Boost.Variant2: A never valueless variant type

Overview

Description

This library implements a type-safe discriminated/tagged union type, variant<T…>, that is API-compatible with the C++17 Standard’s std::variant<T…>.

A variant<T1, T2, …, Tn> variable can hold a value of any of the types T1, T2, …, Tn. For example, variant<int64_t, double, std::string> can hold an int64_t value, a double value, or a string value.

Such a type is sometimes called a "tagged union", because it’s roughly equivalent to

struct V
{
    enum tag { tag_int64_t, tag_double, tag_string };

    tag tag_;

    union
    {
        int64_t     i_;
        double      d_;
        std::string s_;
    };
};

Usage Examples

Variants can be used to represent dynamically-typed values. A configuration file of the form

server.host=test.example.com
server.port=9174
cache.max_load=0.7

can be represented as std::map<std::string, variant<int64_t, double, std::string>>.

Variants can also represent polymorphism. To take a classic example, a polymorphic collection of shapes:

#define _USE_MATH_DEFINES
#include <iostream>
#include <vector>
#include <memory>
#include <cmath>

class Shape
{
public:

    virtual ~Shape() = default;
    virtual double area() const = 0;
};

class Rectangle: public Shape
{
private:

    double width_, height_;

public:

    Rectangle( double width, double height ):
        width_( width ), height_( height ) {}

    virtual double area() const { return width_ * height_; }
};

class Circle: public Shape
{
private:

    double radius_;

public:

    explicit Circle( double radius ): radius_( radius ) {}
    virtual double area() const { return M_PI * radius_ * radius_; }
};

double total_area( std::vector<std::unique_ptr<Shape>> const & v )
{
    double s = 0.0;

    for( auto const& p: v )
    {
        s += p->area();
    }

    return s;
}

int main()
{
    std::vector<std::unique_ptr<Shape>> v;

    v.push_back( std::unique_ptr<Shape>( new Circle( 1.0 ) ) );
    v.push_back( std::unique_ptr<Shape>( new Rectangle( 2.0, 3.0 ) ) );

    std::cout << "Total area: " << total_area( v ) << std::endl;
}

can instead be represented as a collection of variant<Rectangle, Circle> values. This requires the possible Shape types be known in advance, as is often the case. In return, we no longer need virtual functions, or to allocate the values on the heap with new Rectangle and new Circle:

#define _USE_MATH_DEFINES
#include <iostream>
#include <vector>
#include <cmath>

#include <boost/variant2/variant.hpp>
using namespace boost::variant2;

struct Rectangle
{
    double width_, height_;
    double area() const { return width_ * height_; }
};

struct Circle
{
    double radius_;
    double area() const { return M_PI * radius_ * radius_; }
};

double total_area( std::vector<variant<Rectangle, Circle>> const & v )
{
    double s = 0.0;

    for( auto const& x: v )
    {
        s += visit( []( auto const& y ){ return y.area(); }, x );
    }

    return s;
}

int main()
{
    std::vector<variant<Rectangle, Circle>> v;

    v.push_back( Circle{ 1.0 } );
    v.push_back( Rectangle{ 2.0, 3.0 } );

    std::cout << "Total area: " << total_area( v ) << std::endl;
}

Construction and Assignment

If we look at the

    v.push_back( Circle{ 1.0 } );

line, we can deduce that variant<Rectangle, Circle> can be (implicitly) constructed from Circle (and Rectangle), and indeed it can. It can also be assigned a Circle or a Rectangle:

variant<Rectangle, Circle> v = Circle{ 1.0 }; // v holds Circle
v = Rectangle{ 2.0, 3.0 };                    // v now holds Rectangle

If we try to construct variant<int, float> from something that is neither int nor float, say, (short)1, the behavior is "as if" the variant has declared two constructors,

variant::variant(int x);
variant::variant(float x);

and the standard overload resolution rules are used to pick the one that will be used. So variant<int, float>((short)1) will hold an int.

Inspecting the Value

Putting values into a variant is easy, but taking them out is necessarily a bit more convoluted. It’s not possible for variant<int, float> to define a member function get() const, because such a function will need its return type fixed at compile time, and whether the correct return type is int or float will only become known at run time.

There are a few ways around that. First, there is the accessor member function

std::size_t variant::index() const noexcept;

that returns the zero-based index of the current type. For variant<int, float>, it will return 0 for int and 1 for float.

