Member access operators

From cppreference.com
< cpp‎ | language
 
 
C++ language
General topics
Flow control
Conditional execution statements
if
Iteration statements (loops)
for
range-for (C++11)
Jump statements
Functions
Function declaration
Lambda function expression
inline specifier
Dynamic exception specifications (until C++17*)
noexcept specifier (C++11)
Exceptions
Namespaces
Types
Specifiers
decltype (C++11)
auto (C++11)
alignas (C++11)
Storage duration specifiers
Initialization
Expressions
Alternative representations
Literals
Boolean - Integer - Floating-point
Character - String - nullptr (C++11)
User-defined (C++11)
Utilities
Attributes (C++11)
Types
typedef declaration
Type alias declaration (C++11)
Casts
Implicit conversions - Explicit conversions
static_cast - dynamic_cast
const_cast - reinterpret_cast
Memory allocation
Classes
Class-specific function properties
explicit (C++11)
static
Special member functions
Templates
Miscellaneous
 
Expressions
General
Value categories (lvalue, rvalue, xvalue)
Order of evaluation (sequence points)
Constant expressions
Potentially-evaluated expressions
Primary expressions
Lambda expressions(C++11)
Literals
Integer literals
Floating-point literals
Boolean literals
Character literals including escape sequences
String literals
Null pointer literal(C++11)
User-defined literal(C++11)
Operators
a=b, a+=b, a-=b, a*=b, a/=b, a%=b, a&=b, a|=b, a^=b, a<<=b, a>>=b
++a, --a, a++, a--
+a, -a, a+b, a-b, a*b, a/b, a%b, ~a, a&b, a|b, a^b, a<<b, a>>b
a||b, a&&b, !a
a==b, a!=b, a<b, a>b, a<=b, a>=b, a<=>b (since C++20)
a[b], *a, &a, a->b, a.b, a->*b, a.*b
a(...), a,b, a?b:c
new-expression
delete-expression
throw-expression
alignof
sizeof
sizeof...(C++11)
typeid
noexcept(C++11)
Fold expressions(C++17)
Alternative representations of operators
Precedence and associativity
Operator overloading
Default comparisons(C++20)
Conversions
Implicit conversions
Usual arithmetic conversions
const_cast
static_cast
reinterpret_cast
dynamic_cast
Explicit conversions: (T)a, T(a), auto(a), auto{a} (since C++23)
User-defined conversion
 

Accesses a member of its operand.

Operator name Syntax Over​load​able Prototype examples (for class T)
Inside class definition Outside class definition
subscript a[b]

a[...]

(since C++23)
Yes R& T::operator[](S b);

R& T::operator[](...);

(since C++23)
N/A
indirection *a Yes R& T::operator*(); R& operator*(T a);
address-of &a Yes R* T::operator&(); R* operator&(T a);
member of object a.b No N/A N/A
member of pointer a->b Yes R* T::operator->(); N/A
pointer to member of object a.*b No N/A N/A
pointer to member of pointer a->*b Yes R& T::operator->*(S b); R& operator->*(T a, S b);
Notes
  • As with most user-defined overloads, return types should match return types provided by the built-in operators so that the user-defined operators can be used in the same manner as the built-ins. However, in a user-defined operator overload, any type can be used as return type (including void). One exception is operator->, which must return a pointer or another class with overloaded operator-> to be realistically usable.

Explanation

Built-in subscript operator provides access to an object pointed-to by the pointer or array operand.

Built-in indirection operator provides access to an object or function pointed-to by the pointer operand.

Built-in address-of operator creates a pointer pointing to the object or function operand.

Member of object and pointer to member of object operators provide access to a data member or member function of the object operand.

Built-in member of pointer and pointer to member of pointer operators provide access to a data member or member function of the class pointed-to by the pointer operand.

Built-in subscript operator

The subscript operator expressions have the form

expr1 [expr2 ] (1)
expr1 [{expr , ...}] (2) (since C++11)
expr1 [expr2 , expr , ...] (3) (since C++23)
1) For the built-in operator, one of the expressions (either expr1 or expr2) must be a glvalue of type “array of T” or a prvalue of type “pointer to T”, while the other expression (expr2 or expr1, respectively) must be a prvalue of unscoped enumeration or integral type. The result of this expression has the type T. expr2 cannot be an unparenthesized comma expression. (since C++23)
2) The form with brace-enclosed list inside the square brackets is only used to call an overloaded operator[].
3) The form with comma-separated expression list inside the square brackets is only used to call an overloaded operator[].

