Modern C++
Programming
7. Object-Oriented
Programming I
Class Concepts
Federico Busato
2024-03-29
Table of Contents
1 C++ Classes
RAII Idiom
2 Class Hierarchy
3 Access specifiers
Inheritance Access Specifiers
When Use public/protected/private/ for Data Members?
1/66
Table of Contents
4 Class Constructor
Default Constructor
Class Initialization
Uniform Initialization for Objects
Delegate Constructor
explicit Keyword
[[nodiscard]] and Classes
2/66
Table of Contents
5 Copy Constructor
6 Class Destructor
7 Defaulted Constructors, Destructor, and Operators
(=default)
3/66
Table of Contents
8 Class Keywords
this
static
const
mutable
using
friend
delete
4/66
C++ Classes
C++ Classes
C/C++ Structure
A structure (struct) is a collection of variables of the same or different data types
under a single name
C++ Class
A class (class) extends the concept of structure to hold functions as members
struct vs. class
Structures and classes are semantically equivalent.
struct represents passive objects, namely the physical state (set of data)
class represents active objects, namely the logical state (data abstraction)
5/66
Class Members - Data and Function Members
Data Member
Data within a class are called data members or class fields
Function Member
Functions within a class are called function members or methods
6/66
RAII Idiom - Resource Acquisition is Initialization
Holding a resource is a class invariant, and is tied to object
lifetime
RAII Idiom consists in three steps:
Encapsulate a resource into a class (constructor)
Use the resource via a local instance of the class
The resource is automatically released when the object gets out of scope
(destructor)
Implication 1: C++ programming language does not require the garbage collector!!
Implication 2 :The programmer has the responsibility to manage the resources
7/66
struct/class Declaration and Definition
struct declaration and definition
struct A; // struct declaration
struct A { // struct definition
int x; // data member
void f(); // function member
};
class declaration and definition
class A; // class declaration
class A { // class definition
int x; // data member
void f(); // function member
};
8/66
struct/class Function Declaration and Definition
struct A {
void g(); // function member declaration
void f() { // function member declaration
cout << "f"; // inline definition
}
};
void A::g() { // function member definition
cout << "g"; // out-of-line definition
}
9/66
struct/class Members
struct B {
void g() { cout << "g"; } // function member
};
struct A {
int x; // data member
B b; // data member
void f() { cout << "f"; } // function member
};
A a;
a.x;
a.f();
a.b.g();
10/66
Class Hierarchy
Class Hierarchy 1/3
Child/Derived Class or Subclass
A new class that inheriting variables and functions from another class is called a
derived or child class
Parent/Base Class
The closest class providing variables and functions of a derived class is called parent
or base class
Extend a base class refers to creating a new class which retains characteristics of the
base class and on top it can add (and never remove) its own members
Syntax:
class DerivedClass : [<inheritance attribute>] BaseClass {
11/66
Class Hierarchy 2/3
struct A { // base class
int value = 3;
void g() {}
};
struct B : A { // B is a derived class of A (B extends A)
int data = 4; // B inherits from A
int f() { return data; }
};
A a;
B b;
a.value;
b.g();
12/66
Class Hierarchy 3/3
struct A {};
struct B : A {};
void f(A a) {} // copy
void g(B b) {} // copy
void f_ref(A& a) {} // the same for A*
void g_ref(B& b) {} // the same for B*
A a;
B b;
f(a); // ok, also f(b), f_ref(a), g_ref(b)
g(b); // ok, also g_ref(b), but not g(a), g_ref(a)
A a1 = b; // ok, also A& a2 = b
// B b1 = a; // compile error
13/66
Access specifiers
Access specifiers 1/2
The access specifiers define the visibility of inherited members of the subsequent base
class. The keywords public , private , and protected specify the sections of
visibility
The goal of the access specifiers is to prevent direct access to the internal
representation of the class for avoiding wrong usage and potential inconsistency
(access control)
public: No restriction (function members, derived classes, outside the class)
protected: Function members and derived classes access