Once we have the index, we can use the free function get<N> to obtain the value. Since we’re passing the type index to get, it knows what to return. get<0>(v) will return int, and get<1>(v) will return float:

void f( variant<int, float> const& v )
{
    switch( v.index() )
    {
    case 0:

        // use get<0>(v)
        break;

    case 1:

        // use get<1>(v)
        break;

    default:

        assert(false); // never happens
    }
}

If we call get<0>(v), and v.index() is not currently 0, an exception (of type bad_variant_access) will be thrown.

An alternative approach is to use get<int>(v) or get<float>(v). This works similarly.

Another alternative that avoids the possibility of bad_variant_access is to use get_if. Instead of a reference to the contained value, it returns a pointer to it, returning nullptr to indicate type mismatch. get_if takes a pointer to the variant, so in our example we’ll use something along the following lines:

void f( variant<int, float> const& v )
{
    if( int const * p = get_if<int>(&v) )
    {
        // use *p
    }
    else if( float const * p = get_if<float>(&v) )
    {
        // use *p
    }
    else
    {
        assert(false); // never happens
    }
}

Visitation

Last but not least, there’s visit. visit(f, v) calls the a function object f with the value contained in the variant v and returns the result. When v is variant<int, float>, it will call f with either an int or a float. The function object must be prepared to accept both.

In practice, this can be achieved by having the function take a type that can be passed either int or float, such as double:

double f( double x ) { return x; }

double g( variant<int, float> const& v )
{
    return visit( f, v );
}

By using a function object with an overloaded operator():

struct F
{
    void operator()(int x) const { /* use x */ }
    void operator()(float x) const { /* use x */ }
};

void g( variant<int, float> const& v )
{
    visit( F(), v );
}

Or by using a polymorphic lambda, as we did in our Circle/Rectangle example:

void g( variant<int, float> const& v )
{
    visit( [&]( auto const& x ){ std::cout << x << std::endl; }, v );
}

visit can also take more than one variant. visit(f, v1, v2) calls f(x1, x2), where x1 is the value contained in v1 and x2 is the value in v2.

Default Construction

The default constructor of variant value-initializes the first type in the list. variant<int, float>{} holds 0 (of type int), and variant<float, int>{} holds 0.0f.

This is usually the desired behavior. However, in cases such as variant<std::mutex, std::recursive_mutex>, one might legitimately wish to avoid constructing a std::mutex by default. A provided type, monostate, can be used as the first type in those scenarios. variant<monostate, std::mutex, std::recursive_mutex> will default-construct a monostate, which is basically a no-op, as monostate is effectively an empty struct.

Revision History

Changes in 1.73.0

Added support for std::hash, boost::hash.
variant<T…> is now trivial when all types in T… are trivial. This improves performance by enabling it to be passed to, and returned from, functions in registers.

Changes in 1.71.0

After the Boost formal review, the implementation has been changed to provide the strong exception safety guarantee, instead of basic. expected has been removed.

Design

Features

This variant implementation has two distinguishing features:

It’s never "valueless", that is, variant<T1, T2, …, Tn> has an invariant that it always contains a valid value of one of the types T1, T2, …, Tn.
It provides the strong exception safety guarantee on assignment and emplace.

This is achieved with the use of double storage, unless all of the contained types have a non-throwing move constructor.

Rationale

Never Valueless

It makes intuitive sense that variant<X, Y, Z> can hold only values of type X, type Y, or type Z, and nothing else.

If we think of variant as an extension of union, since a union has a state called "no active member", an argument can be made that a variant<X, Y, Z> should also have such an additional state, holding none of X, Y, Z.

This however makes variant less convenient in practice and less useful as a building block. If we really need a variable that only holds X, Y, or Z, the additional empty state creates complications that need to be worked around. And in the case where we do need this additional empty state, we can just use variant<empty, X, Y, Z>, with a suitable struct empty {};.

From a pure design perspective, the case for no additional empty state is solid. Implementation considerations, however, argue otherwise.

When we replace the current value of the variant (of, say, type X) with another (of type Y), since the new value needs to occupy the same storage as the old one, we need to destroy the old X first, then construct a new Y in its place. But since this is C++, the construction can fail with an exception. At this point the variant is in the "has no active member" state that we’ve agreed it cannot be in.

This is a legitimate problem, and it is this problem that makes having an empty/valueless state so appealing. We just leave the variant empty on exception and we’re done.

