When tt(parse) encounters a em(syntax error), it calls the error reporting member function (tt(error)) to print an error message. A very basic in-line implementation is provided by b() in the parser class internal header file (see chapter ref(INTERFACE)), which can easily be redefined by the programmer. The tt(error) member's default implementation is OK for tt(rpn). Once tt(error) returns, b()'s parser may recover from the error and continue parsing if the grammar contains a suitable error rule (see chapter ref(RECOVERY)). Otherwise, the parsing function tt(parse) returns a non-zero value. No error production rules were used for tt(rpn), so the calculator terminates at the first syntax error.. Not very nice in a real-life calculator, but it is acceptable for this first example.