Posts Tagged ‘ lfmf ’

Learn from my fail: check your pointers

I spent quite some time tracking down an issue I was having with my C++ project. Here is a snippet of the code:

class foo
    bool* myArray;

        myArray = new bool[5];

        for(int i=0; i<5; i++)
            myArray = false;

    void init()
        myArray[0] = true;

Every time I ran it, it would throw a memory access violation. I was very puzzled by this since it was just a standard boolean array holding true or false. It wasn’t anything very advanced. I used the Visual Studio debugger to poke around. It goes through the constructor fine and then when it goes into the init() function myArray would become a null pointer. The debugger would report that myArray has a memory address of 0x00000000.

I was very confused because how could the pointer be initialized properly in the constructor, but not be initialized when I step into another function in the same object? I chased it all afternoon and decided to give up and walk away. The next morning, I looked at the code again. Then it hit me: the problem and solution was staring at me in the face.

In the for loop in the constructor I have this code:

myArray = false;

I had forgotten to put in the array subscript! If myArray was a true null pointer, it would have just said it was uninitialized or “nullptr” with a stop icon instead of a zero memory location. What was happening was that the pointer was being set to false, which is zero. That is why the debugger showed that the memory location was 0x00000000. So when I access the pointer to assign a value, I am accessing memory that is already reserved or in-use so it throws a memory access violation. The code is “valid” which is why it doesn’t show any errors during compilation, but it doesn’t produce the desired results at runtime.

With the correct code:

myArray[i] = false;

Everything works fine and I was able to continue my project! After working with C# for a while, it’s an eye opener how much the .Net garbage collector and managed heap does for the programmer.