Section 10.1 References

In the string chapter we mentioned that the a function that takes a string parameter probably should be written as:

void foo(const string& s);

Instead of:

void foo(string s);

So why is the first version preferable? To answer that question, we will start by reviewing how normal variables work. Each regular variable is a name for a different location in memory (a different box to store things in). Given this code:

int a = 5;
int b = a;

There are two storage locations, each with its own name. The second line of main make a new storage area with the name b and copies the value a has to it. When reading that code, you should picture a memory diagram that looks like this:

Changing the value in the storage named by b does not change the value stored in a. This is how things usually work in C++.

A reference changes this. A reference does not name its own storage, instead a reference always “points” to some other storage. We declare a variable to be a reference by adding & to the data type, like int& b;. The & change the type of `b —it no longer is an “int”, it is a “reference to an int”.

A memory diagram for int& b = a; should look like:

b is an alias for a - it is another way to name the exact same storage that a names. If we change b, we change that value. Run this Codelens sample to see its animation of the situation:

You can change either a or b and both variables “see” the new value.