Question

I ran into this bizarre pointer-less situation in my code and am now wondering if there is a way to detect it (other than crashing). The code manages to get into an un-constructed object’s method. When b1 is constructed, a has not been constructed, and b tries to use it. By the time b2 is constructed, a is properly constructed and the code works as expected.

Beyond the obvious, “don’t do this in your code,” I’m wondering if there’s a way to detect this at compile or run time. The compiler didn’t detect it at all, and I just got some obscure and very unhelpful message about running managed code while initializing a DLL when the program crashed.

I tried to test “this”, but it’s not NULL because memory has been asigned — it’s just that the constructor has not been called so the memory is in an indeterminate state.

I would have thought the compiler would stick in some debug code to detect when this happens, but I guess not.

Is there any assertion or test or compile time switch I can use to detect this situation, or does it just come down to, “if it hurts, don’t do that?”

 OUTPUT:
 (NULL)
 test

#include "stdafx.h"
#include "cstring"

class Apple 
{
    char *sometimesinitialized;

    public:
    Apple () { 
        sometimesinitialized = new char[15];
        strcpy_s(sometimesinitialized, 5, "test");
    };
    void test()
    {
        printf("%s\n", sometimesinitialized);
    }
};

class Ball
{
    public:
    Ball();
};

Ball b1; // OOPS!
Apple a;
Ball b2; // Works as expected

Ball::Ball()
{
    a.test();
}

int _tmain(int argc, _TCHAR* argv[])
{
    scanf_s("%i");
    return 0;
}

Answer 1

If an instance of B needs to access an instance of A then it should be passed to B‘s constructor.

struct B {
    B(A &a) {
        a.test();
    }
};


A a;
B b(a);