1.4 — Variable assignment and initialization

In the previous lesson (1.3 -- Introduction to objects and variables), we covered how to define a variable that we can use to store values. In this lesson, we’ll explore how to actually put values into variables and use those values.

As a reminder, here’s a short snippet that first allocates a single integer variable named x, then allocates two more integer variables named y and z:

int x;    // define an integer variable named x
int y, z; // define two integer variables, named y and z

Variable assignment

After a variable has been defined, you can give it a value (in a separate statement) using the = operator. This process is called assignment, and the = operator is called the assignment operator.

int width; // define an integer variable named width
width = 5; // assignment of value 5 into variable width

// variable width now has value 5

By default, assignment copies the value on the right-hand side of the = operator to the variable on the left-hand side of the operator. This is called copy-assignment.

Here’s an example where we use assignment twice:

#include <iostream>

int main()
{
	int width;
	width = 5; // copy assignment of value 5 into variable width

	std::cout << width; // prints 5

	width = 7; // change value stored in variable width to 7

	std::cout << width; // prints 7

	return 0;
}

This prints:

57

When we assign value 7 to variable width, the value 5 that was there previously is overwritten. Normal variables can only hold one value at a time.

Variable initialization

One downside of assignment is that assigning a value to a just-defined object requires two statements: one to define the variable, and another to assign the value.

These two steps can be combined. When an object is defined, you can optionally give it an initial value. The process of specifying an initial value for an object is called initialization, and the syntax used to initialize an object is called an initializer. Informally, the initial value is often called an “initializer” as well.

int width { 5 }; // define variable width and initialize with initial value 5

// variable width now has value 5

In the above initialization of variable width, { 5 } is the initializer, and 5 is the initial value.

Different forms of initialization

Unlike assignment (which is comparably simple), initialization in C++ is surprisingly complex, so we’ll present a simplified view here to get started.

There are 5 commonly used ways to initialize variables in C++:

int a;         // default-initialization (no initializer)

// Traditional initialization forms:
int b = 5;     // copy-initialization (initial value after equals sign)
int c ( 6 );   // direct-initialization (initial value in parenthesis)

// Modern initialization forms (preferred):
int d { 7 };   // direct-list initialization (initial value in braces)
int f {};      // value-initialization (empty braces)

You may see the above forms written with different spacing (e.g. int d{7};). Whether you use extra spaces for readability or not is a matter of personal preference.

Default-initialization

When no initializer is provided (such as for variable a above), this is called default-initialization. In most cases, default-initialization performs no initialization, and leaves the variable with an indeterminate value.

We’ll discuss this case further in lesson (1.6 -- Uninitialized variables and undefined behavior).

Copy-initialization

When an initial value is provided after an equals sign, this is called copy-initialization. This form of initialization was inherited from the C language.

int width = 5; // copy-initialization of value 5 into variable width

Much like copy-assignment, this copies the value on the right-hand side of the equals into the variable being created on the left-hand side. In the above snippet, variable width will be initialized with value 5.

Copy-initialization had fallen out of favor in modern C++ due to being less efficient than other forms of initialization for some complex types. However, C++17 remedied the bulk of these issues, and copy-initialization is now finding new advocates. You will also find it used in older code (especially code ported from C), or by developers who simply think it looks more natural and is easier to read.

Direct-initialization

When an initial value is provided inside parenthesis, this is called direct-initialization.

int width( 5 ); // direct initialization of value 5 into variable width

Direct-initialization was initially introduced to allow for more efficient initialization of complex objects (those with class types, which we’ll cover in a future chapter). Just like copy-initialization, direct-initialization had fallen out of favor in modern C++, largely due to being superseded by direct-list-initialization. However, direct-list-initialization has a few quirks of its own, and so direct-initialization is once again finding use in certain cases.

int x();  // forward declaration of function x
int x(0); // definition of variable x with initializer 0
T(5);     // function call if T is a function, direct-initialization of temporary object if T is a type

Direct-list-initialization and Copy-list-initialization

The modern way to initialize objects in C++ is to use a form of initialization that makes use of curly braces. This is called list-initialization (or uniform initialization or brace initialization).

