What Is a Constant?
Imagine that you are writing a program to calculate the area and the circumference of a circle. The formulas are
area = pi * radius * radius; circumference = 2 * pi * radius
In these formulas,piis a constant of value 22 / 7. You don’t want the value ofpito change anywhere in your program. You also want to avoid any accidental assignments of possibly incorrect values topi. C++ enables you to definepias a constant that cannot be changed after declaration. In other words, after it’s defined, the value of a constant cannot be altered. Assignments to a constant in C++ cause compilation errors.
因此,常量就像变量在c++中除了that they cannot be changed. Much like a variable, a constant also occupies space in memory and has a name to identify the address where the space is reserved. However, the content of this space cannot be overwritten. Constants in C++ can be
■ Literal constants
■ Declared constants using theconstkeyword
■ Constant expressions using theconstexprkeyword
■ Enumerated constants using theenumkeyword
■ Defined constants (although they are not recommended and have been deprecated)
Literal Constants
Literal constants can be of many types: integer, string, and so on. In your first C++ program inListing 1.1, you displayed “Hello World” using the following statement:
std::cout << "Hello World" << std::endl;
In this code,"Hello World"is a string literal constant. You literally have been using literal constants ever since then! When you declare an integersomeNumber, like this:
int someNumber = 10;
the integer variablesomeNumberis assigned the initial value 10. Here decimal 10 is a part of the code, gets compiled into the application, is unchangeable, and is a literal constant, too. You can initialize the integer by using a literal in octal notation, like this:
int someNumber = 012 // octal 12 evaluates to decimal 10
You can also use binary literals, like this:
int someNumber = 0b1010; // binary 1010 evaluates to decimal 10
Declaring Variables as Constants Usingconst
The most important type of constants in C++ are declared by using the keywordconstbefore the variable type. The syntax of a generic declaration looks like this:
const type-name constant-name = value;
Listing 3.7 shows a simple application that displays the value of a constant calledpi.
Input▼
Listing 3.7Declaring a Constant Called pi
1: #include2: 3: int main() 4: { 5: using namespace std; 6: 7: const double pi = 22.0 / 7; 8: cout << "The value of constant pi is: " << pi << endl; 9: 10: // Uncomment next line to fail compilation 11: // pi = 345; // error, assignment to a constant 12: 13: return 0; 14: }
Output▼
The value of constant pi is: 3.14286
Analysis▼
Note the declaration of the constantpiin Line 7. You use theconstkeyword to tell the compiler thatpiis a constant of typedouble. If you uncomment Line 11, which assigns a value to a constant, you get a compile failure that says something similar to, “You cannot assign to a variable that isconst.” Thus, using constants is a powerful way to ensure that certain data cannot be modified.
Constants are useful when declaring the length of a static array, which is fixed at compile time. Listing 4.2 in Lesson 4, “Managing Arrays and Strings,” includes an example that demonstrates the use of aconst intto define the length of an array.
Constant Expressions Usingconstexpr
The keywordconstexprinstructs the compiler to compute the expression, if possible. For example, a simple function that divides two numbers may be declared as aconstexpr:
constexpr double Div_Expr(double a, double b) { return a / b; }
The function can be used by a variable that is also declared as aconstexpr:
constexpr double pi = Div_Expr(22, 7); // Div_Expr() is executed by compiler, pi assigned at compile time
Thus,constexprallows for optimization possibilities where some simple computation might be performed by the compiler. In the example above,Div_Expr()is invoked with arguments that are integral constants 22 and 7. Hence, the compiler is able to computepi. If the arguments were not constants but plain integers, then you would still be able to useDiv_Expr(), but the division would be performed at runtime, and you would not be able to assign it to aconstexpr:
int a = 22, b = 7; const double pi = Div_Expr(a, b); // Div_Expr() executed at runtime because arguments are not constants
C++20 Immediate Functions Usingconsteval
In the previous section, you saw howDiv_Expr()is treated by the compiler as a constant expression when invoked with constants, and you saw how the result of this constant expression is evaluated by the compiler. However, whenDiv_Expr()is invoked with plain integer variables, the compiler treats it as an ordinary function that is executed at runtime.
C++20 introducesimmediate functionsthat are required to be executed by the compiler. You declare an immediate function by using keywordconsteval:
consteval double Div_Eval(double a, double b) { return a / b; }
Div_Eval()can only be invoked with arguments that are constants themselves. The compiler performs the division and assigns the return value to the point in code where the function is called:
const double pi = Div_Eval(22, 7); // compiler assigns the value of pi
Unlike withDiv_Expr(), if you were to invokeDivEval()using plain integers, the compilation would fail:
int a = 22, b = 7; double pi = Div_Eval(a, b); // fail: non-const arguments to consteval fn.
Listing 3.8 demonstrates the use ofconstexprandconsteval.