The built-in subscript expression E1[E2] is exactly identical to the expression *(E1 + E2) except for its value category (see below) and evaluation order (since C++17): the pointer operand (which may be a result of array-to-pointer conversion, and which must point to an element of some array or one past the end) is adjusted to point to another element of the same array, following the rules of pointer arithmetic, and is then dereferenced.

When applied to an array, the subscript expression is an lvalue if the array is an lvalue, and an xvalue if it isn't (since C++11).

When applied to a pointer, the subscript expression is always an lvalue.

The type T is not allowed to be an incomplete type, even if the size or internal structure of T is never used, as in &x[0].

Using an unparenthesized comma expression as second (right) argument of a subscript operator is deprecated.

For example, a[b, c] is deprecated and a[(b, c)] is not.

(since C++20)
(until C++23)

An unparenthesized comma expression cannot be second (right) argument of a subscript operator. For example, a[b, c] is either ill-formed or equivalent to a.operator[](b, c).

Parentheses are needed to for using a comma expression as the subscript, e.g., a[(b, c)].

(since C++23)

In overload resolution against user-defined operators, for every object type T (possibly cv-qualified), the following function signature participates in overload resolution:

T& operator[](T*, std::ptrdiff_t);
T& operator[](std::ptrdiff_t, T*);
#include <iostream>
#include <map>
#include <string>
 
int main()
{
    int a[4] = {1, 2, 3, 4};
    int* p = &a[2];
    std::cout << p[1] << p[-1] << 1[p] << (-1)[p] << '\n';
 
    std::map<std::pair<int, int>, std::string> m;
    m[{1, 2}] = "abc"; // uses the [{...}] version
}

Output:

4242

Built-in indirection operator

The indirection operator expressions have the form

*expr

The operand of the built-in indirection operator must be pointer to object or a pointer to function, and the result is the lvalue referring to the object or function to which expr points.

A pointer to (possibly cv-qualified) void cannot be dereferenced. Pointers to other incomplete types can be dereferenced, but the resulting lvalue can only be used in contexts that allow an lvalue of incomplete type, e.g. when initializing a reference.

In overload resolution against user-defined operators, for every type T that is either object type (possibly cv-qualified) or function type (not const- or ref-qualified), the following function signature participates in overload resolution:

T& operator*(T*);
#include <iostream>
 
int f() { return 42; }
 
int main()
{
    int n = 1;
    int* pn = &n;
 
    int& r = *pn; // lvalue can be bound to a reference
    int m = *pn;  // indirection + lvalue-to-rvalue conversion
 
    int (*fp)() = &f;
    int (&fr)() = *fp; // function lvalue can be bound to a reference
 
    [](...){}(r, m, fr); // removes possible "unused variable" warnings
}

Built-in address-of operator

The address-of operator expressions have the form

&expr (1)
&class ::member (2)
1) If the operand is an lvalue expression of some object or function type T, operator& creates and returns a prvalue of type T*, with the same cv qualification, that is pointing to the object or function designated by the operand. If the operand has incomplete type, the pointer can be formed, but if that incomplete type happens to be a class that defines its own operator&, it is unspecified whether the built-in or the overload is used. For the operands of type with user-defined operator&, std::addressof may be used to obtain the true pointer. Note that, unlike C99 and later C versions, there's no special case for the unary operator& applied to the result of the unary operator*.
If the operand is the name of an overloaded function, the address may be taken only if the overload can be resolved due to context. See Address of an overloaded function for details.

If expr names an explicit object member function, expr must be a qualified identifier. Applying & to an unqualified identifier naming an explicit object member function is ill-formed.

(since C++23)
2) If the operand is a qualified name of a non-static or variant member other than an explicit object member function (since C++23), e.g. &C::member, the result is a prvalue pointer to member function or pointer to data member of type T in class C. Note that neither &member nor C::member nor even &(C::member) may be used to initialize a pointer to member.

