cpp/language/implicit conversion: Difference between revisions

Implicit conversions are performed whenever an expression of some type T1 is used in context that does not accept that type, but accepts some other type T2; in particular:

• when the expression is used as the argument when calling a function that is declared with T2 as parameter;
• when the expression is used as an operand with an operator that expects T2;
• when initializing a new object of type T2, including return statement in a function returning T2;
• when the expression is used in a switch statement (T2 is integral type);
• when the expression is used in an if statement or a loop (T2 is bool).

The program is well-formed (compiles) only if there exists one unambiguous implicit conversion sequence from T1 to T2.

If there are multiple overloads of the function or operator being called, after the implicit conversion sequence is built from T1 to each available T2, overload resolution rules decide which overload is compiled.

Note: in arithmetic expressions, the destination type for the implicit conversions on the operands to binary operators is determined by a separate set of rules: usual arithmetic conversions.

Order of the conversions

Implicit conversion sequence consists of the following, in this order:

When considering the argument to a constructor or to a user-defined conversion function, only one standard conversion sequence is allowed (otherwise user-defined conversions could be effectively chained). When converting from one non-class type to another non-class type, only a standard conversion sequence is allowed.

A standard conversion sequence consists of the following, in this order:

A user-defined conversion consists of zero or one non-explicit single-argument converting constructor or non-explicit conversion function call.

An expression e is said to be implicitly convertible to T2 if and only if T2 can be copy-initialized from e, that is the declaration T2 t = e; is well-formed (can be compiled), for some invented temporary t. Note that this is different from direct initialization (T2 t(e)), where explicit constructors and conversion functions would additionally be considered.

Contextual conversions

In the following contexts, a context-specific type T is expected, and the expression e of class type E is only allowed if

Such expression e is said to be contextually implicitly converted to the specified type T. Note that explicit conversion functions are not considered, even though they are considered in contextual conversions to bool.(since C++11)

• the argument of the delete-expression (T is any object pointer type);
• integral constant expression, where a literal class is used (T is any integral or unscoped enumeration type, the selected user-defined conversion function must be constexpr);
• the controlling expression of the switch statement (T is any integral or enumeration type).

#include <cassert>

template<typename T>
class zero_init
{
    T val;
public:
    zero_init() : val(static_cast<T>(0)) {}
    zero_init(T val) : val(val) {}
    operator T&() { return val; }
    operator T() const { return val; }
};

int main()
{
    zero_init<int> i;
    assert(i == 0);
    
    i = 7;
    assert(i == 7);
    
    switch (i) {}     // error until C++14 (more than one conversion function)
                      // OK since C++14 (both functions convert to the same type int)
    switch (i + 0) {} // always okay (implicit conversion)
}

Value transformations

Value transformations are conversions that change the value category of an expression. They take place whenever an expression appears as an operand of an operator that expects an expression of a different value category:

• Whenever a glvalue appears as an operand of an operator that requires a prvalue for that operand, the lvalue-to-rvalue, array-to-pointer, or function-to-pointer standard conversions are applied to convert the expression to a prvalue.

Lvalue-to-rvalue conversion

An lvalue(until C++11)A glvalue(since C++11) of any non-function, non-array type T can be implicitly converted to an rvalue(until C++11)a prvalue(since C++11):

• If T is not a class type, the type of the rvalue(until C++11)prvalue(since C++11) is the cv-unqualified version of T.
• Otherwise, the type of the rvalue(until C++11)prvalue(since C++11) is T.

If an lvalue-to-rvalue conversion from an incomplete type is required by a program, that program is ill-formed.

Given the object to which the lvalue(until C++11)glvalue(since C++11) refers as obj:

This conversion models the act of reading a value from a memory location into a CPU register.

Array-to-pointer conversion

An lvalue or rvalue of type “array of N T” or “array of unknown bound of T” can be implicitly converted to a prvalue of type “pointer to T”. If the array is a prvalue, temporary materialization occurs.(since C++17) The resulting pointer refers to the first element of the array (see Array-to-pointer decay for details).