As explained, though, this is undesirable from a design perspective as it makes the component less useful and less elegant.

There are several ways around the issue. The most straightforward one is to just disallow types whose construction can throw. Since we can always create a temporary value first, then use the move constructor to initialize the one in the variant, it’s enough to require a nonthrowing move constructor, rather than all constructors to be nonthrowing.

Unfortunately, under at least one popular standard library implementation, node based containers such as std::list and std::map have a potentially throwing move constructor. Disallowing variant<X, std::map<Y, Z>> is hardly practical, so the exceptional case cannot be avoided.

On exception, we could also construct some other value, leaving the variant valid; but in the general case, that construction can also throw. If one of the types has a nonthrowing default constructor, we can use it; but if not, we can’t.

The approach Boost.Variant takes here is to allocate a temporary copy of the value on the heap. On exception, a pointer to that temporary copy can be stored into the variant. Pointer operations don’t throw.

Another option is to use double buffering. If our variant occupies twice the storage, we can construct the new value in the unused half, then, once the construction succeeds, destroy the old value in the other half.

When std::variant was standardized, none of those approaches was deemed palatable, as all of them either introduce overhead or are too restrictive with respect to the types a variant can contain. So as a compromise, std::variant took a way that can (noncharitably) be described as "having your cake and eating it too."

Since the described exceptional situation is relatively rare, std::variant has a special case, called "valueless", into which it goes on exception, but the interface acknowledges its existence as little as possible, allowing users to pretend that it doesn’t exist.

This is, arguably, not that bad from a practical point of view, but it leaves many of us wanting. Rare states that "never" occur are undertested and when that "never" actually happens, it’s usually in the most inconvenient of times.

This implementation does not follow std::variant; it statically guarantees that variant is never in a valueless state. The function valueless_by_exception is provided for compatibility, but it always returns false.

Instead, if the contained types are such that it’s not possible to avoid an exceptional situation when changing the contained value, double storage is used.

Strong Exception Safety

The initial submission only provided the basic exception safety guarantee. If an attempt to change the contained value (via assignment or emplace) failed with an exception, and a type with a nonthrowing default constructor existed among the alternatives, a value of that type was created into the variant. The upside of this decision was that double storage was needed less frequently.

The reviewers were fairly united in hating it. Constructing a random type was deemed too unpredictable and not complying with the spirit of the basic guarantee. The default constructor of the chosen type, even if nonthrowing, may still have undesirable side effects. Or, if not that, a value of that type may have special significance for the surrounding code. Therefore, some argued, the variant should either remain with its old value, or transition into the new one, without synthesizing other states.

At the other side of the spectrum, there were those who considered double storage unacceptable. But they considered it unacceptable in principle, regardless of the frequency with which it was used.

As a result, providing the strong exception safety guarantee on assignment and emplace was declared an acceptance condition.

In retrospect, this was the right decision. The reason the strong guarantee is generally not provided is because it doesn’t compose. When X and Y provide the basic guarantee on assignment, so does struct { X x; Y y; };. Similarly, when X and Y have nonthrowing assignments, so does the struct. But this doesn’t hold for the strong guarantee.

The usual practice is to provide the basic guarantee on assignment and let the user synthesize a "strong" assignment out of either a nonthrowing swap or a nonthrowing move assignment. That is, given x1 and x2 of type X, instead of the "basic" x1 = x2;, use either X(x2).swap(x1); or x1 = X(x2);.

Nearly all types provide a nonthrowing swap or a nonthrowing move assignment, so this works well. Nearly all, except variant, which in the general case has neither a nonthrowing swap nor a nonthrowing move assignment. If variant does not provide the strong guarantee itself, it’s impossible for the user to synthesize it.

So it should, and so it does.

Differences with std::variant

The main differences between this implementation and std::variant are:

No valueless-by-exception state: valueless_by_exception() always returns false.
Strong exception safety guarantee on assignment and emplace.
emplace first constructs the new value and then destroys the old one; in the single storage case, this translates to constructing a temporary and then moving it into place.
A converting constructor from, e.g. variant<int, float> to variant<float, double, int> is provided as an extension.
The reverse operation, going from variant<float, double, int> to variant<int, float> is provided as the member function subset<U…>. (This operation can throw if the current state of the variant cannot be represented.)
The C++20 additions and changes to std::variant have not yet been implemented.

Differences with Boost.Variant

This library is API compatible with std::variant. As such, its interface is different from Boost.Variant’s. For example, visitation is performed via visit instead of apply_visitor.