In overload resolution against user-defined operators, this operator does not introduce any additional function signatures: built-in address-of operator does not apply if there exists an overloaded operator& that is a viable function.

void f(int) {}
void f(double) {}
 
struct A { int i; };
struct B { void f(); };
 
int main()
{
    int n = 1;
    int* pn = &n;    // pointer
    int* pn2 = &*pn; // pn2 == pn
 
    int A::* mp = &A::i;      // pointer to data member
    void (B::*mpf)() = &B::f; // pointer to member function
 
    void (*pf)(int) = &f; // overload resolution due to initialization context
//  auto pf2 = &f; // error: ambiguous overloaded function type
    auto pf2 = static_cast<void (*)(int)>(&f); // overload resolution due to cast
}

Built-in member access operators

The member access operator expressions have the form

expr .template(optional) id-expr (1)
expr ->template(optional) id-expr (2)
expr .pseudo-destructor (3)
expr ->pseudo-destructor (4)
1) The expr must be an expression of complete class type T.
2) The expr must be an expression of pointer to complete class type T*.
3,4) The expr must be an expression of scalar type (see below).

expr is evaluated even if it is not necessary (e.g. when the second operand names a static member).

id-expr is a name of (formally, an id-expression that names) a data member or member function of T or of an unambiguous and accessible base class B of T (e.g. E1.E2 or E1->E2), optionally qualified (e.g. E1.B::E2 or E1->B::E2), optionally using template disambiguator (e.g. E1.template E2 or E1->template E2).

If a user-defined operator-> is called, operator-> is called again on the resulting value, recursively, until an operator-> is reached that returns a plain pointer. After that, built-in semantics are applied to that pointer.

The expression E1->E2 is exactly equivalent to (*E1).E2 for built-in types; that is why the following rules address only E1.E2.

In the expression E1.E2:

1) if E2 is a static data member:
  • if E2 is of reference type T& or T&& (since C++11), the result is an lvalue of type T designating the object or function to which the reference is bound,
  • otherwise, given the type of E2 as T, the result is an lvalue of type T designating that static data member.
Essentially, E1 is evaluated and discarded in both cases;
2) if E2 is a non-static data member:
  • if E2 is of reference type T& or T&& (since C++11), the result is an lvalue of type T designating the object or function to which the corresponding reference member of E1 is bound,
  • otherwise, if E1 is an lvalue, the result is an lvalue designating that non-static data member of E1,
  • otherwise (if E1 is an rvalue (until C++17)xvalue (which may be materialized from prvalue) (since C++17)), the result is an rvalue (until C++11)xvalue (since C++11) designating that non-static data member of E1.
If E2 is not a mutable member, the cv-qualification of the result is the union of the cv-qualifications of E1 and E2, otherwise (if E2 is a mutable member), it is the union of the volatile-qualifications of E1 and E2;
3) if E2 is a static member function, the result is an lvalue designating that static member function. Essentially, E1 is evaluated and discarded in this case;
4) if E2 is a non-static member function including a destructor, the result is a special kind of prvalue designating that non-static member function of E1 that can only be used as the left-hand operand of a member function call operator, and for no other purpose;
5) if E2 is a member enumerator, given the type of E2 as T, the result is an rvalue (until C++11)a prvalue (since C++11) of type T whose value is the value of the enumerator;
6) if E2 is a nested type, the program is ill-formed;
7) if E1 has a ScalarType and E2 is a ~ followed by the type name or decltype specifier designating the same type (minus cv-qualifications), optionally qualified, the result is a special kind of prvalue that can only be used as the left-hand operand of a function call operator, and for no other purpose. The resulting function call expression is called pseudo destructor call. It takes no arguments, returns void, evaluates E1, and ends the lifetime of its result object. This is the only case where the left-hand operand of operator. has non-class type. Allowing pseudo destructor call makes it possible to write code without having to know if a destructor exists for a given type.

operator. cannot be overloaded, and for operator->, in overload resolution against user-defined operators, the built-in operator does not introduce any additional function signatures: built-in operator-> does not apply if there exists an overloaded operator-> that is a viable function.

