25.5 — Early binding and late binding

In this lesson and the next, we are going to take a closer look at how virtual functions are implemented. While this information is not strictly necessary to effectively use virtual functions, it is interesting. Nevertheless, you can consider both sections optional reading.

When a C++ program is executed, it executes sequentially, beginning at the top of main(). When a function call is encountered, the point of execution jumps to the beginning of the function being called. How does the CPU know to do this?

When a program is compiled, the compiler converts each statement in your C++ program into one or more lines of machine language. Each line of machine language is given its own unique sequential address. This is no different for functions -- when a function is encountered, it is converted into machine language and given the next available address. Thus, each function ends up with a unique address.

Binding and dispatching

Our programs contain many names (identifiers, keywords, etc…). Each name has a set of associated properties: for example, if the name represents a variable, that variable has a type, a value, a memory address, etc…

For example, when we say int x, we’re telling the compiler to associate the name x with the type int. Later if we say x = 5, the compiler can use this association to type check the assignment to ensure it is valid.

In general programming, binding is the process of associating names with such properties. Function binding (or method binding) is the process that determines what function definition is associated with a function call. The process of actually invoking a bound function is called dispatching.

In C++, the term binding is used more casually (and dispatching is usually considered part of binding). We’ll explore the C++ use of the terms below.

Early binding

Most of the function calls the compiler encounters will be direct function calls. A direct function call is a statement that directly calls a function. For example:

#include <iostream>

struct Foo
{
    void printValue(int value)
    {
        std::cout << value;
    }
};

void printValue(int value)
{
    std::cout << value;
}

int main()
{
    printValue(5);   // direct function call to printValue(int)

    Foo f{};
    f.printValue(5); // direct function call to Foo::printValue(int)
    return 0;
}

In C++, when a direct call is made to a non-member function or a non-virtual member function, the compiler can determine which function definition should be matched to the call. This is sometimes called early binding (or static binding), as it can be performed at compile-time. The compiler (or linker) can then generate machine language instructions that tells the CPU to jump directly to the address of the function.

        mov     edi, 5           ; copy argument 5 into edi register in preparation for function call
        call    printValue(int)  ; directly call printValue(int)

Calls to overloaded functions and function templates can also be resolved at compile-time:

#include <iostream>

template <typename T>
void printValue(T value)
{
    std::cout << value << '\n';
}

void printValue(double value)
{
    std::cout << value << '\n';
}

void printValue(int value)
{
    std::cout << value << '\n';
}

int main()
{
    printValue(5);   // direct function call to printValue(int)
    printValue<>(5); // direct function call to printValue<int>(int)

    return 0;
}

Let’s take a look at a simple calculator program that uses early binding:

#include <iostream>

int add(int x, int y)
{
    return x + y;
}

int subtract(int x, int y)
{
    return x - y;
}

int multiply(int x, int y)
{
    return x * y;
}

int main()
{
    int x{};
    std::cout << "Enter a number: ";
    std::cin >> x;

    int y{};
    std::cout << "Enter another number: ";
    std::cin >> y;

    int op{};
    std::cout << "Enter an operation (0=add, 1=subtract, 2=multiply): ";
    std::cin >> op;

    int result {};
    switch (op)
    {
        // call the target function directly using early binding
        case 0: result = add(x, y); break;
        case 1: result = subtract(x, y); break;
        case 2: result = multiply(x, y); break;
        default:
            std::cout << "Invalid operator\n";
            return 1;
    }

    std::cout << "The answer is: " << result << '\n';

    return 0;
}

Because add(), subtract(), and multiply() are all direct function calls to non-member functions, the compiler will match these function calls to their respective function definitions at compile-time.

Note that because of the switch statement, which function is actually called is not determined until runtime. However, that is a path of execution issue, not a binding issue.

Late binding

In some cases, a function call can’t be resolved until runtime. In C++, this is sometimes known as late binding (or in the case of virtual function resolution, dynamic dispatch).

In C++, one way to get late binding is to use function pointers. To review function pointers briefly, a function pointer is a type of pointer that points to a function instead of a variable. The function that a function pointer points to can be called by using the function call operator () on the pointer.

For example, the following code calls the printValue() function through a function pointer:

#include <iostream>

void printValue(int value)
{
    std::cout << value << '\n';
}

int main()
{
    auto fcn { printValue }; // create a function pointer and make it point to function printValue
    fcn(5);                  // invoke printValue indirectly through the function pointer

    return 0;
}

Calling a function via a function pointer is also known as an indirect function call. At the point where fcn(5) is actually called, the compiler does not know at compile-time what function is being called. Instead, at runtime, an indirect function call is made to whatever function exists at the address held by the function pointer.

        lea     rax, [rip + printValue(int)] ; determine address of printValue and place into rax register
        mov     qword ptr [rbp - 8], rax     ; move value in rax register into memory associated with variable fcn

        mov     edi, 5                       ; copy argument 5 into edi register in preparation for function call
        call    qword ptr [rbp - 8]          ; invoke the function at the address held by variable fcn

The following calculator program is functionally identical to the calculator example above, except it uses a function pointer instead of a direct function call:

#include <iostream>

int add(int x, int y)
{
    return x + y;
}

int subtract(int x, int y)
{
    return x - y;
}

int multiply(int x, int y)
{
    return x * y;
}

int main()
{
    int x{};
    std::cout << "Enter a number: ";
    std::cin >> x;

    int y{};
    std::cout << "Enter another number: ";
    std::cin >> y;

    int op{};
    std::cout << "Enter an operation (0=add, 1=subtract, 2=multiply): ";
    std::cin >> op;

    using FcnPtr = int (*)(int, int); // alias ugly function pointer type
    FcnPtr fcn { nullptr }; // create a function pointer object, set to nullptr initially

    // Set fcn to point to the function the user chose
    switch (op)
    {
        case 0: fcn = add; break;
        case 1: fcn = subtract; break;
        case 2: fcn = multiply; break;
        default:
            std::cout << "Invalid operator\n";
            return 1;
    }

    // Call the function that fcn is pointing to with x and y as parameters
    std::cout << "The answer is: " << fcn(x, y) << '\n';

    return 0;
}

In this example, instead of calling the add(), subtract(), or multiply() function directly, we’ve instead set fcn to point at the function we wish to call. Then we call the function through the pointer.

The compiler is unable to use early binding to resolve the function call fcn(x, y) because it can not tell which function fcn will be pointing to at compile time!

Late binding is slightly less efficient since it involves an extra level of indirection. With early binding, the CPU can jump directly to the function’s address. With late binding, the program has to read the address held in the pointer and then jump to that address. This involves one extra step, making it slightly slower. However, the advantage of late binding is that it is more flexible than early binding, because decisions about what function to call do not need to be made until runtime.

In the next lesson, we’ll take a look at how late binding is used to implement virtual functions.