private: Function members only access (internal)
struct has default public members
class has default private members
14/66
Access specifiers 2/2
struct A1 {
int value; // public (by default)
protected:
void f1() {} // protected
private:
void f2() {} // private
};
class A2 {
int data; // private (by default)
};
struct B : A1 {
void h1() { f1(); } // ok, "f1" is visible in B
// void h2() { f2(); } // compile error "f2" is private in A1
};
A1 a;
a.value; // ok
// a.f1() // compile error protected
// a.f2() // compile error private
15/66
Inheritance Access Specifiers 1/3
The access specifiers are also used for defining how the visibility is propagated from
the base class to a specific derived class in the inheritance
Member
declaration
Inheritance Derived classes
public
public
public
protected protected
private \
public
protected
protected
protected protected
private \
public
private
private
protected private
private \
struct has default public members
class has default private members
16/66
Inheritance Access Specifiers 2/3
struct A {
int var1; // public
protected:
int var2; // protected
};
struct B : protected A {
int var3; // public
};
B b;
// b.var1; // compile error, var1 is protected in B
// b.var2; // compile error, var2 is protected in B
b.var3; // ok, var3 is public in B
17/66
Inheritance Access Specifiers 3/3
class A {
public:
int var1;
protected:
int var2;
};
class B1 : A {}; // private inheritance
class B2 : public A {}; // public inheritance
B1 b1;
// b1.var1; // compile error, var1 is private in B1
// b1.var2; // compile error, var2 is private in B1
B2 b2;
b2.var1; // ok, var1 is public in B2
// b2.var2; // compile error, var2 is protected in B2
18/66
When Use public/protected/private/ for Data Members?
When use protected/private data members:
They are not part of the interface, namely the logical state of the object (not
useful for the user)
They must preserve the const correctness (e.g. for pointer), see Advanced
Concepts I
When use public data members:
They can potentially change any time
const correctness is preserved for values and references, as opposite to pointers.
Data members should be preferred to member functions in this case
19/66
Class Constructor
Class Constructor
Constructor [ctor]
A constructor is a special member function of a class that is executed when a new
instance of that class is created
Goals: initialization and resource acquisition
Syntax: T(...) same named of the class and no return type
A constructor is supposed to initialize all data members
We can define multiple constructors with different signatures
Any constructor can be constexpr
20/66
Default Constructor
Default Constructor
The default constructor T() is a constructor with no argument
Every class has always either an implicit, explicit, or deleted default constructor
struct A {
A() {} // explicit default constructor
A(int) {} // user-defined (non-default) constructor
};
struct A {
int x = 3; // implicit default constructor
};
A a{}; // call the default constructor, equivalent to: A a;
Note: an implicit default constructor is constexpr
21/66
Default Constructor Examples
struct A {
A() { cout << "A"; } // default constructor
};
A a1; // call the default constructor
// A a2(); // interpreted as a function declaration!!
A a3{}; // ok, call the default constructor
// direct-list initialization (C++11)
A array[3]; // print "AAA"
A* ptr = new A[4]; // print "AAAA"
22/66
Deleted Default Constructor 1/2
The implicit default constructor of a class is marked as deleted if (simplified):
It has any user-defined constructor
struct A {
A(int x) {}
};
// A a; // compile error
It has a non-static member/base class of reference/const type
struct NoDefault { // deleted default constructor
int& x;
const int y;
};
23/66
Deleted Default Constructor 2/2
It has a non-static member/base class which has a deleted (or inaccessible)
default constructor
struct A {
NoDefault var; // deleted default constructor
};
struct B : NoDefault {}; // deleted default constructor
It has a non-static member/base class with a deleted or inaccessible destructor
struct A {
private:
A() {}
};
24/66
Initializer List
The Initializer list is used for initializing the data members of a class or explicitly call
the base class constructor before entering the constructor body