Recursive variants are not supported.

Double storage is used instead of temporary heap backup. This variant is always "stack-based", it never allocates, and never throws bad_alloc on its own.

Implementation

Dependencies

This implementation only depends on Boost.Config and Boost.Mp11.

Supported Compilers

GCC 4.8 or later with -std=c++11 or above
Clang 3.5 or later with -std=c++11 or above
Visual Studio 2015, 2017, 2019

Tested on Travis and Appveyor.

Reference

<boost/variant2/variant.hpp>

Synopsis

namespace boost {
namespace variant2 {

// in_place_type

template<class T> struct in_place_type_t {};
template<class T> constexpr in_place_type_t<T> in_place_type{};

// in_place_index

template<std::size_t I> struct in_place_index_t {};
template<std::size_t I> constexpr in_place_index_t<I> in_place_index{};

// variant

template<class... T> class variant;

// variant_size

template<class T> struct variant_size {};

template<class T> struct variant_size<T const>: variant_size<T> {};
template<class T> struct variant_size<T volatile>: variant_size<T> {};
template<class T> struct variant_size<T const volatile>: variant_size<T> {};

template<class T> struct variant_size<T&>: variant_size<T> {}; // extension
template<class T> struct variant_size<T&&>: variant_size<T> {}; // extension

template<class T>
  inline constexpr size_t variant_size_v = variant_size<T>::value;

template<class... T>
  struct variant_size<variant<T...>>:
    std::integral_constant<std::size_t, sizeof...(T)> {};

// variant_alternative

template<size_t I, class T> struct variant_alternative {};

template<size_t I, class T> struct variant_alternative<I, T const>;
template<size_t I, class T> struct variant_alternative<I, T volatile>;
template<size_t I, class T> struct variant_alternative<I, T const volatile>;

template<size_t I, class T> struct variant_alternative<I, T&>; // extension
template<size_t I, class T> struct variant_alternative<I, T&&>; // extension

template<size_t I, class T>
  using variant_alternative_t = typename variant_alternative<I, T>::type;

template<size_t I, class... T>
  struct variant_alternative<I, variant<T...>>;

// variant_npos

constexpr std::size_t variant_npos = -1;

// holds_alternative

template<class U, class... T>
  constexpr bool holds_alternative(const variant<T...>& v) noexcept;

// get

template<size_t I, class... T>
  constexpr variant_alternative_t<I, variant<T...>>&
    get(variant<T...>& v);
template<size_t I, class... T>
  constexpr variant_alternative_t<I, variant<T...>>&&
    get(variant<T...>&& v);
template<size_t I, class... T>
  constexpr const variant_alternative_t<I, variant<T...>>&
    get(const variant<T...>& v);
template<size_t I, class... T>
  constexpr const variant_alternative_t<I, variant<T...>>&&
    get(const variant<T...>&& v);

template<class U, class... T>
  constexpr U& get(variant<T...>& v);
template<class U, class... T>
  constexpr U&& get(variant<T...>&& v);
template<class U, class... T>
  constexpr const U& get(const variant<T...>& v);
template<class U, class... T>
  constexpr const U&& get(const variant<T...>&& v);

// get_if

template<size_t I, class... T>
  constexpr add_pointer_t<variant_alternative_t<I, variant<T...>>>
    get_if(variant<T...>* v) noexcept;
template<size_t I, class... T>
  constexpr add_pointer_t<const variant_alternative_t<I, variant<T...>>>
    get_if(const variant<T...>* v) noexcept;

template<class U, class... T>
  constexpr add_pointer_t<U>
    get_if(variant<T...>* v) noexcept;
template<class U, class... T>
  constexpr add_pointer_t<const U>
    get_if(const variant<T...>* v) noexcept;

// relational operators

template<class... T>
  constexpr bool operator==(const variant<T...>& v, const variant<T...>& w);
template<class... T>
  constexpr bool operator!=(const variant<T...>& v, const variant<T...>& w);
template<class... T>
  constexpr bool operator<(const variant<T...>& v, const variant<T...>& w);
template<class... T>
  constexpr bool operator>(const variant<T...>& v, const variant<T...>& w);
template<class... T>
  constexpr bool operator<=(const variant<T...>& v, const variant<T...>& w);
template<class... T>
  constexpr bool operator>=(const variant<T...>& v, const variant<T...>& w);