List-initialization comes in two forms:

int width { 5 };    // direct-list-initialization of initial value 5 into variable width (preferred)
int height = { 6 }; // copy-list-initialization of initial value 6 into variable height (rarely used)

Prior to the introduction of list-initialization, some types of initialization required using copy-initialization, and other types of initialization required using direct-initialization. Copy-initialization can be hard to differentiate from copy-assignment (because they both use an =). And direct-initialization can be hard to differentiate from function-related operations (because they both use parenthesis).

List-initialization was introduced to provide a consistent initialization syntax (which is why it is sometimes called “uniform initialization”) that makes it clear we’re doing initialization.

Additionally, list initialization provides a way to initialize objects with a list of values rather than a single value (which is why it is called “list initialization”). We show an example of this in lesson 16.2 -- Introduction to std::vector and list constructors.

List initialization disallows narrowing conversions

The primary benefit of list-initialization is that “narrowing conversions” are disallowed. This means that if you try to list-initialize a variable using a value that the variable can not safely hold, the compiler is required to produce a diagnostic (usually an error). For example:

int main()
{
    // An integer can only hold non-fractional values
    int w1 { 4.5 }; // compile error: list init does not allow narrowing conversion of 4.5 to 4

    int w2 = 4.5;   // compiles: copy-init initializes width with 4
    int w3(4.5);    // compiles: direct-init initializes width with 4

    return 0;
}

On line 4 of the above program, we’re using a value (4.5) with a fractional component (.5) to list-initialize an integer variable (which can only hold non-fractional values). Because this is a narrowing conversion, the compiler is required to generate a diagnostic in such cases.

Copy-initialization (line 6) and direct-initialization (line 7) both silently drop the .5 and initialize the variable with the value 4 (which probably isn’t what we want). Your compiler may warn you about this (since losing data is rarely desired), but it also may not.

Note that this restriction on narrowing conversions only applies to the list-initialization, not to any subsequent assignments to the variable:

int main()
{
    int w1 { 4.5 }; // compile error: list-init does not allow narrowing conversion of 4.5 to 4

    w1 = 4.5;       // okay: copy-assignment allows narrowing conversion of 4.5 to 4

    return 0;
}

List-initialization is the preferred form of initialization in modern C++

Value-initialization and zero-initialization

When a variable is initialized using empty braces, value initialization takes place. In most cases, value initialization will initialize the variable to zero (or empty, if that’s more appropriate for a given type). In such cases where zeroing occurs, this is called zero-initialization.

int width {}; // value-initialization / zero-initialization to value 0

int x { 0 };    // direct-list-initialization with initial value 0
std::cout << x; // we're using that 0 value here

int x {};      // value initialization
std::cin >> x; // we're immediately replacing that value so an explicit 0 would be meaningless

Initialize your variables

Initialize your variables upon creation. You may eventually find cases where you want to ignore this advice for a specific reason (e.g. a performance critical section of code that uses a lot of variables), and that’s okay, as long as the choice is made deliberately.

We explore what happens if you try to use a variable that doesn’t have a well-defined value in lesson 1.6 -- Uninitialized variables and undefined behavior.

Initializing multiple variables

In the last section, we noted that it is possible to define multiple variables of the same type in a single statement by separating the names with a comma:

int a, b; // create variables a and b, but do not initialize them

We also noted that best practice is to avoid this syntax altogether. However, since you may encounter other code that uses this style, it’s still useful to talk a little bit more about it, if for no other reason than to reinforce some of the reasons you should be avoiding it.

