Pointers

Pointers and OOPS Concept

POINTERS What are Pointers? Pointers are basically the same as any other variable. However, what is different about them is that instead of containing actual data, they contain a pointer to the memory location where information can be found. This is a very important concept, and many programs and ideas rely on pointers as the basis of their design, linked lists for example. Getting Started How do I define a pointer? Well, the same as any other variable, except you add an asterisk before its name. So, for example, the following code creates two pointers, both of which point to an integer: int* pNumberOne; int* pNumberTwo; Notice the 'p' prefix in front of the two variable names? This is a convention used to indicate that the variable is a pointer. Now, lets make these pointers actually point to something: pNumberOne = &some_number; pNumberTwo = &some_other_number; The & (ampersand) sign should be read as 'the address of', and causes the address in memory of a variable to be returned, instead of the variable itself. So, in this example, pNumberOne is set to equal the address of some_number, so pNumberOne now points to some_number; Now, if we want to refer to the address of some_number, we can use pNumberOne. If we want to refer to the value of some_number from pNumberOne, we would have to say *pNumberOne. The * dereferences the pointer, and should be read as 'the memory location pointed to by', unless in a declaration, as in the line 'int *pNumber'. What we've learnt so far: an example: Phew! That's a lot to take in, I'd recommend that if you don't understand any of those concepts to give it another read through; pointers are a complex subject, and it can take a while to master them. Here is an example which demonstrates the ideas discussed above. It is written in C, without the C++ extensions. #include <stdio.h> SUNSAT The Perfect Team

Pointers and OOPS Concept

void main() { // declare the variables: int nNumber; int *pPointer; // now, give a value to them: nNumber = 15; pPointer = &nNumber; // print out the value of nNumber: printf("nNumber is equal to : %d\n", nNumber); // now, alter nNumber through pPointer: *pPointer = 25; // prove that nNumber has changed as a result of the above by // printing its value again: printf("nNumber is equal to : %d\n", nNumber); } Read through, and compile the above code sample, and make sure you understand why it works. Then, when you are ready, read on! A trap! See if you can spot the fault in the program below: #include <stdio.h> int *pPointer; void SomeFunction(); { int nNumber; nNumber = 25; // make pPointer point to nNumber: pPointer = &nNumber; } void main() { SomeFunction(); // make pPointer point to something

SUNSAT The Perfect Team

Pointers and OOPS Concept // why does this fail? printf("Value of *pPointer: %d\n", *pPointer); } This program firstly calls the SomeFunction function, which creates a variable called nNumber, and then makes the pPointer point to it. Then, however, is where the problem is. When the function leaves, nNumber is deleted, because it is a local variable. Local variables are always deleted when execution leaves the block they were defined in. This means when SomeFunction returns to main(), the variable is deleted, so pPointer is pointing at where the variable used to be, which no longer belongs to this program. If you do not understand this, it may be wise to read back over on local and global variables, and on scope. This concept is also important. So how can the problem be solved? The answer is by using a technique known as dynamic allocation. Please be aware that this is different between C and C++. Since most developers are now using C++, this is the dialect that the code below is using. Dynamic Allocation Dynamic allocation is perhaps the key to pointers. It is used to allocate memory without having to define variables and then make pointers point to them. Although the concept may appear confusing, it is really simple. The following code demonstrates how to allocate memory for an integer: int *pNumber; pNumber = new int; The first line declares the pointer, pNumber. The second line then allocates memory for an integer, and then makes pNumber point to this new memory. Here is another example, this time using a double: double *pDouble; pDouble = new double; The formula is the same every time, so you can't really fail with this bit. What is different about dynamic allocation, however, is that the memory you allocate is not deleted when the function returns, or when execution leaves the current block. So, if we rewrite the above example using dynamic allocation, we can see that it works fine now: #include <stdio.h> int *pPointer; void SomeFunction()

