C++11 rvalue reference calling copy constructor too

Question

I've been testing some C++11 features from some some. I came across r-value references and move constructors.

I implemented my first move constructor, here it is:

#include <iostream>
#include <vector>
using namespace std;

class TestClass{

public:
    TestClass(int s):
        size(s), arr(new int[s]){
    }
    ~TestClass(){
        if (arr)
            delete arr;
    }
    // copy constructor
    TestClass(const TestClass& other):
            size(other.size), arr(new int[other.size]){
        std::copy(other.arr, other.arr + other.size, arr);
    }

    // move constructor
    TestClass(TestClass&& other){
        arr=other.arr;
        size=other.size;

        other.arr=nullptr;
        other.size=0;
    }

private:
    int size;
    int * arr;
};

int main(){
    vector<TestClass> vec;

    clock_t start=clock();
    for(int i=0;i<500000;i++){
        vec.push_back(TestClass(1000));
    }
    clock_t stop=clock();
    cout<<stop-start<<endl;

    return 0;
}

The code works fine. Anyway putting a std::cout inside the copy constructor i noticed that it gets called! And a lot of times.. (move constructor 500000 times, copy constructor 524287 times).

What surprised me more is that if i comment out the copy constructor from the code, the whole program gets a lot faster, and this time the move constructor is called 1024287 times.

Any clue?

@pinoscotto Just a side note: you have undefined behavior, you should use delete[], and you don't need the if (arr) ... in the destructor. So the destructor should be simply ~TestClass(){ delete[] arr; }. — Ali
@pinoscotto It is always safe to call delete[] without the if: (1) if the new throws in the initialization, the destructor won't be called (2) otherwise you either point to a valid address (3) or have a nullptr and it is always safe to delete a nullptr (guaranteed by the standard). — Ali

Howard Hinnant Howard Hinnant · Accepted Answer · 2013-08-06T16:25:17

Put noexcept on your move constructor:

TestClass(TestClass&& other) noexcept {

Elaboration: I was going to give this one Pierre, but unfortunately the cppreference source is only approximately correct.

In C++03

vector<T>::push_back(T)

has the "strong exception guarantee". That means that if the push_back throws an exception, the vector is left in the same state it had prior to the call to push_back.

This guarantee is problematic if the move constructor throws an exception.

When the vector reallocates, it would like to move the elements from the old buffer to the new. However if any one of those moves throws an exception (besides the first), then it is left in a state where the old buffer has been modified, and the new buffer doesn't yet contain everything it is supposed to. The vector can't restore the old buffer to its original state because it would have to move elements back to do so, those moves might also fail.

So a rule was laid down for C++11:

If T has a noexcept move constructor, that can be used to move the elements from the old buffer to the new.
Otherwise if T has a copy constructor, that will be used instead.
Otherwise (if there is no accessible copy constructor), then the move constructor will be used after all, however in this case, the strong exception safety guarantee is no longer given.

Clarification: "copy constructor" in rule 2 means a constructor taking a const T&, not one of those weenie so-called T& copy constructors. :-)

C++11 rvalue reference calling copy constructor too

4 Answers