/* C/C++ union example A union is a struct that maps different datatypes to the same memory address. As a result a change in one variable is reflected in the variables of other datatype. This implies when using union one must ensure they are convertible to each other else the results can be undefined. Unions were created to handle memory constrains like that often encountered in Kernel and OS development and embedded programming. Also remember code using union is generally not portable. Properties of union 1. All data member share the same memory location. 2. The size of union datatype is equal to size of its largest member. 3. When declaring a union only one member can be initialized. See the code, only long variable is initialized, it could have been int or char variable but not more than one at a time. Initialized variable need not be the first declared variable. 4. union is C language heritage but it is available in C++ also. Since union is a struct which itself is a class in C++ there are some restrictions on union when used in C++ 4a. They can have constructor but only one variable can be initialized either in the constructor body or member initializer list. 4b. Cannot have destructors. 4c. Can have member methods but cannot have virtual methods. 4d. Cannot be used as Base classes. 4e. In C++ union cannot have reference member variables. 4f. In C++, unions can have member access specifier likes private, public and protected with public being the default access specifier. */ #include using namespace std; union MultiData { long l=0; int i; char c; // member methods can be there in // C++ only. In C you can have // pointer to functions. But as the // memory location is shared, one // needs to be very careful of what // they are doing. int getInt(){return i;} }; void Display(MultiData const* ptr) { std::cout<<"ptr->l = "<< ptr->l << "\n" << std::dec << "ptr->i = " << ptr->i << "\n" << "ptr->c = " << ptr->c << "\n\n"; } int main() { MultiData multi; MultiData* ptr = &multi; std::cout << std::hex << std::showbase << "address of ptr " << ptr << "\n"; // In C++ reinterpret_cast or C style // cast (int*) is required to display // the address of char datatype. This // is because pointer to char get // special treatment in C/C++ as they // are treated as char stream or char // arrays. In C you could just use // "%p" or "%x" format specifier to // display the address of any variable // of any datatype std::cout << "address of ptr->l : " << &ptr->l << "\n"; std::cout << "address of ptr->i : " << &ptr->i << "\n"; std::cout << "address of ptr->c : " << reinterpret_cast(&ptr->c) << "\n\n"; // display initial values. // the value of ptr->c appears empty // because ascii value 0 is not a // displayable character. std::cout << "initial values\n"; Display(ptr); // as mentioned earlier, changing the // value of any variable is reflected // in other variables also. ptr->l = 80; std::cout << "after long variable changed\n"; Display(ptr); // change value of int variable std::cout << "after int variable changed\n"; ptr->i = 50; Display(ptr); // change value of char variable std::cout << "after char variable changed\n"; ptr->c = 'B'; Display(ptr); // Getting the value of int variable using member method std::cout << "Integer value using member method\n"; std::cout << "ptr->i = " << ptr->getInt(); return 0; }

C_C++_Union_example.cpp