(Not to be confused with std::initializer list )
struct A {
int x, y;
A(int x1) : x(x1) {} // ": x(x1)" is the Initializer list
// direct initialization syntax
A(int x1, int y1) : // ": x{x1}, y{y1}"
x{x1}, // is the Initializer list
y{y1} {} // direct-list initialization syntax
}; // (C++11)
25/66
In-Class Member Initializer
C++11 In-class non-static data members initialization (NSDMI) allows initializing
the data members where they are declared. A user-defined constructor can be used to
override their default values
struct A {
int x = 0; // in-class member initializer
const char* str = nullptr; // in-class member initializer
A() {} // "x" and "str" are well-defined if
// the default constructor is called
A(const char* str1) : str{str1} {}
};
26/66
Data Member Initialization
const and reference data members must be initialized by using the initialization list
or by using in-class brace-or-equal-initializer syntax (C++11)
struct A {
int x;
const char y; // must be initialized
int& z; // must be initialized
int& v = x; // equal-initializer (C++11)
const int w{4}; // brace initializer (C++11)
A() : x(3), y('a'), z(x) {}
};
27/66
Initialization Order
Class member initialization follows the order of declarations and not the order in the
initialization list
struct ArrayWrapper {
int* array;
int size;
ArrayWrapper(int user_size) :
size{user_size},
array{new int[size]} {}
// wrong!!: "size" is still undefined
};
ArrayWrapper a(10);
cout << a.array[4]; // segmentation fault
28/66
Uniform Initialization for Objects
Uniform Initialization (C++11)
Uniform Initialization {}, also called list-initialization, is a way to fully initialize any
object independently of its data type
Minimizing Redundant Typenames
- In function arguments
- In function returns
Solving the “Most Vexing Parse” problem
- Constructor interpreted as function prototype
mbevin.wordpress.com/2012/11/16/uniform-initialization
29/66
Minimizing Redundant Typenames
struct Point {
int x, y;
Point(int x1, int y1) : x(x1), y(y1) {}
};
C++03
Point add(Point a, Point b) {
return Point(a.x + b.x, a.y + b.y);
}
Point c = add(Point(1, 2), Point(3, 4));
C++11
Point add(Point a, Point b) {
return { a.x + b.x, a.y + b.y }; // here
}
auto c = add({1, 2}, {3, 4}); // here
30/66
“Most Vexing Parse” problem 1/2
struct A {
A(int) {}
};
struct B {
// A a(1); // compile error It works in a function scope
A a{2}; // ok, call the constructor
};
31/66
“Most Vexing Parse” problem
2/2
struct A {};
struct B {
B(A a) {}
void f() {}
};
B b( A() ); // "b" is interpreted as function declaration
// with a single argument A (*)() (func. pointer)
// b.f() // compile error "Most Vexing Parse" problem
// solved with B b{ A{} };
32/66
Constructors and Inheritance
Class constructors are never inherited
A Derived class must call implicitly or explicitly a Base constructor before the current
class constructor
Class constructors are called in order from the top Base class to the most
Derived class (C++ objects are constructed like onions)
struct A {
A() { cout << "A" };
};
struct B1 : A { // call "A()" implicitly
int y = 3; // then, "y = 3"
};
struct B2 : A { // call "A()" explicitly
B2() : A() { cout << "B"; }
};
B1 b1; // print "A"
B2 b2; // print "A", then print "B"
33/66
Delegate Constructor
The problem:
Most constructors usually perform identical initialization steps before executing
individual operations
C++11 A delegate constructor calls another constructor of the same class to reduce
the repetitive code by adding a function that does all the initialization steps
struct A {
int a;
float b;
bool c;
// standard constructor:
A(int a1, float b1, bool c1) : a(a1), b(b1), c(c1) {
// do a lot of work
}
A(int a1, float b1) : A(a1, b1, false) {} // delegate construtor
A(float b1) : A(100, b1, false) {} // delegate construtor
};
34/66
explicit Keyword 1/2
explicit
The explicit keyword specifies that a constructor or conversion operator (C++11)
does not allow implicit conversions or copy-initialization from single arguments or
braced initializers
The problem:
struct MyString {
MyString(int n); // (1) allocate n bytes for the string
MyString(const char *p); // (21) initializes starting from a raw string
};
MyString string = 'a'; // call (1), implicit conversion!!