// visit

template<class F, class... V>
  constexpr /*see below*/ visit(F&& f, V&&... v);

// monostate

struct monostate {};

constexpr bool operator==(monostate, monostate) noexcept { return true; }
constexpr bool operator!=(monostate, monostate) noexcept { return false; }
constexpr bool operator<(monostate, monostate) noexcept { return false; }
constexpr bool operator>(monostate, monostate) noexcept { return false; }
constexpr bool operator<=(monostate, monostate) noexcept { return true; }
constexpr bool operator>=(monostate, monostate) noexcept { return true; }

// swap

template<class... T>
  void swap(variant<T...>& v, variant<T...>& w) noexcept( /*see below*/ );

// bad_variant_access

class bad_variant_access;

} // namespace variant2
} // namespace boost

variant

namespace boost {
namespace variant2 {

template<class... T> class variant
{
public:

  // constructors

  constexpr variant() noexcept( /*see below*/ );

  constexpr variant( variant const & r ) noexcept( /*see below*/ );
  constexpr variant( variant&& r ) noexcept( /*see below*/ );

  template<class U>
    constexpr variant( U&& u ) noexcept( /*see below*/ );

  template<class U, class... A>
    constexpr explicit variant( in_place_type_t<U>, A&&... a );
  template<class U, class V, class... A>
    constexpr explicit variant( in_place_type_t<U>,
      std::initializer_list<V> il, A&&... a );

  template<size_t I, class... A>
    constexpr explicit variant( in_place_index_t<I>, A&&... a );
  template<size_t I, class V, class... A>
    constexpr explicit variant( in_place_index_t<I>,
      std::initializer_list<V> il, A&&... a );

  // destructor

  ~variant();

  // assignment

  constexpr variant& operator=( variant const & r ) noexcept( /*see below*/ );
  constexpr variant& operator=( variant&& r ) noexcept( /*see below*/ );

  template<class U> constexpr variant& operator=( U&& u ) noexcept( /*see below*/ );

  // modifiers

  template<class U, class... A>
    constexpr U& emplace( A&&... a );
  template<class U, class V, class... A>
    constexpr U& emplace( std::initializer_list<V> il, A&&... a );

  template<size_t I, class... A>
    constexpr variant_alternative_t<I, variant<T...>>&
      emplace( A&&... a );
  template<size_t I, class V, class... A>
    constexpr variant_alternative_t<I, variant<T...>>&
      emplace( std::initializer_list<V> il, A&&... a );

  // value status

  constexpr bool valueless_by_exception() const noexcept;
  constexpr size_t index() const noexcept;

  // swap

  void swap( variant& r ) noexcept( /*see below*/ );

  // converting constructors (extension)

  template<class... U> variant( variant<U...> const& r )
    noexcept( /*see below*/ );

  template<class... U> variant( variant<U...>&& r )
    noexcept( /*see below*/ );

  // subset (extension)

  template<class... U> constexpr variant<U...> subset() & ;
  template<class... U> constexpr variant<U...> subset() && ;
  template<class... U> constexpr variant<U...> subset() const& ;
  template<class... U> constexpr variant<U...> subset() const&& ;
};

} // namespace variant2
} // namespace boost

In the descriptions that follow, let i be in the range [0, sizeof…(T)), and Ti be the i-th type in T….

Constructors

constexpr variant() noexcept( std::is_nothrow_default_constructible_v<T0> );

Effects:

Constructs a variant holding a value-initialized value of type T0.

Ensures:

index() == 0.

Throws:

Any exception thrown by the value-initialization of T0.

Remarks:

This function does not participate in overload resolution unless std::is_default_constructible_v<T0> is true.

constexpr variant( variant const & w )
  noexcept( mp_all<std::is_nothrow_copy_constructible<T>...>::value );

Effects:

Initializes the variant to hold the same alternative and value as w.

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless std::is_copy_constructible_v<Ti> is true for all i.

constexpr variant( variant&& w )
  noexcept( mp_all<std::is_nothrow_move_constructible<T>...>::value );

Effects:

Initializes the variant to hold the same alternative and value as w.

Throws:

Any exception thrown by the move-initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless std::is_move_constructible_v<Ti> is true for all i.

template<class U> constexpr variant( U&& u ) noexcept(/*see below*/);

Let Tj be a type that is determined as follows: build an imaginary function FUN(Ti) for each alternative type Ti. The overload FUN(Tj) selected by overload resolution for the expression FUN(std::forward(u)) defines the alternative Tj which is the type of the contained value after construction.
Effects:

Initializes *this to hold the alternative type Tj and initializes the contained value from std::forward(u).

