Checking null vector pointers

If I were to declare a vector of complex classes e.g.

Code:

---

std::vector<MyClass*> vec;

---

, resize it (or not. either way it returns the same segmentation fault), and then check it by saying

Code:

---

MyClass*& a = vec[0];

if (a != (MyClass*)NULL) { etc; }

---

. Why would checking for null return a segfault? I've never had this problem before with vectors or just dynamically declared pointers. I tried resizing it, and initializing all the members to NULL.

Did you check that the vector is not empty but actually has an element vec[0] ?

I did not check whether the vector was empty but I did make sure it wasn't by resizing it. What are you getting at?

you don't need to cast NULL to your pointer type. if you're using G++ >=4.6 or visual studio 2010, you can use the nullptr keyword instead of NULL. it is part of the new standard that was just approved this year. I also think it's unnecessary to make a reference to a pointer, unless you're planning to modify the pointer inside the vector.

Quote:

---

Originally Posted by Rodaxoleaux

I did not check whether the vector was empty but I did make sure it wasn't by resizing it. What are you getting at?

---

Probably that about the only way it can cause a seg fault is if you've essentially created a null-reference, one of those things that logically does not exist, but at the low level bad code can end up effectively putting you there. You'd only get that by dereferencing NULL, which should only happen if the container was empty I expect.
Otherwise, you're probably not getting a segfault precisely where you think you are.

Code:

---

#include <vector>

class MyClass
{
public:
        int some_int;
};

int main()
{
        std::vector<MyClass*> vec;

        vec.resize( 100, NULL );

        MyClass*& a = vec[0];
 
        if (a != NULL)
        {
                //etc;
        }

        system( "pause" );

        return 0;
}

---

This does not segfault on my system. If it does on yours... I don't know :) Otherwise, check the rest of your code, maybe you aren't resizing or you are clearing it somewhere else.

Please note that "MyClass*&" is quite useless... a simple pointer would most likely be enough.

Cool points for nvoigt and iMalc. True. It being empty does create segfaults. I didn't know that. @Elkvis: C++0x. Oh, the greatness that is sure to produce.