Prefixing Versus Postfixing

Clearly, whether you use the prefix or postfix form makes a difference if the value is used for some purpose, such as a function argument or assigning to a variable. But what if the value of an increment or decrement expression isn’t used? For example, are

x++;

and

++x;

different from one another? Or are

for (n = lim; n > 0; --n)
    ...;

and

for (n = lim; n > 0; n--)
    ...;

different from one another?

Logically, whether the prefix or postfix forms are used makes no difference in these two situations. The values of the expressions aren’t used, so the only effects are the side effects. Here the expressions using the operators are full expressions, so the side effects of incrementing x and decrementing n are guaranteed to be performed by the time the program moves on to the next step; the prefix form and postfix form lead to the same final result.

However, although the choice between prefix and postfix forms has no effect on the program’s behavior, it is possible for the choice to have a small effect on execution speed. For built-in types and modern compilers, this seems to be a non issue. But C++ lets you define these operators for classes. In that case, the user defines a prefix function that works by incrementing a value and then returning it. But the postfix version works by first stashing a copy of the value, incrementing the value, and then returning the stashed copy. Thus, for classes, the prefix version is a bit more efficient than the postfix version.

In short, for built-in types, it most likely makes no difference which form you use. For user-defined types having user-defined increment and decrement operators, the prefix form is more efficient.