Ensures:

holds_alternative<Tj>(*this).

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:
The expression inside noexcept is equivalent to std::is_nothrow_constructible_v<Tj, U>. This function does not participate in overload resolution unless
sizeof…(T) is nonzero,

std::is_same_v<std::remove_cvref_t, variant> is false,

std::remove_cvref_t is neither a specialization of in_place_type_t nor a specialization of in_place_index_t,

std::is_constructible_v<Tj, U> is true, and

the expression FUN(std::forward(u)) is well-formed.

template<class U, class... A>
  constexpr explicit variant( in_place_type_t<U>, A&&... a );

Effects:

Initializes the contained value of type U with the arguments std::forward<A>(a)….

Ensures:

holds_alternative(*this).

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless there is exactly one occurrence of U in T… and std::is_constructible_v<U, A…> is true.

template<class U, class V, class... A>
  constexpr explicit variant( in_place_type_t<U>, std::initializer_list<V> il,
    A&&... a );

Effects:

Initializes the contained value of type U with the arguments il, std::forward<A>(a)….

Ensures:

holds_alternative(*this).

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless there is exactly one occurrence of U in T… and std::is_constructible_v<U, initializer_list<V>&, A…> is true.

template<size_t I, class... A>
  constexpr explicit variant( in_place_index_t<I>, A&&... a );

Effects:

Initializes the contained value of type TI with the arguments std::forward<A>(a)….

Ensures:

index() == I.

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless I < sizeof…(T) and std::is_constructible_v<TI, A…> is true.

template<size_t I, class V, class... A>
  constexpr explicit variant( in_place_index_t<I>, std::initializer_list<V> il,
    A&&... a );

Effects:

Initializes the contained value of type TI with the arguments il, std::forward<A>(a)….

Ensures:

index() == I.

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless I < sizeof…(T) and std::is_constructible_v<TI, initializer_list<V>&, A…> is true.

Destructor

~variant();

Effects:

Destroys the currently contained value.

Assignment

constexpr variant& operator=( const variant& r )
  noexcept( mp_all<std::is_nothrow_copy_constructible<T>...>::value );

Let j be r.index().

Effects:

emplace<j>(get<j>(r)).

Returns:

*this.

Ensures:

index() == r.index().

Remarks:

This operator does not participate in overload resolution unless std::is_copy_constructible_v<Ti> && std::is_copy_assignable_v<Ti> is true for all i.

constexpr variant& operator=( variant&& r )
  noexcept( mp_all<std::is_nothrow_move_constructible<T>...>::value );

Let j be r.index().

Effects:

emplace<j>(get<j>(std::move(r))).

Returns:

*this.

Ensures:

index() == r.index().

Remarks:

This operator does not participate in overload resolution unless std::is_move_constructible_v<Ti> && std::is_move_assignable_v<Ti> is true for all i.

template<class U> constexpr variant& operator=( U&& u )
  noexcept( /*see below*/ );

Let Tj be a type that is determined as follows: build an imaginary function FUN(Ti) for each alternative type Ti. The overload FUN(Tj) selected by overload resolution for the expression FUN(std::forward(u)) defines the alternative Tj which is the type of the contained value after construction.
Effects:

emplace<j>(std::forward(u)).

Returns:

*this.

Ensures:

index() == j.

Remarks:
The expression inside noexcept is std::is_nothrow_constructible_v<Tj, U&&>. This operator does not participate in overload resolution unless
std::is_same_v<std::remove_cvref_t<T>, variant> is false,

std::is_constructible_v<Tj, U&&> && std::is_assignable_v<Tj&, U&&> is true, and

the expression FUN(std::forward(u)) (with FUN being the above-mentioned set of imaginary functions) is well-formed.

Modifiers

template<class U, class... A>
  constexpr U& emplace( A&&... a );

Let I be the zero-based index of U in T….

Effects:

Equivalent to: return emplace(std::forward<A>(a)…);

Remarks:

This function shall not participate in overload resolution unless std::is_constructible_v<U, A&&…> is true and U occurs exactly once in T….

template<class U, class V, class... A>
  constexpr U& emplace( std::initializer_list<V> il, A&&... a );

Let I be the zero-based index of U in T….