explicit cannot be applied to copy/move-constructors
Most C++ constructors should be explicit
35/66
explicit Keyword 2/2
struct A {
A() {}
A(int) {}
A(int, int) {}
};
void f(const A&) {}
A a1 = {}; // ok
A a2(2); // ok
A a3 = 1; // ok (implicit)
A a4{4, 5}; // ok. Selected A(int, int)
A a5 = {4, 5}; // ok. Selected A(int, int)
f({}); // ok
f(1); // ok
f({1}); // ok
struct B {
explicit B() {}
explicit B(int) {}
explicit B(int, int) {}
};
void f(const B&) {}
// B b1 = {}; // error implicit conversion
B b2(2); // ok
// B b3 = 1; // error implicit conversion
B b4{4, 5}; // ok. Selected B(int, int)
// B b5 = {4, 5}; // error implicit conversion
B b6 = (B) 1; // OK: explicit cast
// f({}); // error implicit conversion
// f(1); // error implicit conversion
// f({1}); // error implicit conversion
f(B{1}); // ok
36/66
[[nodiscard]] and Classes
C++17 allows setting [[nodiscard]] for the entire class/struct
[[nodiscard]] struct A {};
A f() { return A{}; }
auto x = f(); // ok
f(); // compiler warning
C++20 allows to set [[nodiscard]] for constructors
struct A {
[[nodiscard]] A() {} // C++20 also allows [[nodiscard]] with a reason
};
void f(A {})
A a{}; // ok
f(A{}); // ok
A{}; // compiler warning
37/66
Copy Constructor
Copy Constructor
Copy Constructor
A copy constructor T(const T&) creates a new object as a deep copy of an
existing object
struct A {
A() {} // default constructor
A(int) {} // non-default constructor
A(const A&) {} // copy constructor direct initialization
}
38/66
Copy Constructor Details
Every class always defines an implicit or explicit copy constructor, potentially
deleted
The copy constructor implicitly calls the default Base class constructor
Even the copy constructor is considered a user-defined constructor
The copy constructor doesn’t have template parameters, otherwise it is a standard
member function
The copy constructor must not be confused with the assignment operator
operator=
MyStruct x;
MyStruct y{x}; // copy constructor
y = x; // call the assignment operator=, not the copy constructor
// copy initialization, see next lecture
39/66
Copy Constructor Example
struct Array {
int size;
int* array;
Array(int size1) : size{size1} {
array = new int[size];
}
// copy constructor, ": size{obj.size}" initializer list
Array(const Array& obj) : size{obj.size} {
array = new int[size];
for (int i = 0; i < size; i++)
array[i] = obj.array[i];
}
};
Array x{100}; // do something with x.array ...
Array y{x}; // call "Array::Array(const Array&)"
40/66
Copy Constructor Usage
The copy constructor is used to:
Initialize one object from another one having the same type
- Direct constructor
- Assignment operator
A a1;
A a2(a1); // Direct copy initialization
A a3{a1}; // Direct copy initialization
A a4 = a1; // Copy initialization
A a5 = {a1}; // Copy list initialization
Copy an object which is passed by-value as input parameter of a function
void f(A a);
Copy an object which is returned as result from a function*
A f() { return A(3); } // * see RVO optimization
41/66
Copy Constructor Usage Examples
struct A {
A() {}
A(const A& obj) { cout << "copy"; }
};
void f(A a) {} // pass by-value
A g1(A& a) { return a; }
A g2() { return A(); }
A a;
A b = a; // copy constructor (assignment) "copy"
A c(b); // copy constructor (direct) "copy"
f(b); // copy constructor (argument) "copy"
g1(a); // copy constructor (return value) "copy"
A d = g2(); // * see RVO optimization (Advanced Concepts I)
42/66
Pass by-value and Copy Constructor
struct A {
A() {}
A(const A& obj) { cout << "expensive copy"; }
};
struct B : A {
B() {}
B(const B& obj) { cout << "cheap copy"; }
};
void f1(B b) {}
void f2(A a) {}
B b1;
f1(b1); // cheap copy
f2(b1); // expensive copy!! It calls A(const A&) implicitly
43/66
Deleted Copy Constructor
The implicit copy constructor of a class is marked as deleted if (simplified):
It has a non-static member/base class of reference/const type
struct NonDefault { int& x; }; // deleted copy constructor
It has a non-static member/base class which has a deleted (or inaccessible) copy
constructor
struct B { // deleted copy constructor
NonDefault a;
};
struct B : NonDefault {}; // delete copy constructor
It has a non-static member/base class with a deleted or inaccessible destructor
The class has the move constructor (next lectures)
44/66
Class Destructor
Class Destructor 1/3
Destructor [dtor]
A destructor is a special member function that is executed whenever an object is
out-of-scope or whenever the delete/delete[] expression is applied to a pointer
of that class
Goals: resources releasing
Syntax: T() same name of the class and no return type
Any object has exactly one destructor, which is always implicitly or explicitly
