The advantage of schema is in annotations (both documentation and metadata). An additional advantage is that the XML schema is itself written in XML, which makes its processing and reading much easier.
The list above is for reference only. You are most likely to define an XML schema using the PDE schema editor that will take care of generating the correct file.