Effects:

Equivalent to: return emplace(il, std::forward<A>(a)…);

Remarks:

This function shall not participate in overload resolution unless std::is_constructible_v<U, std::initializer_list<V>&, A&&…> is true and U occurs exactly once in T….

template<size_t I, class... A>
  constexpr variant_alternative_t<I, variant<T...>>&
    emplace( A&&... a );

Requires:

I < sizeof…(T).

Effects:

Destroys the currently contained value, then initializes a new contained value as if using the expression Ti(std::forward<A>(a)…).

Ensures:

index() == I.

Returns:

A reference to the new contained value.

Throws:

Nothing unless the initialization of the new contained value throws.

Exception Safety:

Strong. On exception, the contained value is unchanged.

Remarks:

This function shall not participate in overload resolution unless std::is_constructible_v<Ti, A&&…> is true.

template<size_t I, class V, class... A>
  constexpr variant_alternative_t<I, variant<T...>>&
    emplace( std::initializer_list<V> il, A&&... a );

Requires:

I < sizeof…(T).

Effects:

Destroys the currently contained value, then initializes a new contained value as if using the expression Ti(il, std::forward<A>(a)…).

Ensures:

index() == I.

Returns:

A reference to the new contained value.

Throws:

Nothing unless the initialization of the new contained value throws.

Exception Safety:

Strong. On exception, the contained value is unchanged.

Remarks:

This function shall not participate in overload resolution unless std::is_constructible_v<Ti, std::initializer_list<V>&, A&&…> is true.

Value Status

constexpr bool valueless_by_exception() const noexcept;

Returns:

false.

Note	This function is provided purely for compatibility with `std::variant`.

constexpr size_t index() const noexcept;

Returns:

The zero-based index of the active alternative.

Swap

void swap( variant& r ) noexcept( mp_all<std::is_nothrow_move_constructible<T>...,
  is_nothrow_swappable<T>...>::value );

Effects:
If index() == r.index(), calls swap(get(*this), get(r)), where I is index().

Otherwise, as if variant tmp(std::move(*this)); *this = std::move(r); r = std::move(tmp);

Converting Constructors (extension)

template<class... U> variant( variant<U...> const& r )
  noexcept( mp_all<std::is_nothrow_copy_constructible<U>...>::value );

Effects:

Initializes the contained value from the contained value of r.

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless all types in U… are in T… and std::is_copy_constructible_v<Ui>::value is true for all Ui.

template<class... U> variant( variant<U...>&& r )
  noexcept( mp_all<std::is_nothrow_move_constructible<U>...>::value );

Effects:

Initializes the contained value from the contained value of std::move(r).

Throws:

Any exception thrown by the initialization of the contained value.

Remarks:

This function does not participate in overload resolution unless all types in U… are in T… and std::is_move_constructible_v<Ui>::value is true for all Ui.

Subset (extension)

template<class... U> constexpr variant<U...> subset() & ;

template<class... U> constexpr variant<U...> subset() const& ;

Returns:

A variant<U…> whose contained value is copy-initialized from the contained value of *this and has the same type.

Throws:
If the active alternative of *this is not among the types in U…, bad_variant_access.

Otherwise, any exception thrown by the initialization of the contained value.
Remarks:

This function does not participate in overload resolution unless all types in U… are in T… and std::is_copy_constructible_v<Ui>::value is true for all Ui.

template<class... U> constexpr variant<U...> subset() && ;

template<class... U> constexpr variant<U...> subset() const&& ;

Returns:

A variant<U…> whose contained value is move-initialized from the contained value of *this and has the same type.

Throws:
If the active alternative of *this is not among the types in U…, bad_variant_access.

Otherwise, any exception thrown by the initialization of the contained value.
Remarks:

This function does not participate in overload resolution unless all types in U… are in T… and std::is_move_constructible_v<Ui>::value is true for all Ui.

variant_alternative

template<size_t I, class T> struct variant_alternative<I, T const>;

template<size_t I, class T> struct variant_alternative<I, T volatile>;

template<size_t I, class T> struct variant_alternative<I, T const volatile>;

template<size_t I, class T> struct variant_alternative<I, T&>; // extension

template<size_t I, class T> struct variant_alternative<I, T&&>; // extension

If typename variant_alternative<I, T>::type exists and is U,
- variant_alternative<I, T const>::type is U const;
- variant_alternative<I, T volatile>::type is U volatile;
- variant_alternative<I, T const volatile>::type is U const volatile.
- variant_alternative<I, T&>::type is U&.
- variant_alternative<I, T&&>::type is U&&.
Otherwise, these structs have no member type.

template<size_t I, class... T>
  struct variant_alternative<I, variant<T...>>;

When I < sizeof…(T), the nested type type is an alias for the I-th (zero-based) type in T…. Otherwise, there is no member type.

holds_alternative

template<class U, class... T>
  constexpr bool holds_alternative(const variant<T...>& v) noexcept;

Requires:

The type U occurs exactly once in T…. Otherwise, the program is ill-formed.