SUNSAT The Perfect Team

Pointers and OOPS Concept { // make pPointer point to a new integer pPointer = new int; *pPointer = 25; } void main() { SomeFunction(); // make pPointer point to something printf("Value of *pPointer: %d\n", *pPointer); } Read through, and compile the above code sample, and make sure you understand why it works. When SomeFunction is called, it allocates some memory, and makes pPointer point to it. This time, when the function returns, the new memory is left intact, so pPointer still points to something useful. That's it for dynamic allocation! Make sure you understand this, and then read on to learn about the fly in the ointment, and why there is still a serious error in the code above. Memory comes, memory goes There's always a complication, and this one could become quite serious, although it's very easy to remedy. The problem is that although the memory that you allocate using dynamic allocation is conveniently left intact, it actually never gets deleted automatically. That is, the memory will stay allocated until you tell the computer that you've finished with it. The upshot of this is that if you don't tell the computer that you've finished with the memory, it will be wasting space that other applications, or other parts of your application could be using. This eventually will lead to a system crash through all the memory being used up, so it's pretty important. Freeing the memory when you've finished with it is very simple: delete pPointer; That's all there is to it. You have to be careful however, that you pass a valid pointer, i.e. that is a pointer that actually points to some memory you've allocated, and not just any old rubbish. Trying to delete memory that's already been freed is dangerous and can lead to your program crashing. So here is the example again, updated this time so it doesn't waste any memory: #include <stdio.h> int *pPointer; void SomeFunction() {

SUNSAT The Perfect Team

Pointers and OOPS Concept // make pPointer point to a new integer pPointer = new int; *pPointer = 25; } void main() { SomeFunction(); // make pPointer point to something printf("Value of *pPointer: %d\n", *pPointer); delete pPointer; } One line difference is all it took, but this line is essential. If you don't delete the memory, you get what is known as a 'memory leak', where memory is gradually leaking away, and can't be reused unless the application is closed. Passing pointers to functions The ability to pass pointers to function is very useful, but very easy to master. If we were to make a program that takes a number and adds five to it, we might write something like the following: #include <stdio.h> void AddFive(int Number) { Number = Number + 5; } void main() { int nMyNumber = 18; printf("My original number is %d\n", nMyNumber); AddFive(nMyNumber); printf("My new number is %d\n", nMyNumber); } However, the problem with this is that the Number referred to in AddFive is a copy of the variable nMyNumber passed to the function, not the variable itself. Therefore, the line 'Number = Number + 5' adds five to the copy of the variable, leaving the original variable, in main(), unaffected. Try running the program to prove this. To get around this problem, we can pass a pointer to where the number is kept in memory to the function, but we'll have to alter the function so that it expects a pointer to a

SUNSAT The Perfect Team

Pointers and OOPS Concept number, not a number. To do this, we change 'void AddFive(int Number)' to 'void AddFive(int* Number)', adding the asterisk. Here is the program again, with the changes made. Notice that we have to make sure we pass the address of nMyNumber instead of the number itself? This is done by adding the & sign, which (as you will recall) is read as 'the address of'. #include <stdio.h> void AddFive(int* Number) { *Number = *Number + 5; } void main() { int nMyNumber = 18; printf("My original number is %d\n", nMyNumber); AddFive(&nMyNumber); printf("My new number is %d\n", nMyNumber); } Try coming up with an example of your own to demonstrate this. Notice the importance of the * before Number in the AddFive function? This is needed to tell the compiler that we want to add five to the number pointed to by the variable Number, rather than add five to the pointer itself. The final thing to note about functions is that you can return pointers from them as well, like this: int * MyFunction(); In this example, MyFunction returns a pointer to an integer. Pointers to Classes There are a couple of other caveats with pointers, one of which is structures or classes. You can define a class as follows: class MyClass { public: int m_Number; char m_Character; }; Then, you can define a variable of type MyClass as follows:

SUNSAT The Perfect Team

Pointers and OOPS Concept MyClass thing; You should already know this, if not try reading up on this area. To define a pointer to MyClass you would use: MyClass *thing; as you would expect. Then, you would allocate some memory and make this pointer point to the memory: thing = new MyClass; This is where the problem comes in, how then would you use this pointer. Well, normally you would write 'thing.m_Number', but you can't with a pointer because thing is not a MyClass, but a pointer to it, so thing itself does not contain a variable called m_Number; it is the structure that it points to that contains m_Number. Therefore, we must use a different convention. This is to replace the '.' (dot) with a -> (dash followed by a greater than sign). An example showing this is below: class MyClass { public: int m_Number; char m_Character; }; void main() { MyClass *pPointer; pPointer = new MyClass; pPointer->m_Number = 10; pPointer->m_Character = 's'; delete pPointer; } Pointers to Arrays You can also make pointers that point to arrays. This is done as follows: int *pArray; pArray = new int[6]; This will create a pointer, pArray, and make it point to an array of six elements. The other method, not using dynamic allocation, is as follows:

SUNSAT The Perfect Team

Pointers and OOPS Concept int *pArray; int MyArray[6]; pArray = &MyArray[0]; Note that, instead of writing &MyArray[0], you can simply write MyArray. This, of course, only applies to arrays, and is a result of how they implemented in the C/C++ language. A common pitfall is to write pArray = &MyArray;, but this is incorrect. If you write this, you will end up with a pointer to a pointer to an array (no typo), which is certainly not what you want. Using pointers to arrays Once you have a pointer to an array, how do you use it? Well, lets say you have a pointer to an array of ints. The pointer will initially point to the first value in the array, as the following example shows: #include <stdio.h> void main() { int Array[3]; Array[0] = 10; Array[1] = 20; Array[2] = 30; int *pArray; pArray = &Array[0]; printf("pArray points to the value %d\n", *pArray); } To make the pointer move to the next value in the array, we can say pArray++. We can also, as some of you probably guessed by now, say pArray + 2, which would move the array pointer on by two elements. The thing to be careful of is that you know the what the upper bound of the array is (3 in this example), because the compiler cannot check that you have not gone past the end of array when you are using pointers, so you could easily end up crashing the system. Here is the example again, showing this time the three values that we set: #include <stdio.h> void main() { int Array[3]; Array[0] = 10; Array[1] = 20;

SUNSAT The Perfect Team

Pointers and OOPS Concept Array[2] = 30; int *pArray; pArray = &Array[0]; printf("pArray points to the value %d\n", *pArray); pArray++; printf("pArray points to the value %d\n", *pArray); pArray++; printf("pArray points to the value %d\n", *pArray); } You can also subtract values as well, so pArray - 2 is 2 elements from where pArray is currently pointing. Make sure, however, that you add or subtract to the pointer and not to its value. This kind of manipulation using pointers and arrays is most useful when used in loops, such as the for or while loops. Note that, also, if you have a pointer to a value, e.g. int* pNumberSet, you can treat it as an array, for example pNumberSet[0] is equivalent to *pNumberSet, and pNumberSet[1] is equivalent to *(pNumberSet + 1). One final word of warning for arrays, is that if you allocate memory for an array using new, as in the following example: int *pArray; pArray = new int[6]; it must be deleted using the following: delete[] pArray; Notice the [] after delete. This tells the compiler that it is deleting a whole array, and not just a single item. You must use this method whenever arrays are involved, otherwise you will end up with a memory leak. Last Words One final note: you must not delete memory that you did not allocate using new, as in the following example: void main() { int number; int *pNumber = number; delete pNumber; // wrong - *pNumber wasn't allocated using new.

SUNSAT The Perfect Team

Pointers and OOPS Concept } Common Questions and FAQ Q: Why do I get 'symbol undefined' errors on new and delete? A: This is most likely caused by your source file being interpreted by the compiler as being a plain C file, and the new and delete operators are a new feature of C++. This is usually remedied by ensuring that you are using a .cpp extension on your source code files. Q: What's the difference between new and malloc? A: new is a keyword only present in C++, and is now the standard way (other than using Windows' memory allocation routines) to allocate memory. You should never use malloc within a C C++ application unless absolutely necessary. Because malloc is not designed for object-oriented features of C++, using it to allocate memory for classes will prevent the constructor of the class being called, as just one example of the problems that can arise. As a result of the problems that arise from the use of malloc and free, and because they are now for all intents and purposes obsolete, they are not discussed in any detail in this article, and I would discourage their use wherever possible. Q: Can I use free and delete together? A: You should free memory with the equivalent routine to that used to allocated it. For instance use free only on memory allocated with malloc, delete only on memory allocated with new and so on. References References are, to a certain degree, out of the scope of this article. However, since I've been asked many times by people reading this about them, I will discuss them briefly. They are very much related to pointers in that, in many cases, they can be used as a simpler alternative. If you recall above, I mentioned that the ampersand (&) is read as 'the address of' unless in a declaration. In the case of its presence in a declaration such as that shown below, it should be read as 'a reference to'. int& Number = myOtherNumber; Number = 25; The reference is like a pointer to myOtherNumber, except that it is automatically dereferenced, so it behaves just as it were the actual value type rather than a pointer type. The equivalent code to this, using pointers, is shown below: int* pNumber = &myOtherNumber; *pNumber = 25;

