Most programs that use b() parse only one language and therefore only need one b() parser. But what if you want to parse more than one language with the same program? Since b() constructs a em(class) rather than a em(parsing function), this problem can easily be solved: simply define your second (third, fourth, ...) parser class, each having its own unique class-name, using the tt(%class-name) directive, and construct parser objects of each of the defined classes. In multi-threaded programs each thread can create its own parser object, and then call that object's tt(parse) function: b() does not use static data that are modified by the generated parser. In multi-threaded programs, when the parser uses polymorphic semantic values, the option tt(--thread-safe) or the directive tt(%thread-safe) should be specified to ensure that the code handling the polymorphic semantic data types accesses the error-counter data member of their own parser object.