Peekee wrote: >A type of variable is a property that the rules define as a varibale, >and specify the following:
You're mixing up three concepts here: variable, class of variable, and type of datum that a variable can store. Please make them all distinct. >b) The type of variable which can be one of: Date, Integer, Float, or >String. "Float" refers to an implementation technique for storing and manipulating numerical values. A floating-point type stores numbers as a fixed-length fraction in binary (or occasionally some other radix) along with an integral exponent. It seems unlikely that we'd actually want to limit a variable to the range of any particular floating-point type. More likely we'd want to allow any rational number (which is a restriction we have in a couple of places). Please define exactly what the range of each of these types is. You refer to "Null" later on, which is not naturally a value of any of these types. Actually, to handle Null you might want to introduce disjunction (union) types: define the type Void that has only one value Null, then a variable can be restricted to type Integer+Void which means that it can store either an integer value or Null. It seems to me that switches are a special case of your variable concept, where the restricting type is an enumeration. -zefram
