Language Reference/Exception Handling

The basic part of the exception handling system is based on the built-in predicate errorExit/1 and the try-catch language construction.

• errorExit/1 raises an exception
• try-catch sets an exception handler for a certain computation.

When errorExit/1 is called the currently active exception handler is invoked. This exception handler is executed in its original context, i.e. in the context where it was set rather than in the context where the exception is raised.

The argument that errorExit/1 is invoked on is transferred to the exception handler. This argument must somehow provide the needed description of the exception.

Together with additional runtime routines, it is possible to build high-level exception mechanisms on top of this system.

It is however out of the scope of this document to describe runtime system access routines.

It is likewise out of the scope of this document to describe how the runtime system deals with exceptions occurring inside the runtime system.

The first argument of try-catch is the term to execute with new exception handler. The second argument must be a variable. This variable will be bound to the value errorExit/1 is invoked on, if it is invoked while this exception handler is active. The third argument is the exception handler, which will be invoked if errorExit/1 is called while this exception handler is active.

The exception handler can access the variable stated in the second argument thereby examining the exception that was raised.

Example

clauses
    p(X) :-
        try
            dangerous(X)
        catch Exception do
            handleDangerous(Exception)
        end try.

If an exception is raised while executing dangerous, then Exception will be bound to the exception value, and control will be transferred to the third argument of try-catch. In this case Exception is passed to handleDangerous.