#include <cassert>
#include <iostream>
#include <memory>
 
struct P
{
    template<typename T>
    static T* ptr() { return new T; }
};
 
template<typename T>
struct A
{
    A(int n): n(n) {}
 
    int n;
    static int sn;
 
    int f() { return 10 + n; }
    static int sf() { return 4; }
 
    class B {};
    enum E {RED = 1, BLUE = 2};
 
    void g()
    {
        typedef int U;
 
        // keyword template needed for a dependent template member
        int* p = T().template ptr<U>();
        p->~U(); // U is int, calls int's pseudo destructor
        delete p;
    }
};
 
template<>
int A<P>::sn = 2;
 
struct UPtrWrapper
{
    std::unique_ptr<std::string> uPtr;
    std::unique_ptr<std::string>& operator->() { return uPtr; }
};
 
int main()
{
    A<P> a(1);
    std::cout << a.n << ' '
              << a.sn << ' '   // A::sn also works
              << a.f() << ' ' 
              << a.sf() << ' ' // A::sf() also works
//            << a.B << ' '    // error: nested type not allowed
              << a.RED << ' '; // enumerator
 
    UPtrWrapper uPtrWrap{std::make_unique<std::string>("wrapped")};
    assert(uPtrWrap->data() == uPtrWrap.operator->().operator->()->data());
}

Output:

1 2 11 4 1

Built-in pointer-to-member access operators

The member access operator expressions through pointers to members have the form

lhs .*rhs (1)
lhs ->*rhs (2)
1) lhs must be an expression of class type T.
2) lhs must be an expression of type pointer to class type T*.

The second operand of both operators is an expression of type pointer to member ( data or function) of T or pointer to member of an unambiguous and accessible base class B of T.

The expression E1->*E2 is exactly equivalent to (*E1).*E2 for built-in types; that is why the following rules address only E1.*E2.

In the expression E1.*E2:

1) if E2 is a pointer to data member,
  • if E1 is an lvalue, the result is an lvalue designating that data member,
  • otherwise (if E1 is an rvalue (until C++17)xvalue (which may be materialized from prvalue) (since C++17)), the result is an rvalue (until C++11)xvalue (since C++11) designating that data member;
2) if E2 is a pointer to member function, the result is a special kind of prvalue designating that member function that can only be used as the left-hand operand of a member function call operator, and for no other purpose;
3) cv-qualification rules are the same as for member of object operator, with one additional rule: a pointer to member that refers to a mutable member cannot be used to modify that member in a const object;
4) if E2 is a null pointer-to-member value, the behavior is undefined;
5) if the dynamic type of E1 does not contain the member to which E2 refers, the behavior is undefined;
6) if E1 is an rvalue and E2 points to a member function with ref-qualifier &, the program is ill-formed unless the member function has the cv-qualifier const but not volatile (since C++20);
7) if E1 is an lvalue and E2 points to a member function with ref-qualifier &&, the program is ill-formed.
(since C++11)

In overload resolution against user-defined operators, for every combination of types D, B, R, where class type B is either the same class as D or an unambiguous and accessible base class of D, and R is either an object or function type, the following function signature participates in overload resolution:

R& operator->*(D*, R B::*);

where both operands may be cv-qualified, in which case the return type's cv-qualification is the union of the cv-qualification of the operands.

#include <iostream>
 
struct S
{
    S(int n) : mi(n) {}
    mutable int mi;
    int f(int n) { return mi + n; }
};
 
struct D : public S
{
    D(int n) : S(n) {}
};
 
int main()
{
    int S::* pmi = &S::mi;
    int (S::* pf)(int) = &S::f;
 
    const S s(7);
//  s.*pmi = 10; // error: cannot modify through mutable
    std::cout << s.*pmi << '\n';
 
    D d(7); // base pointers work with derived object
    D* pd = &d;
    std::cout << (d.*pf)(7) << ' '
              << (pd->*pf)(8) << '\n';
}

Output:

7
14 15

Standard library

Subscript operator is overloaded by many standard container classes:

accesses specific bit
(public member function of std::bitset<N>)
provides indexed access to the managed array
(public member function of std::unique_ptr<T,Deleter>)
accesses the specified character
(public member function of std::basic_string<CharT,Traits,Allocator>)
access specified element
(public member function of std::array<T,N>)
access specified element
(public member function of std::deque<T,Allocator>)
access specified element
(public member function of std::vector<T,Allocator>)
access or insert specified element
(public member function of std::map<Key,T,Compare,Allocator>)
access or insert specified element
(public member function of std::unordered_map<Key,T,Hash,KeyEqual,Allocator>)
accesses an element by index
(public member function of std::reverse_iterator<Iter>)
accesses an element by index
(public member function of std::move_iterator<Iter>)
get/set valarray element, slice, or mask
(public member function of std::valarray<T>)
returns specified sub-match
(public member function of std::match_results<BidirIt,Alloc>)

