Problem
At runtime, you need to interrogate dynamically the type of particular class.
Solution
Use runtime type identification (commonly referred to as RTTI) to query the address of the object for the type of object it points to. Example 8-6 shows how.
Example 8-6. Using runtime type identification
#include #include using namespace std; class Base {}; class Derived : public Base {}; int main( ) { Base b, bb; Derived d; // Use typeid to test type equality if (typeid(b) == typeid(d)) { // No cout << "b and d are of the same type. "; } if (typeid(b) == typeid(bb)) { // Yes cout << "b and bb are of the same type. "; } if (typeid(d) == typeid(Derived)) { // Yes cout << "d is of type Derived. "; } }
Discussion
Example 8-6 shows you how to use the operator typeid to determine and compare the type of an object. typeid takes an expression or a type and returns a reference to an object of type_info or a subclass of it (which is implementation defined). You can use what is returned to test for equality or retrieve a string representation of the type's name. For example, you can compare the types of two objects like this:
if (typeid(b) == typeid(d)) {
This will return true if the type_info objects returned by both of these are equal. This is because typeid returns a reference to a static object, so if you call it on two objects that are the same type, you will get two references to the same thing, which is why the equality test returns true.
You can also use typeid with the type itself, as in:
if (typeid(d) == typeid(Derived)) {
This allows you to explicitly test for a particular type.
Probably the most common use of typeid is for debugging. To write out the name of the type, use type_info::name, like this:
std::cout << typeid(d).name( ) << std::endl;
When you are passing objects around of varying types, this can be a useful debugging aid. The null-terminated string returned by name is implementation defined, but you can expect (but not depend on) the name of the type most of the time. This works for native types, too.
Do not abuse this technique by basing program logic on type information unless you absolutely have to. In general, it is considered bad design to have logic that does something along the lines of:
If obj has a type of X, do something else, if obj has a type of Y, do something else.
This approach is a bad design because the client code now contains superfluous dependencies on the type of the object being used. It also results in a lot of messy if/then code that is duplicated everywhere you want particular behavior for an object of type X or Y. Object-oriented programming and polymorphic behavior exist in large part so you don't have to write this kind of logic. If you want type-specific, dynamic behavior for some family of related classes, then they should all subclass the same base class and use virtual functions to dynamically invoke potentially different behavior based on the type.
RTTI adds overhead, so compilers don't usually enable it by default. Chances are your compiler has a command-line parameter to turn on RTTI. Also, this isn't the only way you can query type information, see Recipe 8.7 for another technique.
See Also
Recipe 8.7
Building C++ Applications
Code Organization
Numbers
Strings and Text
Dates and Times
Managing Data with Containers
Algorithms
Classes
Exceptions and Safety
Streams and Files
Science and Mathematics
Multithreading
Internationalization
XML
Miscellaneous
Index