SUNSAT The Perfect Team

Pointers and OOPS Concept The other difference between pointers and references is that you cannot 'reseat' a reference, that is to say that you cannot change what it is pointing to after its declaration. For instance, the following code would output '20': int myFirstNumber = 25; int mySecondNumber = 20; int &myReference = myFirstNumber; myReference = mySecondNumber; printf("%d", myFristNumber); When in a class, the value of the reference must be set by the constructor in the following way: CMyClass::CMyClass(int &variable) : m_MyReferenceInCMyClass(variable) { // constructor code here } Summary This topic is very hard to master initially, so it is worth looking over at least twice: most people do not understand it immediately. Here are the main points again: 1. Pointers are variables that point to an area in memory You define a pointer by adding an asterisk (*) in front of the variable name (i.e. int *number). 2. You can get the address of any variable by adding an ampersand (&) in front of it (i.e. pNumber = &my_number). 3. The asterisk, unless in a declaration (such as int *number), should be read as 'the memory location pointed to by'. 4. The ampersand, unless in a declaration (such as int &number), should be read 'the address of'. 5. You can allocate memory using the 'new' keyword. 6. Pointers MUST be of the same type as the variables you want them to point to, so int *number will not point to a MyClass. 7. You can pass pointers to functions. 8. You must delete memory that you have allocated by using the 'delete' keyword. 9. You can get a pointer to array that already exists by using &array[0];. 10. You must delete an array that is dynamically allocated using delete[], not just delete. This is not an absolutely complete guide to pointers, there are a few other things that I could have covered in more detail, such as pointers to pointers, and also things that I chose not to cover at all, such as function pointers, which I believe are too complex for a

SUNSAT The Perfect Team

Pointers and OOPS Concept beginner's article, and things that are used rarely enough so that a beginner would be better off not flummoxed by these unwieldy details. That's it! Try running some of the programs presented here, and come up with some examples of your own.

OOPS Concept Introduction This article explains why C++ (among other Object Oriented Languages) was invented. What its advantages over procedural languages are, and it shows an example of this. What is Object Orientation (OO)? Once upon a time... In a simple line: OO is a big improvement in the organization of code. Object Oriented Programming (OOP) is kind of a philosophy. Let me explain: suppose there is a company, a BIG company. And every employee in this company has access to everything (is it still a company?). Even the cleaner has access to the science department ready to mess around with some dangerous toxins. Luckily, everyone is very faithful and will not do such a thing someday soon. Then, one day, after this company grows even bigger and more complex, the whole organization collapses. Quickly now, what's the problem here? Exactly, the organization has become too complicated. Everything is confusing, the scientist couldn't find the toxins anymore because the cleaner messed around with them. The salesmen went crazy because the telephonist accidentally sold everything, etc. Now, think of this company as a program. This program has no real structure. Everything knows everything, this is not much of a problem in a simple Hello World program. But like I said, this is a BIG program (company). The cleaner, scientist, salesmen, and the telephonist in this program are functions. And the toxins and the products for sale are the data. There is of course much more, use your imagination. The point here is that every function has access to all data, which is the case in procedural languages like C (although C has some features to prevent this, they aren't much I guess). Meaning that, in a situation, data could easily be lost or corrupted when a new function is made. And if a data item needed to be edited, you would also have to edit every function which works with that particular data item... This is not very fun when you have worked on a program for like let's say... a few months. OO to the rescue! This is why OO is here, to make irrational programs like the one above more rational. To make the chaotic ordered etc. Let's use our imagination again... suppose there is a SUNSAT The Perfect Team