The indirection and member operators are overloaded by many iterators and smart pointer classes:

dereferences pointer to the managed object
(public member function of std::unique_ptr<T,Deleter>)
dereferences the stored pointer
(public member function of std::shared_ptr<T>)
accesses the managed object
(public member function of std::auto_ptr<T>)
dereferences the iterator
(public member function of std::raw_storage_iterator<OutputIt,T>)
dereferences the decremented underlying iterator
(public member function of std::reverse_iterator<Iter>)
no-op
(public member function of std::back_insert_iterator<Container>)
no-op
(public member function of std::front_insert_iterator<Container>)
no-op
(public member function of std::insert_iterator<Container>)
(C++11)(C++11)(deprecated in C++20)
accesses the pointed-to element
(public member function of std::move_iterator<Iter>)
returns the current element
(public member function of std::istream_iterator<T,CharT,Traits,Distance>)
no-op
(public member function of std::ostream_iterator<T,CharT,Traits>)
obtains a copy of the current character
(public member function of std::istreambuf_iterator<CharT,Traits>)
no-op
(public member function of std::ostreambuf_iterator<CharT,Traits>)
accesses the current match
(public member function of std::regex_iterator<BidirIt,CharT,Traits>)
accesses current submatch
(public member function of std::regex_token_iterator<BidirIt,CharT,Traits>)

No standard library classes overload operator&. The best known example of overloaded operator& is the Microsoft COM class CComPtr, although it can also appear in EDSLs such as boost.spirit.

No standard library classes overload operator->*. It was suggested that it could be part of smart pointer interface, and in fact is used in that capacity by actors in boost.phoenix, but is more common in EDSLs such as cpp.react.

Notes

Feature-test macro Value Std Comment
__cpp_multidimensional_subscript 202110L (C++23) Multidimensional subscript operator

Defect reports

The following behavior-changing defect reports were applied retroactively to previously published C++ standards.

DR Applied to Behavior as published Correct behavior
CWG 1213 C++11 subscripting an array rvalue resulted in lvalue reclassified as xvalue
CWG 1458 C++98 applying & to an lvalue of incomplete class type which
declares operator& resulted in undefined behavior
it is unspecified
which & is used
CWG 1800 C++98 when applying & to a non-static data member of a
member anonymous union, it was unclear whether
the anonymous union take a part in the result type
the anonymous union
is not included in
the result type
CWG 2614 C++98 the result of E1.E2 was unclear if E2 is a reference member or enumerator made clear

See also

Operator precedence

Operator overloading

Common operators
assignment increment
decrement
arithmetic logical comparison member
access
other

a = b
a += b
a -= b
a *= b
a /= b
a %= b
a &= b
a |= b
a ^= b
a <<= b
a >>= b

++a
--a
a++
a--

+a
-a
a + b
a - b
a * b
a / b
a % b
~a
a & b
a | b
a ^ b
a << b
a >> b

!a
a && b
a || b

a == b
a != b
a < b
a > b
a <= b
a >= b
a <=> b

a[...]
*a
&a
a->b
a.b
a->*b
a.*b

function call
a(...)
comma
a, b
conditional
a ? b : c
Special operators

static_cast converts one type to another related type
dynamic_cast converts within inheritance hierarchies
const_cast adds or removes cv-qualifiers
reinterpret_cast converts type to unrelated type
C-style cast converts one type to another by a mix of static_cast, const_cast, and reinterpret_cast
new creates objects with dynamic storage duration
delete destructs objects previously created by the new expression and releases obtained memory area
sizeof queries the size of a type
sizeof... queries the size of a parameter pack (since C++11)
typeid queries the type information of a type
noexcept checks if an expression can throw an exception (since C++11)
alignof queries alignment requirements of a type (since C++11)