2.9 — Naming collisions and an introduction to namespaces

Let’s say you are driving to a friend’s house for the first time, and the address given to you is 245 Front Street in Mill City. Upon reaching Mill City, you take out your map, only to discover that Mill City actually has two different Front Streets across town from each other! Which one would you go to? Unless there were some additional clue to help you decide (e.g. you remember his house is near the river) you’d have to call your friend and ask for more information. Because this would be confusing and inefficient (particularly for your mailman), in most countries, all street names and house addresses within a city are required to be unique.

Similarly, C++ requires that all identifiers be non-ambiguous. If two identical identifiers are introduced into the same program in a way that the compiler or linker can’t tell them apart, the compiler or linker will produce an error. This error is generally referred to as a naming collision (or naming conflict).

An example of a naming collision

a.cpp:

#include <iostream>

void myFcn(int x)
{
    std::cout << x;
}

main.cpp:

#include <iostream>

void myFcn(int x)
{
    std::cout << 2 * x;
}

int main()
{
    return 0;
}

When the compiler compiles this program, it will compile a.cpp and main.cpp independently, and each file will compile with no problems.

However, when the linker executes, it will link all the definitions in a.cpp and main.cpp together, and discover conflicting definitions for function myFcn. The linker will then abort with an error. Note that this error occurs even though myFcn is never called!

Most naming collisions occur in two cases:

1. Two (or more) definitions for a function (or global variable) are introduced into separate files that are compiled into the same program. This will result in a linker error, as shown above.
2. Two (or more) definitions for a function (or global variable) are introduced into the same file (often via an #include). This will result in a compiler error.

As programs get larger and use more identifiers, the odds of a naming collision being introduced increases significantly. The good news is that C++ provides plenty of mechanisms for avoiding naming collisions. Local scope, which keeps local variables defined inside functions from conflicting with each other, is one such mechanism. But local scope doesn’t work for function names. So how do we keep function names from conflicting with each other?

What is a namespace?

Back to our address analogy for a moment, having two Front Streets was only problematic because those streets existed within the same city. On the other hand, if you had to deliver mail to two addresses, one at 209 Front Street in Mill City, and another address at 417 Front Street in Jonesville, there would be no confusion about where to go. Put another way, cities provide groupings that allow us to disambiguate addresses that might otherwise conflict with each other. Namespaces act like the cities do in this analogy.

A namespace is a region that allows you to declare names inside of it for the purpose of disambiguation. The namespace provides a scope region (called namespace scope) to the names declared inside of it -- which simply means that any name declared inside the namespace won’t be mistaken for identical names in other scopes.

Within a namespace, all names must be unique, otherwise a naming collision will result.

Namespaces are often used to group related identifiers in a large project to help ensure they don’t inadvertently collide with other identifiers. For example, if you put all your math functions in a namespace called math, then your math functions won’t collide with identically named functions outside the math namespace.

We’ll talk about how to create your own namespaces in a future lesson.

The global namespace

In C++, any name that is not defined inside a class, function, or a namespace is considered to be part of an implicitly defined namespace called the global namespace (sometimes also called the global scope).

In the example at the top of the lesson, functions main() and both versions of myFcn() are defined inside the global namespace. The naming collision encountered in the example happens because both versions of myFcn() end up inside the global namespace, which violates the rule that all names in the namespace must be unique.

Only declarations and definition statements can appear in the global namespace. This means we can define variables in the global namespace, though this should generally be avoided (we cover global variables in lesson 6.4 -- Introduction to global variables). This also means that other types of statements (such as expression statements) cannot be placed in the global namespace (initializers for global variables being an exception):

#include <iostream> // handled by preprocessor

// All of the following statements are part of the global namespace
void foo();    // okay: function forward declaration in the global namespace
int x;         // compiles but strongly discouraged: uninitialized variable definition in the global namespace
int y { 5 };   // compiles but discouraged: variable definition with initializer in the global namespace
x = 5;         // compile error: executable statements not allowed in the global namespace

int main()     // okay: function definition in the global namespace
{
    return 0;
}

void goo();    // okay: another function forward declaration in the global namespace

The std namespace

When C++ was originally designed, all of the identifiers in the C++ standard library (including std::cin and std::cout) were available to be used without the std:: prefix (they were part of the global namespace). However, this meant that any identifier in the standard library could potentially conflict with any name you picked for your own identifiers (also defined in the global namespace). Code that was working might suddenly have a naming conflict when you #included a new file from the standard library. Or worse, programs that would compile under one version of C++ might not compile under a future version of C++, as new identifiers introduced into the standard library could have a naming conflict with already written code. So C++ moved all of the functionality in the standard library into a namespace named “std” (short for standard).

It turns out that std::cout‘s name isn’t really std::cout. It’s actually just cout, and std is the name of the namespace that identifier cout is part of. Because cout is defined in the std namespace, the name cout won’t conflict with any objects or functions named cout that we create in the global namespace.

Similarly, when accessing an identifier that is defined in a namespace (e.g. std::cout) , you need to tell the compiler that we’re looking for an identifier defined inside the namespace (std).

There are a few different ways to do this.

Explicit namespace qualifier std::

The most straightforward way to tell the compiler that we want to use cout from the std namespace is by explicitly using the std:: prefix. For example:

#include <iostream>

int main()
{
    std::cout << "Hello world!"; // when we say cout, we mean the cout defined in the std namespace
    return 0;
}

The :: symbol is an operator called the scope resolution operator. The identifier to the left of the :: symbol identifies the namespace that the name to the right of the :: symbol is contained within. If no identifier to the left of the :: symbol is provided, the global namespace is assumed.

So when we say std::cout, we’re saying “the cout that lives in namespace std“.

This is the safest way to use cout, because there’s no ambiguity about which cout we’re referencing (the one in the std namespace).

Using namespace std (and why to avoid it)

Another way to access identifiers inside a namespace is to use a using directive statement. Here’s our original “Hello world” program with a using directive:

#include <iostream>

using namespace std; // this is a using directive that allows us to access names in the std namespace with no namespace prefix

int main()
{
    cout << "Hello world!";
    return 0;
}

A using directive allows us to access the names in a namespace without using a namespace prefix. So in the above example, when the compiler goes to determine what identifier cout is, it will match with std::cout, which, because of the using directive, is accessible as just cout.

Many texts, tutorials, and even some compilers recommend or use a using-directive at the top of the program. However, used in this way, this is a bad practice, and highly discouraged.

Consider the following program:

#include <iostream> // imports the declaration of std::cout

using namespace std; // makes std::cout accessible as "cout"
 
int cout() // declares our own "cout" function in the global namespace
{
    return 5;
}
 
int main()
{
    cout << "Hello, world!"; // Compile error!  Which cout do we want here?  The one in the std namespace or the one we defined above?
 
    return 0;
}

The above program doesn’t compile, because the compiler now can’t tell whether we want the cout function that we defined, or the cout that is defined inside the std namespace.

When using a using-directive in this manner, any identifier we define may conflict with any identically named identifier in the std namespace. Even worse, while an identifier name may not conflict today, it may conflict with new identifiers added to the std namespace in future language revisions. This was the whole point of moving all of the identifiers in the standard library into the std namespace in the first place!