Function-to-pointer conversion

An lvalue of function type can be implicitly converted to a prvalue pointer to that function. This does not apply to non-static member functions because lvalues that refer to non-static member functions do not exist.

struct S { int m; };
int i = S().m; // member access expects glvalue as of C++17;
               // S() prvalue is converted to xvalue

Integral promotion

prvalues of small integral types (such as char) and unscoped enumeration types may be converted to prvalues of larger integral types (such as int). In particular, arithmetic operators do not accept types smaller than int as arguments, and integral promotions are automatically applied after lvalue-to-rvalue conversion, if applicable. This conversion always preserves the value.

The following implicit conversions in this section are classified as integral promotions.

Note that for a given source type, the destination type of integral promotion is unique, And all other conversions are not promotions. For example, overload resolution chooses char -> int (promotion) over char -> short (conversion).

Promotion from integral types

A prvalue of type bool can be converted to a prvalue of type int, with false becoming 0 and true becoming 1.

For a prvalue val of an integral type T except bool:

Promotion from enumeration types

A prvalue of an unscoped enumeration type whose underlying type is not fixed can be converted to a prvalue of the first type from the following list able to hold their entire value range:

• int
• unsigned int
• long
• unsigned long

Floating-point promotion

A prvalue of type float can be converted to a prvalue of type double. The value does not change.

This conversion is called floating-point promotion.

Numeric conversions

Unlike the promotions, numeric conversions may change the values, with potential loss of precision.

Integral conversions

A prvalue of an integer type or of an unscoped enumeration type can be converted to any other integer type. If the conversion is listed under integral promotions, it is a promotion and not a conversion.

• If the destination type is unsigned, the resulting value is the smallest unsigned value equal to the source value modulo 2n
  where n is the number of bits used to represent the destination type.

• That is, depending on whether the destination type is wider or narrower, signed integers are sign-extended^[1] or truncated and unsigned integers are zero-extended or truncated respectively.

• If the destination type is signed, the value does not change if the source integer can be represented in the destination type. Otherwise the result is implementation-defined(until C++20)the unique value of the destination type equal to the source value modulo 2n
  where n is the number of bits used to represent the destination type(since C++20) (note that this is different from signed integer arithmetic overflow, which is undefined).
• If the source type is bool, the value false is converted to zero and the value true is converted to the value one of the destination type (note that if the destination type is int, this is an integer promotion, not an integer conversion).
• If the destination type is bool, this is a boolean conversion (see below).

Floating-point conversions

If the conversion is listed under floating-point promotions, it is a promotion and not a conversion.

• If the source value can be represented exactly in the destination type, it does not change.
• If the source value is between two representable values of the destination type, the result is one of those two values (it is implementation-defined which one, although if IEEE arithmetic is supported, rounding defaults to nearest).
• Otherwise, the behavior is undefined.

Floating–integral conversions

A prvalue of floating-point type can be converted to a prvalue of any integer type. The fractional part is truncated, that is, the fractional part is discarded.

• If the truncated value cannot fit into the destination type, the behavior is undefined (even when the destination type is unsigned, modulo arithmetic does not apply).
• If the destination type is bool, this is a boolean conversion (see below).

A prvalue of integer or unscoped enumeration type can be converted to a prvalue of any floating-point type. The result is exact if possible.

• If the value can fit into the destination type but cannot be represented exactly, it is implementation defined whether the closest higher or the closest lower representable value will be selected, although if IEEE arithmetic is supported, rounding defaults to nearest.
• If the value cannot fit into the destination type, the behavior is undefined.
• If the source type is bool, the value false is converted to zero, and the value true is converted to one.

Pointer conversions

A null pointer constant can be converted to any pointer type, and the result is the null pointer value of that type. Such conversion (known as null pointer conversion) is allowed to convert to a cv-qualified type as a single conversion, that is, it is not considered a combination of numeric and qualifying conversions.