declared
C++20 The destructor can be constexpr
45/66
Class Destructor 2/3
struct Array {
int* array;
Array() { // constructor
array = new int[10];
}
Array() { // destructor
delete[] array;
}
};
int main() {
Array a; // call the constructor
for (int i = 0; i < 5; i++)
Array b; // call 5 times the constructor + destructor
} // call the destructor of "a"
46/66
Class Destructor - Order of Calls 3/3
Class destructor is never inherited. Base class destructor is invoked after the
current class destructor
Class destructors are called in reverse order. From the most Derived to the top
Base class
struct A {
A() { cout << "A"; }
};
struct B {
B() { cout << "B"; }
};
struct C : A {
B b; // call B()
C() { cout << "C"; }
};
int main() {
C b; // print "C", then "B", then "A"
}
47/66
Defaulted
Constructors,
Destructor, and
Operators
(=default)
Defaulted Constructors, Destructor, and Operators (=default) 1/3
C++11 The compiler can automatically generate
default/copy/move constructors
A() = default
A(const A&) = default
A(A&&) = default
destructor
A() = default
copy/move assignment operators A& operator=(const A&) = default
A& operator=(A&&) = default
spaceship operator
auto operator<=>(const A&) const = default
= default implies constexpr , but not noexcept or explicit
48/66
Defaulted Constructors, Destructor, and Operators (=default) 1/3
When the compiler-generated constructors, destructors, and operators are useful:
Change the visibility of non-user provided constructors and assignment operators
( public , protected , private )
Make visible the declarations of such members
The defaulted default constructor has a
::::::
similar effect as a user-defined constructor
with empty body and empty initializer list
When the compiler-generated constructor is useful:
Any user-provided constructor disables implicitly-generated default constructor
Force the default values for the class data members
49/66
Defaulted Constructors, Destructor, and Operators (=default) 3/3
struct A {
A(int v1) {} // delete implicitly-defined default ctor because
// a user-provided constructor is defined
A() = default; // now, A has the default constructor
};
struct B {
protected:
B() = default; // now it is protected
};
struct C {
int x;
// C() {} // 'x' is undefined
C() = default; // 'x' is zero
};
50/66
Class Keywords
this Keyword
this
Every object has access to its own address through the pointer this
Explicit usage is not mandatory (and not suggested)
this is necessary when:
The name of a local variable is equal to some member name
Return reference to the calling object
struct A {
int x;
void f(int x) {
this->x = x; // without "this" has no effect
}
const A& g() {
return *this;
}
};
51/66
static Keyword 1/5
static Keyword
The keyword static declares members (fields or methods) that are not bound to
class instances. A static member is shared by all objects of the class
struct A {
int x;
int f() { return x; }
static int g() { return 3; } // g() cannot access 'x' as it is associated
}; // with class instances
A a{4};
a.f(); // call the class instance method
A::g(); // call the static class method
a.g(); // as an alternative, a class instance can access static class members
52/66
static Keyword - Constant Members 2/5
struct A {
static const int a = 4; // C++03
static constexpr float b = 4.2f; // better, C++11
// static const float c = 4.2f; // only GNU extension (GCC)
static constexpr int f() { return 1; } // ok, C++11
// static const int g() { return 1; } // 'const' refers to the return type
};
53/66
static Keyword - Mutable Members 3/5
Non- const static data members cannot be directly initialized inline (see
Translation Units lecture)...before C++17
struct A {
// static int a = 4; // compiler error
static int a; // ok, declaration only
static inline int b = 4; // ok from C++17
static int f() { return 2; }
static int g(); // ok, declaration only
};
int A::a = 4; // ok, undefined reference without this definition
int A::g() { return 3; } // ok, undefined reference without this definition
54/66
static Keyword - Example 4/5
struct A {
static int x; // declaration
static int f() { return x; }
static int& g() { return x; }
};
int A::x = 3; // definition
//---------------------------------------------------------------------------------
A::f(); // return 3
A::x++;
A::f(); // return 4
A::g() = 7;
A::f(); // return 7
55/66
static Keyword - Member Visibility 5/5
A static member function can only access static class members
A non- static member function can access static class members
struct A {
int x = 3;
static inline int y = 4;
int f1() { return x; } // ok
// static int f2() { return x; } // compiler error, 'x' is not visible