Pointers and OOPS Concept company, a BIG company. This company is very successful and is divided into multiple departments. Each department has its own purpose and its own security measures. This time, the telephonist cannot sell any products because all the products are located in a storage facility to which only the salesmen have access (HAH!). And the cleaner doesn't have access to the science department either, because only the scientists have the key to it! This, can be called a company. There is a rational organization with security measures to prevent any corruption or sloppiness within the company. Now, the boss (yes, this time there is a boss) can go home relaxed without having to worry about the wild fantasies of the cleaner. One word: OOP. We could do this to the program by using classes. This is what we're gonna do. The Company example OOP-less Let's take a look at what disaster could occur in a program without OO. (The example is small and unreal, but it will help you get the point). Note: it is still written in C++ but without the OO concept in mind. Collapse //FILE: Evil_Workers.cpp #include #include //for getche const int LastChecked = 10; //unaltered value int toxin = LastChecked; //toxin data item int products = LastChecked; //products in stock void cleaner() { cout << "\nHAHAH I AM AN EVIL CLEANER!!\ntoxin = " << toxin << "\n*Alters toxin!*\n"; //yes he is evil! toxin += 2; //cleaner CAN alter toxin! cout << "toxin = " << toxin << endl; } void telephonist() { cout << "\nHAHAH I AM AN EVIL TELEPHONIST!!\nproducts = " << products << "\n*Sells all products!*\n"; //NOOOOO!! products -= 10; //telephonist CAN sell products! cout << "products = " << products << endl; } void scientist()

SUNSAT The Perfect Team

Pointers and OOPS Concept { cout << "\nScientist:\n"; //scientist checks if toxin is still unaltered if(toxin == LastChecked) cout << "I'm glad nobody messed with my toxin!\n"; else cout << "Oh my god, somebody messed with my toxin!!!!\n"; } void salesman() { cout << "\nSalesman:\n"; //salesman checks if no products are sold if(products == LastChecked) cout << "I'm glad nobody sold anything!\n"; else cout << "Oh my god, somebody sold stuff!!!!!\n"; } void main() { scientist(); //scientist checks salesman(); //salesman checks cleaner(); //cleaner alters telephonist(); //telephonist sells scientist(); //scientist checks salesman(); //salesman checks cout << "\n\nPress enter."; cout.flush(); getche(); //just so the program doesnt terminate immediatly } As you can see here, toxin and products are defined globally. Meaning they are accessible by any part of the program. So now, cleaner() and telephonist() can both access these variables. Which is of course what we don't want. We only want scientist() and salesman() to be able to access these variables. Then again, we don't want scientist() to be able to mess with products and the same goes for salesman() and toxin. There is another problem with the above code. The program doesn't really fit in real life situations. Take scientist(), it is a function. Now in real life, a scientist would be more of an object instead of a function. And the scientist checking up on the toxins would be the function of the scientist. So how does OO solve all this? In C++, it solves this by using classes. I will not go very in-depth on classes in this article (if you do want in-depth info on classes, go here). OOP-ness Now, let's see how OOP solves the problem using classes in this example:

SUNSAT The Perfect Team

Pointers and OOPS Concept Collapse //FILE: Good_Workers.cpp #include //a class named Scientist will be defined here class Scientist { private: //important! int toxin; public: //constructor with initialization list Scientist(int Toxin_Value) : toxin(Toxin_Value) {} //member functions which do something with toxin int GetToxin() //return the value of toxin { return toxin; } void AlterToxin(int Toxin_Value) //set toxin to Toxin_Value { toxin = Toxin_Value; } }; //for OO's sake let's create a Cleaner class class Cleaner { public: void Evil_Altertoxin(Scientist& victim, int value) { victim.toxin = value; //this will generate a compiler error } }; void main() { const int DaveVar = 10; //correct toxin value Scientist Dave(DaveVar); //create object Dave Cleaner John; //create object John //Dave checks to see if toxin is still unaltered cout << "\nScientist Dave:\n"; if(Dave.GetToxin() == DaveVar) cout << "I'm glad nobody messed with my toxin!\n";