A prvalue pointer to any (optionally cv-qualified) object type T can be converted to a prvalue pointer to (identically cv-qualified) void. The resulting pointer represents the same location in memory as the original pointer value.

• If the original pointer is a null pointer value, the result is a null pointer value of the destination type.

A prvalue ptr of type “pointer to (possibly cv-qualified) Derived” can be converted to a prvalue of type “pointer to (possibly cv-qualified) Base”, where Base is a base class of Derived, and Derived is a complete class type. If the Base is inaccessible or ambiguous, the program is ill-formed.

• If ptr is a null pointer value, the result is also a null pointer value.
• Otherwise, if Base is a virtual base class of Derived and ptr does not point to an object whose type is similar to Derived and that is within its lifetime or within its period of construction or destruction, the behavior is undefined.
• Otherwise, the result is a pointer to the base class subobject of the derived class object.

Pointer-to-member conversions

A null pointer constant can be converted to any pointer-to-member type, and the result is the null member pointer value of that type. Such conversion (known as null member pointer conversion) is allowed to convert to a cv-qualified type as a single conversion, that is, it is not considered a combination of numeric and qualifying conversions.

A prvalue of type “pointer to member of Base of type (possibly cv-qualified) T” can be converted to a prvalue of type “pointer to member of Derived of type (identically cv-qualified) T”, where Base is a base class of Derived, and Derived is a complete class type. If Base is inaccessible, ambiguous, or virtual base of Derived or is a base of some intermediate virtual base of Derived, the program is ill-formed.

• If Derived does not contain the original member and is not a base class of the class containing the original member, the behavior is undefined.
• Otherwise, the resulting pointer can be dereferenced with a Derived object, and it will access the member within the Base base subobject of that Derived object.

Boolean conversions

A prvalue of integral, floating-point, unscoped enumeration, pointer, and pointer-to-member types can be converted to a prvalue of type bool.

The value zero (for integral, floating-point, and unscoped enumeration) and the null pointer and the null pointer-to-member values become false. All other values become true.

Qualification conversions

Generally speaking:

• A prvalue of type pointer to cv-qualified type T can be converted to a prvalue pointer to a more cv-qualified same type T (in other words, constness and volatility can be added).
• A prvalue of type pointer to member of cv-qualified type T in class X can be converted to a prvalue pointer to member of more cv-qualified type T in class X.

The formal definition of “qualification conversion” is given below.

Similar types

Informally, two types are similar if, ignoring top-level cv-qualification:

• they are the same type; or
• they are both pointers, and the pointed-to types are similar; or
• they are both pointers to member of the same class, and the types of the pointed-to members are similar; or
• they are both arrays and the array element types are similar.

For example:

• const int* const * and int** are similar;
• int (*)(int*) and int (*)(const int*) are not similar;
• const int (*)(int*) and int (*)(int*) are not similar;
• int (*)(int* const) and int (*)(int*) are similar (they are the same type);
• std::pair<int, int> and std::pair<const int, int> are not similar.

Formally, type similarity is defined in terms of qualification-decomposition.

A qualification-decomposition of a type T is a sequence of components cv_i and P_i such that T is “cv_0 P_0 cv_1 P_1 ... cv_n−1 P_n−1 cv_n U” for non-negative n, where

• each cv_i is a set of const and volatile, and
• each P_i is

• “pointer to”,
• “pointer to member of class C_i of type”,
• “array of N_i”, or
• “array of unknown bound of”.

If P_i designates an array, the cv-qualifiers cv_i+1 on the element type are also taken as the cv-qualifiers cv_i of the array.