int g1() { return y; } // ok
static int g2() { return y; } // ok
struct B {
int h() { return y + g2(); } // ok
}; // 'x', 'f1()', 'g1()' are not visible within 'B'
};
56/66
const Keyword 1/3
Const member functions
Const member functions (inspectors or observers) are functions marked with
const that are not allowed to change the object logical state
The compiler prevents from inadvertently mutating/changing the data members of
observer functions All data members are marked const within an observer
method, including the this pointer
The physical state can still be modified, see mutable member functions
Member functions without a const suffix are called non-const member functions
or mutators/modifiers
57/66
const Keyword 2/3
struct A {
int x = 3;
int* p;
int get() const {
// x = 2; // compile error class variables cannot be modified
// p = nullptr; // compile error class variables cannot be modified
p[0] = 3; // ok, p is 'int* const' -> its content is
// not protected
return x;
}
};
A common case where const member functions are useful is to enforce const correctness when
accessing pointers, see Advanced Concepts I, Const Correctness
58/66
const Keyword - const Overloading 3/3
The const keyword is part of the function signature. Therefore, a class can
implement two similar methods, one which is called when the object is const , and
one that is not
class A {
int x = 3;
public:
int& get1() { return x; } // read and write
int get1() const { return x; } // read only
int& get2() { return x; } // read and write
};
A a1;
cout << a1.get1(); // ok
cout << a1.get2(); // ok
a1.get1() = 4; // ok
const A a2;
cout << a2.get1(); // ok
// cout << a2.get2(); // compile error "a2" is const
//a2.get1() = 5; // compile error only "get1() const" is available
59/66
mutable Keyword
mutable
mutable data members of const class instances are modifiable. They should be
part of the object physical state, but not of the logical state
It is particularly useful if most of the members should be constant but a few need to be
modified
Conceptually, mutable members should not change anything that can be retrieved from
the class interface
struct A {
int x = 3;
mutable int y = 5;
};
const A a;
// a.x = 3; // compiler error const
a.y = 5; // ok
60/66
using Keyword for type declaration
The using keyword is used to declare a type alias tied to a specific class
struct A {
using type = int;
};
typename A::type x = 3; // "typename" keyword is needed when we refer to types
struct B : A {};
typename B::type x = 4; // B can use "type" as it is public in A
61/66
using Keyword for Inheritance
The using keyword can be also used to change the inheritance attribute of member
data or functions
struct A {
protected:
int x = 3;
};
struct B : A {
public:
using A::x;
};
B b;
b.x = 3; // ok, "b.x" is public
62/66
friend Keyword 1/3
friend Class
A friend class can access the private and protected members of the class in
which it is declared as a friend
Friendship properties:
Not Symmetric: if class A is a friend of class B, class B is not automatically a
friend of class A
Not Transitive: if class A is a friend of class B, and class B is a friend of class C,
class A is not automatically a friend of class C
Not Inherited: if class Base is a friend of class X, subclass Derived is not
automatically a friend of class X; and if class X is a friend of class Base, class X is
not automatically a friend of subclass Derived
63/66
friend Keyword 2/3
class B; // class declaration
class A {
friend class B;
int x; // private
};
class B {
int f(A a) { return a.x; } // ok, B is friend of A
};
class C : B {
// int f(A a) { return a.x; } // compile error not inherited
};
64/66
friend Keyword 3/3
friend Method
A non-member function can access the private and protected members of a class
if it is declared a friend of that class
class A {
int x = 3; // private
friend int f(A a); // friendship declaration, no implementation
};
//'f' is not a member function of any class
int f(A a) {
return a.x; // A is friend of f(A)
}
friend methods are commonly used for implementing the stream operator operator<<
65/66
delete Keyword
delete Keyword (C++11)
The delete keyword explicitly marks a member function as deleted and any use
results in a compiler error. When it is applied to copy/move constructor or
assignment, it prevents the compiler from implicitly generating these functions
The default copy/move functions for a class can produce unexpected results. The
keyword delete prevents these errors
struct A {
A() = default;
A(const A&) = delete; // e.g. deleted because unsafe or expensive
};
void f(A a) {} // implicit call to copy constructor
A a;
// f(a); // compile error marked as deleted
66/66