SUNSAT The Perfect Team

Pointers and OOPS Concept else cout << "Oh my god, somebody altered my toxin!\n"; //John attempts to alter toxin cout << "\nEVIL Cleaner John:\nLet's try to alter Dave's toxin!"; John.Evil_Altertoxin(Dave, 0); //just for demonstration toxin = 0; //another compiler error } Good_Workers.cpp will generate 2 compiler errors. The first one for this: victim.toxin = value; on line 32. And another one, for this: toxin = 0; on line 54. First of all, let's look at our class. The Scientist class, has one data item, two member functions and a constructor. The member functions are defined under the public scope resolution operator. When something is defined under the public scope resolution operator, the whole program has access to the data, however, whatever data or member functions are under it, they are still not globally defined. This is why the statement toxin = 0; generates a compiler error, because the compiler can't see a global toxin. And it would say something like: error C2065: 'toxin' : undeclared identifier But then, how can the members of a class be accessed, you ask? By the dot operator, as you can see with Dave.GetToxin() and John.Evil_Altertoxin(Dave, 0). Speaking of which, let's take a look at that member function Evil_Altertoxin(): Encapsulation void Evil_Altertoxin(Scientist& victim, int value) { victim.toxin = value; //this will generate a compiler error } This particular function takes an argument of type (class) Scientist, and one of type int. Then it tries to alter victim's toxin by assigning it with value. Now everything looks okay, since we formally used the dot operator to access victim's toxin. However, the compiler gives us an error that says we still can't access it! This is because we have declared toxin under the private access specifier. When a private access specifier is stated somewhere in the class, it means that everything following the access specifier can only be accessed by

SUNSAT The Perfect Team

Pointers and OOPS Concept other parts within the class itself (until another access specifier is stated). In the Scientist class, this means that only the functions GetToxin() and AlterToxin() (and the constructor) can access toxin. This also means that a function like Evil_Altertoxin() CAN'T access toxin because it is not a member of the Scientist class (this is called encapsulation). Just be nice! There is, however, one way for John to "access" toxin. He could ask Dave to alter toxin to a desired value. Then instead of: victim.toxin = value; //this will generate a compiler error We would have to state: victim.AlterToxin(value); //this will compile Even then, John doesn't really access toxin, because Dave still does all the work here. New datatypes! const int DaveVar = 10; //correct toxin value Scientist Dave(DaveVar); //create object Dave Cleaner John; //create object John Have you also noticed the declaration of Dave and John? Yes people, we have created 2 new datatypes: Scientist and Cleaner! Now, instead of functions, they are actually instance data, which is a lot more comparable to the real world. Everything fits together as one! "Thank God, it's just 1!" What about the I-don't-have-to-rewrite-every-function thing? Suppose you have a function that returns the square number of an argument, and you want that function to be able to operate on as many datatypes as possible. With or without overloading, you would still end up writing a lot of functions. Thanks to a new feature, templates, we only have to write one! //FILE: Templates.cpp #include #include //for getche template //warning: translate class as type! T SqrNum(T argNum) {

SUNSAT The Perfect Team

Pointers and OOPS Concept return (argNum * argNum); //return type is of T } void main() { int inum = 2; long lnum = 4; double dnum = 5.6; cout << "inum * inum = " << SqrNum(inum) << endl << "lnum * lnum = " << SqrNum(lnum) << endl << "dnum * dnum = " << SqrNum(dnum) << endl << endl << "Press enter."; cout.flush(); getche(); //just so the program doesnt terminate immediatly } What's happening here!? I'll explain it to you, however, I won't go in-depth on the syntax. The function SqrNum() isn't actually being created on definition. It is more of a blueprint for creating other functions. When the compiler receives a "call" to a template function, it won't actually call the template function. Instead, the compiler will create the right function for the right datatypes using the template function. In the first "call" to SqrNum(inum), the compiler will start a process like this: • •

Hm... inum is of type int, so we'll treat T as int. Hm... let's create a function that works on our int argument.

(it will look like this:) int SqrNum(int argNum) //notice how all the Ts have been replaced by ints { return (argNum * argNum); //return type is of int } And when that is done, the actual function call is being made. Of course, the same goes for SqrNum(lnum) and SqrNum(dnum). Also note: the compiler won't actually replace the template in the source file with the functions! Infinity... Now, there are loads of more stuff that you can do with classes, by classes, or in classes. Like, operator-overloading, function-overloading, inheritance and polymophism, abstract classes and even more. It's just beyond the scope of this article. OO is what you do with OO, and beyond..

SUNSAT The Perfect Team