// T is “pointer to pointer to const int”, it has 3 qualification-decompositions:
// n = 0 -> cv_0 is empty, U is “pointer to pointer to const int”
// n = 1 -> cv_0 is empty, P_0 is “pointer to”,
//          cv_1 is empty, U is “pointer to const int”
// n = 2 -> cv_0 is empty, P_0 is “pointer to”,
//          cv_1 is empty, P_1 is “pointer to”,
//          cv_2 is “const", U is “int”
using T = const int**;

// substitute any of the following type to U gives one of the decompositions:
// U = U0 -> the decomposition with n = 0: U0
// U = U1 -> the decomposition with n = 1: pointer to [U1]
// U = U2 -> the decomposition with n = 2: pointer to [pointer to [const U2]]
using U2 = int;
using U1 = const U2*;
using U0 = U1*;

Two types T1 and T2 are similar if there exists a qualification-decomposition for each of them, where all following conditions are satisfied for the two qualification-decompositions:

• They have the same n.
• The types denoted by U are the same.
• The corresponding P_i components are the same or one is “array of N_i” and the other is “array of unknown bound of”(since C++20) for all i.

// the qualification-decomposition with n = 2:
// pointer to [volatile pointer to [const int]]
using T1 = const int* volatile *;

// the qualification-decomposition with n = 2:
// const pointer to [pointer to [int]]
using T2 = int** const;

// For the two qualification-decompositions above
// although cv_0, cv_1 and cv_2 are all different,
// they have the same n, U, P_0 and P_1,
// therefore types T1 and T2 are similar.

Combining cv-qualifications

In the description below, the longest qualification-decomposition of type Tn is denoted as Dn, and its components are denoted as cvn_i and Pn_i.

// longest qualification-decomposition of T1 (n = 2):
// pointer to [pointer to [char]]
using T1 = char**;

// longest qualification-decomposition of T2 (n = 2):
// pointer to [pointer to [const char]]
using T2 = const char**;

// Determining the cv3_i and T_i components of D3 (n = 2):
// cv3_1 = empty (union of empty cv1_1 and empty cv2_1)
// cv3_2 = “const” (union of empty cv1_2 and “const” cv2_2)
// P3_0 = “pointer to” (no array of unknown bound, use P1_0)
// P3_1 = “pointer to” (no array of unknown bound, use P1_1)
// All components except cv_2 are the same, cv3_2 is different from cv1_2,
// therefore add “const” to cv3_k for each k in [1, 2): cv3_1 becomes “const”.
// T3 is “pointer to const pointer to const char”, i.e., const char* const *.
using T3 = /* the qualification-combined type of T1 and T2 */;

int main()
{
    const char c = 'c';
    char* pc;
    T1 ppc = &pc;
    T2 pcc = ppc; // Error: T3 is not the same as cv-unqualified T2,
                  //        no implicit conversion.
    
    *pcc = &c;
    *pc = 'C';    // If the erroneous assignment above is allowed,
                  // the const object “c” may be modified.
}

Note that in the C programming language, const/volatile can be added to the first level only:

char** p = 0;
char * const* p1 = p;       // OK in C and C++
const char* const * p2 = p; // error in C, OK in C++

void (*p)();
void (**pp)() noexcept = &p; // error: cannot convert to pointer to noexcept function

struct S
{
    typedef void (*p)();
    operator p();
};
void (*q)() noexcept = S(); // error: cannot convert to pointer to noexcept function

The safe bool problem

Until C++11, designing a class that should be usable in boolean contexts (e.g. if (obj) { ... }) presented a problem: given a user-defined conversion function, such as T::operator bool() const;, the implicit conversion sequence allowed one additional standard conversion sequence after that function call, which means the resultant bool could be converted to int, allowing such code as obj << 1; or int i = obj;.

One early solution for this can be seen in std::basic_ios, which initially defines operator void*, so that the code such as if (std::cin) {...} compiles because void* is convertible to bool, but int n = std::cout; does not compile because void* is not convertible to int. This still allows nonsense code such as delete std::cout; to compile.

Many pre-C++11 third party libraries were designed with a more elaborate solution, known as the Safe Bool idiom. std::basic_ios also allowed this idiom via LWG issue 468, and operator void* was replaced (see notes).

Since C++11, explicit bool conversion can also be used to resolve the safe bool problem.

Defect reports

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

See also

• const_cast
• static_cast
• dynamic_cast
• reinterpret_cast
• explicit cast
• user-defined conversion