Red Hat Training

A Red Hat training course is available for Red Hat JBoss Data Virtualization

Chapter 7. User Defined Functions

7.1. User Defined Functions

You can extend the Red Hat JBoss Data Virtualization function library by creating User Defined Functions (UDFs), as well as User Defined Aggregate Functions (UDAFs).
The following are used to define a UDF:
  • Function Name - When you create the function name, keep these requirements in mind:
    • You cannot overload existing Red Hat JBoss Data Virtualization functions.
    • The function name must be unique among user-defined functions in its model for the number of arguments. You can use the same function name for different numbers of types of arguments. Hence, you can overload your user-defined functions.
    • The function name cannot contain the '.' character.
    • The function name cannot exceed 255 characters.
  • Input Parameters - defines a type specific signature list. All arguments are considered required.
  • Return Type - the expected type of the returned scalar value.
  • Pushdown - can be one of REQUIRED, NEVER, ALLOWED. Indicates the expected pushdown behavior. If NEVER or ALLOWED are specified then a Java implementation of the function should be supplied. If REQUIRED is used, then user must extend the Translator for the source and add this function to its pushdown function library.
  • invocationClass/invocationMethod - optional properties indicating the static method to invoke when the UDF is not pushed down.
  • Deterministic - if the method will always return the same result for the same input parameters. Defaults to false. It is important to mark the function as deterministic if it returns the same value for the same inputs as this will lead to better performance. See also the Relational extension boolean metadata property "deterministic" and the DDL OPTION property "determinism".

Note

If using the pushdown UDF in Teiid Designer, the user must create a source function on the source model, so that the parsing will work correctly. Pushdown scalar functions differ from normal user-defined functions in that no code is provided for evaluation in the engine. An exception will be raised if a pushdown required function cannot be evaluated by the appropriate source.