With the requirements listed in the previous section it seems sensible to split functionality into two sections, that of semantic computation and that of pragmatic. The class diagram shown in Figure 3.3 reflects this segregation.
Neither class require internal state to be maintained since all computation is done on a per input basis so neither class require any attributes. For the semantics class, the only methods required are an input parse function to receive the data and build the semantic output, and a private method for passing the semantic output onto the pragmatics class.
The pragmatics class requires a greater number of methods, a public function to receive the semantics and several action functions by which it can signal the actions the animation engine should undertake. Since the number and name of these functions will be determined by the number of specific actions the animation engine is capable of, the class diagram replaces them with a single entry called signal_action*().