Input▼
Listing 3.8Usingconstevalandconstexprto Calculate Pi and Multiples of Pi
0: #include1: consteval double GetPi() { return 22.0 / 7; } 2: constexpr double XPi(int x) { return x * GetPi(); } 3: 4: int main() 5: { 6: using namespace std; 7: constexpr double pi = GetPi(); 8: 9: cout << "constexpr pi evaluated by compiler to " << pi << endl; 10: cout << "constexpr XPi(2) evaluated by compiler to " << XPi(2) << endl; 11: 12: int multiple = 5; 13: cout << "(non-const) integer multiple = " << multiple << endl; 14: cout << "constexpr is ignored when XPi(multiple) is invoked, "; 15: cout << "returns " << XPi(multiple) << endl; 16: 17: return 0; 18: }
Output▼
constexpr pi evaluated by compiler to 3.14286 constexpr XPi(2) evaluated by compiler to 6.28571 (non-const) integer multiple = 5 constexpr is ignored when XPi(multiple) is invoked, returns 15.7143
Analysis▼
In Lines 1 and 2, the program demonstrates the use ofconstevalandconstexpr, respectively.GetPi()in Line 1 is an immediate function. When the compiler encountersGetPi()in Line 7,constevalinstructs the compiler to compute the value of pi resulting from the division and initialize constantpiwith this value,3.14286, in Line 7.GetPi()never makes it to the compiled executable. Line 2 contains aconstexprinXPi(int). Its usage in Line 10 results in the compiler substitutingXPi(2)with6.28571becauseXPi()has been invoked with a constant, the integer value2. The same functionXPi(), when invoked in Line 15 with the variablemultiple, results in the compiler ignoringconstexprand integratingXPi(multiple)into the code as a regular function call.
If you were to change the declaration ofXPi()in Line 2 fromconstexprtoconsteval, you would require the compiler to necessarily compute its return value at every usage ofXPi()in the code and replace it with the computed value. This would not be possible for the usage ofXPiin Line 15 with a non-constinteger, and the compilation would fail. This little example therefore also demonstrates the subtle differences betweenconstevalandconstexpr.
Enumerations
There are situations in which a particular variable should be allowed to accept only a certain set of values. For example, you might not want the colors of the rainbow to contain turquoise or the directions on a compass to contain left. In both these cases, you need a type of variable whose values are restricted to a certain set defined by you.Enumerations, which are characterized by the keywordenum, are exactly the tool you need in such situations. An enumeration comprises a set of constants calledenumerators.
In the following example, the enumerationRainbowColorscontains individual colors such asVioletas enumerators:
enum RainbowColors { Violet = 0, Indigo, Blue, Green, Yellow, Orange, Red };
Here’s another enumeration for the cardinal directions:
enum CardinalDirections { North, South, East, West };
Enumerations are used as user-defined types. Variables of this type can be assigned a range of values restricted to the enumerators contained in the enumeration. So, if defining a variable that contains the colors of a rainbow, you declare the variable like this:
RainbowColors myFavoriteColor = Blue; // Initial value
In this line of code, you declaremyFavoriteColorto be of typeRainbowColors.This variable is restricted to store any of the specified VIBGYOR colors and no other values.
Listing 3.9 demonstrates how enumerated constants are used to hold the four cardinal directions, with an initializing value supplied to the first one.
Input▼
Listing 3.9Using Enumerated Values to Indicate Cardinal Wind Directions
1: #include2: using namespace std; 3: 4: enum CardinalDirections 5: { 6: North = 25, 7: South, 8: East, 9: West 10: }; 11: 12: int main() 13: { 14: cout << "Displaying directions and their symbolic values" << endl; 15: cout << "North: " << North << endl; 16: cout << "South: " << South << endl; 17: cout << "East: " << East << endl; 18: cout << "West: " << West << endl; 19: 20: CardinalDirections windDirection = South; 21: cout << "Variable windDirection = " << windDirection << endl; 22: 23: return 0; 24: }
Output▼
Displaying directions and their symbolic values North: 25 South: 26 East: 27 West: 28 Variable windDirection = 26
Analysis▼
Note that this listing enumerates the four cardinal directions but gives the first direction,North, an initial value of25(see Line 6). This automatically ensures that the following constants are assigned values26,27, and28by the compiler, as demonstrated in the output. In Line 20, you create a variable of typeCardinalDirectionsthat is assigned an initial valueSouth. When displayed on the screen in Line 21, the compiler uses the integer value associated withSouth, which is26.
Scoped Enumerations
The enumerated typeCardinalDirectionsis defined as anunscopedenumeration. The compiler lets you convert variables of this type into integers, and therefore the following statement would be valid:
int someNumber = South;
This flexibility, however, defeats the very purpose of using enumerations. You’re therefore advised to use scoped enumerations instead. Introduced in 2011 as part of C++11, scoped enumerations are declared using theclassorstructkeyword followingenum:
enum class CardinalDirections {North, South, East, West};
When declaring a variable of typeCardinalDirections, you then use the scope resolution operator::as follows:
CardinalDirections dir = CardinalDirections::South;
Scoped enumerations are safer because the compiler ensures strict type safety, which makes the following assignments invalid:
int someNumber = CardinalDirections::South; // error int someNumber = dir; // error
As a scoped enumeration,CardinalDirectionsensures that variables of its type can only be assigned directly to other variables of the same type:
CardinalDirections dir2 = dir; // OK
Defining Constants by Using#define
First and foremost, don’t use#defineif you are writing a program from scratch. The only reason this book explains the definition of constants using#defineis to help you understand legacy code that define constants using this format:
#define pi 3.14286
#defineis a preprocessor macro. In the example above, it causes all following mentions ofpito be replaced by3.14286for the compiler to process. Note that this is a text replacement (read: non-intelligent replacement) done by the preprocessor. The compiler neither knows nor cares about the actual type of the constant in question.