num_type

num_type 
      ::= 
      exact_num_type 
      | approximate_num_type 

This rule is called by

-> data_type