Returns:

true if index() is equal to the zero-based index of U in T….

get

template<size_t I, class... T>
  constexpr variant_alternative_t<I, variant<T...>>&
    get(variant<T...>& v);

template<size_t I, class... T>
  constexpr variant_alternative_t<I, variant<T...>>&&
    get(variant<T...>&& v);

template<size_t I, class... T>
  constexpr const variant_alternative_t<I, variant<T...>>&
    get(const variant<T...>& v);

template<size_t I, class... T>
  constexpr const variant_alternative_t<I, variant<T...>>&&
    get(const variant<T...>&& v);

Effects:

If v.index() is I, returns a reference to the object stored in the variant. Otherwise, throws bad_variant_access.

Remarks:

These functions do not participate in overload resolution unless I < sizeof…(T).

template<class U, class... T>
  constexpr U& get(variant<T...>& v);

template<class U, class... T>
  constexpr U&& get(variant<T...>&& v);

template<class U, class... T>
  constexpr const U& get(const variant<T...>& v);

template<class U, class... T>
  constexpr const U&& get(const variant<T...>&& v);

Requires:

The type U occurs exactly once in T…. Otherwise, the program is ill-formed.

Effects:

If v holds a value of type U, returns a reference to that value. Otherwise, throws bad_variant_access.

get_if

template<size_t I, class... T>
  constexpr add_pointer_t<variant_alternative_t<I, variant<T...>>>
    get_if(variant<T...>* v) noexcept;

template<size_t I, class... T>
  constexpr add_pointer_t<const variant_alternative_t<I, variant<T...>>>
    get_if(const variant<T...>* v) noexcept;

Effects:

A pointer to the value stored in the variant, if v != nullptr && v->index() == I. Otherwise, nullptr.

Remarks:

These functions do not participate in overload resolution unless I < sizeof…(T).

template<class U, class... T>
  constexpr add_pointer_t<U>
    get_if(variant<T...>* v) noexcept;

template<class U, class... T>
  constexpr add_pointer_t<const U>
    get_if(const variant<T...>* v) noexcept;

Requires:

The type U occurs exactly once in T…. Otherwise, the program is ill-formed.

Effects:

Equivalent to: return get_if(v); with I being the zero-based index of U in T….

Relational Operators

template<class... T>
  constexpr bool operator==(const variant<T...>& v, const variant<T...>& w);

Returns:

v.index() == w.index() && get(v) == get(w), where I is v.index().

template<class... T>
  constexpr bool operator!=(const variant<T...>& v, const variant<T...>& w);

Returns:

!(v == w).

template<class... T>
  constexpr bool operator<(const variant<T...>& v, const variant<T...>& w);

Returns:

v.index() < w.index() || (v.index() == w.index() && get(v) < get(w)), where I is v.index().

template<class... T>
  constexpr bool operator>(const variant<T...>& v, const variant<T...>& w);

Returns:

w < v.

template<class... T>
  constexpr bool operator<=(const variant<T...>& v, const variant<T...>& w);

Returns:

v.index() < w.index() || (v.index() == w.index() && get(v) <= get(w)), where I is v.index().

template<class... T>
  constexpr bool operator>=(const variant<T...>& v, const variant<T...>& w);

Returns:

w <= v.

visit

template<class F, class... V>
  constexpr /*see below*/ visit(F&& f, V&&... v);

Returns:

std::forward<F>(f)(get(std::forward<V>(v))…), where I… is v.index()….

swap

template<class... T>
  void swap(variant<T...>& v, variant<T...>& w) noexcept( /*see below*/ );

Effects:

Equivalent to v.swap(w).

bad_variant_access

class bad_variant_access: public std::exception
{
public:

    bad_variant_access() noexcept = default;

    char const * what() const noexcept
    {
        return "bad_variant_access";
    }
};