Unresolved external symbol error :-(

[NorthernMonkey]

Hi,

I was just wondering if anyone could explain in simple terms what these errors are, and why they come about.

More specifically if you could explain why this code:

static const int SIZE = 8*sizeof(unsigned);
	static int* mask[SIZE];
public: static void InitMask()
	{
		int num = 1;
		for (int i=0;i<SIZE;i++)
		{
			mask[i] = &num;
			num=num*2;
		}
	}

produces this error

main.obj : error LNK2001: unresolved external symbol "private: static int * * Binary::mask" (?mask@Binary@@0PAPAHA)
fatal error LNK1120: 1 unresolved externals

whenever I try to call it from my main.cpp using Binary::InitMas k(); (Binary is the class name where this function is declared. I've also tried calling this function using a Binary object eg binObj.InitMask (); The error is the same.

I've only had a few weeks of C++ so pointers etc are a little confusing after learning Java, so the simpler the explanation the better!!

[AmeL]

main.obj : error LNK2001: unresolved external symbol "private: static int * * Binary::mask" (?mask@Binary@@0PAPAHA)
fatal error LNK1120: 1 unresolved externals

Well, the error happens when your linker tries to link your object file ( let says main.o) with the library and with something you have implemented. When the linker read Binary::InitMas k(), it can not find where the implementation of InitMask() is. So the linking error takes place.
Well, let says your design looks like this :

• Binary.h : this is where you have declare the interface of class Binary.
• Binary.cpp : this is where you implement the class members.
• main.cpp : this is where you instantiate the object from class Binary.

If your design looks like this, make sure that in your main.cpp, u did include the header file (Binary.h), else your linker can't find the implementation.

Thanks
/AmeL

[JosAH]

When you define your class in your .h file you only declare its members. You
have to define them in your .cpp file; so you have to define your static int array
in your .cpp file and you're in business.

kind regards,

Jos

ps. only inline members are defined inside the clas definition.

[NorthernMonkey]

Thanks for your input. I'm still new to C++ andthe whole .h .cpp files and when to use which.

So in the .h file I should declare all the possible variable that my class will use, private and public.

In the .cpp of the class I should implement all of my functions?

Can the .cpp file access the private data members declared in the .h then? Because that is the main confusion I have. When I create an accessor method I put it in the .h because I always thought the .cpp was unable to access the private members.

With the code above it is all in the Binary.h file and the calling is in the main.cpp. The main.cpp cannot be touched by me, it is a standard test, so I can't declare things within it or anything. I am just experimenting to try and see if the mask function I implemented is actually working, and so far it isn't :(

[Banfa]

It is normal when defining a class to use 2 files, 1 header (.h) and 1 source (.cpp) file.

The header contains the class declaration, that is the

class MyClass
{
public
    MyClass();
    ~MyClass();
    Print();

private:
    int value;
}

the source file contains the function (or method) definitions for the class as well as static data declarations if the class has any static data members

#include <iostream>

MyClass::MyClass() : value(42)
{
}

MyClass::~MyClass()
{
}

MyClass::Print()
{
    std::cout << value;
}

the implementation of the function can be placed in the header file inside the class declaration but this makes them inline functions. As such you need to be careful and generally I only put functions that have 1 or 2 lines of source into the class declaration in the header file (accessor methods often fit this category).

Access to the private data and function members of the class is nothing to do with what file you are in but is to do with the scope you are in. You have to be in the scope of the class to access private members, for instance in the MyClass::Print function definition I can access value because the code for that function is in the scope of the class because the function is a member of the class. This function

int GetValueAdd1(MyClass &mc)
{
    return mc.value+1;
}

will always cause a compile error no matter where it is placed (class header, class source or any other file) because it is try to access a private member of MyClass but it is not a member (or friend) of MyClass so it's definition is not in scope for the class.

[Banfa]

BTW in your original code the way you are declaring and initialising mask is wrong and you will not get the results you expect (assuming that the program doesn't just crash).

[NorthernMonkey]

Thanks very much again for your insight. It now makes more sense. We had none of this explained to us, which is evident based on the amount of people on this unit having real trouble with this task.

Without going into coding detail or how to do it correctly, could you explain why my mask is wrong?

I thought I could declare an array by simply saying

int myArray[5]; like I could in Java.

Then in a simple loop I could add a number. So in this case I thought that the code

int myArray[5];
for (int i=0;i<5;i++)
{
  myArray[i]=i;
}

would have the same effect of int myArray[] = {0,1,2,3,4];

in that both of them would produce an array of size 5 which I could then access.

On a side note. If I declared my array in m header, then in my .cpp I created a Binary::InitMas k() function. Am I correct in assuming I could then use it within other functions by simply calling InitMask() , or does it still require a calling object?

[Banfa]

I thought I could declare an array by simply saying

int myArray[5]; like I could in Java.

Yes you could declare it like that but you didn't, look at your own code in the first post.

On a side note. If I declared my array in m header, then in my .cpp I created a Binary::InitMas k() function. Am I correct in assuming I could then use it within other functions by simply calling InitMask() , or does it still require a calling object?

By other functions do you mean other class member functions or just other functions in the program?

In other member functions of the class you can access the array directly because it is in scope for them. From the definition of (code for) another class member function you can call InitMask directly without requiring a new instance of the class as you are already working with an instance of the class.

If InitMask is a public function of the class you can call it from any function in the program that has access to a instance of the class.

Sorry your question is vague it makes it hard to answer.

[NorthernMonkey]

Yes sorry about that. The whole pointer thing was me trying to figure out my error to no avail. I just forgot to take it back out again before pasting the code here.

Regarding the calling of InitMask() within another function.

In my Binary.h If I simply have:

static void InitMask();
static unsigned getMaskValue(int v);

Then in Binary.cpp I would declare my mask object?

private static unsigned mask[5];

my InitMask() function, for this example
{
    for (int i=0;i < 5;i++)
         {
          mask[i] = i;
         }
}

static unsigned getMaskValue(int v)
{
    return mask[v];
}

Could I then, in my main method simply run:

Binary::InitMask();
unsigned test = Binary::getMaskValue(2);
cout << test << endl;

in order to access and output a value in my array mask? Or do I still need a declared calling object to call InitMask?

[Banfa]

Are you creating a class to hold the mask variable or are you just using global functions?

In the first case you are missing all the class nomenclature from your code (the MyClass:: etc), in both cases the keyword private on your array declaration is not required (private is a keyword that I believe is only valid when used in/with a class declaration) it should be used when declaring the array in the class declaration.

From your main code it appears you are putting these things into the Binary class, take a look at the example MyClass I posted, which is (I hope) compilable, for syntax.

[NorthernMonkey]

I'm beginning to get the hang of it now thank you very much for all your patience and understanding.

I finally figured out what the mask is used for. We are finding the inverse of a number. For that we need to be able to set individual bits on and off using an On(int i) function and an Off(int i) function respectively.

To do this it seems a little clearer now. Say I wanted to set bit(3) to 1.

I can just add my current Binary object/unsigned to the mask[3].

eg I have Binary object binNumber of int value 1, binary 0......001;

if I did binNumber += mask[3]; // mask[3] being int 8, binary 100;

am I correct in assuming it would add 8 to 1, 100 to 001 giving 9 or 101?

I'm getting there, finally understanding the correct calling procedure, I think it was not referring to the class that was holding me up quite a lot!