Is there a reference_wrapper<> for rvalue references?

Question

I wonder how the following can be done

void f(string &&s) { 
  std::string i(move(s)); 
  /* other stuff */ 
} 

int main() { 
  std::string s; 
  bi         


        

Answer 1

How can I pass an rvalue reference and store it as an rvalue reference in the call wrapper?

The problem here is that such a bind function object can be invoked multiple times. If the function object forwarded a bound parameter as rvalue this would obviously only work once. So, this is a bit of a safety issue.

But in some cases this kind of forwarding is exactly what you want. You could use a lambda as an intermediary:

bind([](string& s){f(move(s));},move(s));

Basically, I came up with this bind+lambda combination as a workaround for a missing "move-capture".

Answer 2

I was googling for "reference_wrapper for rvalues" when I stumbled on this question. Not sure whether my answer would be useful, it is not related to std::bind and actually doesn't work with it, but for some other use cases it might help somebody.

Here's my attempt to implement rvalue_reference_wrapper:

#pragma once

#include <type_traits>
#include <memory>
#include <utility>

template<class T>
class rvalue_reference_wrapper
{
public:
    static_assert(::std::is_object<T>::value, "rvalue_reference_wrapper<T> requires T to be an object type.");

    using type = T;

    rvalue_reference_wrapper(T& ref_value) = delete;

    rvalue_reference_wrapper(T&& ref_value) noexcept
        : _pointer(::std::addressof(ref_value))
    {
    }

    operator T&&() && noexcept
    {
        return ::std::move(*_pointer);
    }

    T&& get() && noexcept
    {
        return ::std::move(*_pointer);
    }

    template<class... ArgTypes>
    auto operator()(ArgTypes&&... args) &&
        -> decltype(::std::invoke(::std::declval<rvalue_reference_wrapper<T>>().get(), ::std::forward<ArgTypes>(args)...))
    {
        return (::std::invoke(::std::move(*this).get(), ::std::forward<ArgTypes>(args)...));
    }

private:
    T* _pointer;
};

template<class T>
inline rvalue_reference_wrapper<T> rv_ref(T& ref_value) = delete;

template<class T>
inline ::std::enable_if_t<!(::std::is_lvalue_reference<T>::value), rvalue_reference_wrapper<T>> rv_ref(T&& ref_value) noexcept
{
    return rvalue_reference_wrapper<T>(::std::forward<T>(ref_value));
}

#ifdef _MSC_VER
namespace std
{
    template<class T>
    struct _Unrefwrap_helper<rvalue_reference_wrapper<T>>
    {
        using type = T &&;
        static constexpr bool _Is_refwrap = true;
    };
}
#else
#pragma error("TODO : implement...")
#endif

The last specialization in namespace std allows MSVC's implementation of standard library to work with my type, e.g. when using std::make_tuple:

int a = 42;
auto p_int = std::make_unique<int>(42);
auto test_tuple = std::make_tuple(42, std::ref(a), rv_ref(std::move(p_int)));
static_assert(std::is_same<decltype(test_tuple), std::tuple<int, int &, std::unique_ptr<int>&&>>::value, "unexpected result");

I believe it would not be hard to implement similar "unwrapping" logic for other standard library implementations.

Answer 3

You can use a mutable lambda object.

auto func = [=]() mutable {
    f(std::move(s));
};

Answer 4

My take on this.

20.8.10.1.2/10 in N3225

The values of the bound arguments v1, v2, ..., vN and their corresponding types V1, V2, ..., VN depend on the types TiD derived from the call to bind and the cv-qualifiers cv of the call wrapper g as follows:

• if TiD is reference_wrapper, the argument is tid.get() and its type Vi is T&;
• if the value of is_bind_expression::value is true, the argument is tid(std::forward(uj)...) and its type Vi is result_of::type;
• if the value j of is_placeholder::value is not zero, the argument is std::forward(uj) and its type Vi is Uj&&;
• otherwise, the value is tid and its type Vi is TiD cv &.

So the only possibility to have a rvalue reference is to have is_bind_expression<TiD>::value true or is_placeholder<TiD>::value not zero. The second possibility has implications you don't want and achieving the wanted result with the first would imply that the problem we are trying to solve is solved if we restrict to the standard provided types. So, the only possibility would be to provide your own wrapper and a specialisation for is_bind_expression<TiD> (that is allowed by 20.8.10.1.1/1) as I don't see one.