You can initialize multiple variables defined on the same line:

int a = 5, b = 6;          // copy-initialization
int c( 7 ), d( 8 );        // direct-initialization
int e { 9 }, f { 10 };     // direct-list-initialization
int i {}, j {};            // value-initialization

Unfortunately, there’s a common pitfall here that can occur when the programmer mistakenly tries to initialize both variables by using one initialization statement:

int a, b = 5;     // wrong: a is not initialized to 5!
int a = 5, b = 5; // correct: a and b are initialized to 5

In the top statement, variable a will be left uninitialized, and the compiler may or may not complain. If it doesn’t, this is a great way to have your program intermittently crash or produce sporadic results. We’ll talk more about what happens if you use uninitialized variables shortly.

The best way to remember that this is wrong is to note that each variable can only be initialized by its own initializer:

int a = 4, b = 5; // correct: a and b both have initializers
int a, b = 5;     // wrong: a doesn't have its own initializer

Unused initialized variables warnings

Modern compilers will typically generate warnings if a variable is initialized but not used (since this is rarely desirable). And if “treat warnings as errors” is enabled, these warnings will be promoted to errors and cause the compilation to fail.

Consider the following innocent looking program:

int main()
{
    int x { 5 }; // variable x defined

    // but not used anywhere

    return 0;
}

When compiling this with GCC and “treat warnings as errors” on, the following error is generated:

prog.cc: In function 'int main()':
prog.cc:3:9: error: unused variable 'x' [-Werror=unused-variable]

and the program fails to compile.

There are a few easy ways to fix this.

1. If the variable really is unused and you don’t need it, then the easiest option is to remove the definition of x (or comment it out). After all, if it’s not used, then removing it won’t affect anything.
2. Another option is to simply use the variable somewhere:

#include <iostream>

int main()
{
    int x { 5 };

    std::cout << x; // variable now used somewhere

    return 0;
}

But this requires some effort to write code that uses it, and has the downside of potentially changing your program’s behavior.

The [[maybe_unused]] attribute C++17

In some cases, neither of the above options are desirable. Consider the case where we have a set of math/physics values that we use in many different programs:

#include <iostream>

int main()
{
    // Here's some math/physics values that we copy-pasted from elsewhere
    double pi { 3.14159 };
    double gravity { 9.8 };
    double phi { 1.61803 };

    std::cout << pi << '\n';  // pi is used
    std::cout << phi << '\n'; // phi is used

    // The compiler will likely complain about gravity being defined but unused

    return 0;
}

If we use these values a lot, we probably have these saved somewhere and copy/paste/import them all together.

However, in any program where we don’t use all of these values, the compiler will likely complain about each variable that isn’t actually used. In the above example, we could easily just remove the definition of gravity. But what if there were 20 or 30 variables instead of 3? And what if we use them in multiple places? Going through the list of variables to remove/comment out the unused ones takes time and energy. And later if we need one that we’ve previously removed, we’ll have to spend more time and energy to go back and re-add/uncomment it.

To address such cases, C++17 introduced the [[maybe_unused]] attribute, which allows us to tell the compiler that we’re okay with a variable being unused. The compiler will not generate unused variable warnings for such variables.

The following program should generate no warnings/errors:

#include <iostream>

int main()
{
    [[maybe_unused]] double pi { 3.14159 };  // Don't complain if pi is unused
    [[maybe_unused]] double gravity { 9.8 }; // Don't complain if gravity is unused
    [[maybe_unused]] double phi { 1.61803 }; // Don't complain if phi is unused

    std::cout << pi << '\n';
    std::cout << phi << '\n';

    // The compiler will no longer warn about gravity not being used

    return 0;
}

Additionally, the compiler will likely optimize these variables out of the program, so they have no performance impact.

The [[maybe_unused]] attribute should only be applied selectively to variables that have a specific and legitimate reason for being unused (e.g. because you need a list of named values, but which specific values are actually used in a given program may vary).

Quiz time

Question #1

What is the difference between initialization and assignment?

Show Solution

Question #2

What form of initialization should you prefer when you want to initialize a variable with a specific value?

Show Solution

Question #3

What are default-initialization and value-initialization? What is the behavior of each? Which should you prefer?

Show Solution