The basic version of my situation looks like this.

class MyClass 
{
public:
    MyClass(std::vector<int> v)
        : m_vector(v) {}

    void doSomething();

private:
    std::vector<int> m_vector;
}

int main() 
{
    std::vector<int> myVector;
    // ...
    // Do stuff with myVector
    // ...
    
    MyClass myItem(myVector);
    myItem.doSomething();
}

Obviously it is more efficient to move the vector into MyClass rather than copy it, but I am never sure how to define the constructor of MyClass and how to pass the argument when instantiating it.

The following versions all compile correctly.

A:

MyClass(const std::vector<int>& v)
    : m_vector(std::move(v)) {}

MyClass myItem(myVector);

B:

MyClass(std::vector<int> v)
    : m_vector(std::move(v)) {}

MyClass myItem(myVector);

C:

MyClass(const std::vector<int>& v)
    : m_vector(v) {}

MyClass myItem(std::move(myVector));

D:

MyClass(const std::vector<int>& v)
    : m_vector(std::move(v)) {}

MyClass myItem(std::move(myVector));

E:

MyClass(std::vector<int>&& v)
    : m_vector(std::move(v)) {}

MyClass myItem(std::move(myVector));

…and I’m just getting more confused. I guess the question is not even which one is “correct”, but which one guarantees the least copies?