Pointer Variables A pointer is a variable that contains a memory address The address is commonly the location of another variable in memory This pointer “points” to the other variable in memory A pointer is declared using the general statement: type *name; e.g. int *i1;

Important Uses of Pointers Allows functions to modify their arguments Supports dynamic memory allocation functions Improves the efficiency of certain routines Allows communication to other functions, programs, and hardware using shared memory and memory transfers Pointers are one of the most powerful features of C/C++ but are also the most dangerous

Pointer Operators &b return the memory address of variable b *p return the value of the variable at address p * is the complement of & so that *(&b)=b These operators have a higher precedence than other arithmetic operators except urinary minus which is equal int b = 7777, *p; p = &b; cout << *p;  7777

A Variable Pointing to Another Variable Memory address Variable in memory 1000 1003 1001 1002 7777 1004 p i1 *p

Pointer Expressions p = q pointer assignment p++, p-- increment and decrement p = p + i addition of an integer i p = p - i subtraction of an integer i p == q test for equality p < q , p > q pointer comparison

char *ch = 3000; int *i = 3000; expression Memory (bytes) ch 3000 i ch + 1 3001 ch + 2 3002 ch + 3 3003 ch + 4 3004 i + 1 ch + 5 3005

Pointers and Arrays A 1D array is equivalent to a pointer to the first element There are two ways of accessing arrays:  pointer arithmetic and array indexing, s1[i]=*(s1+i) Indexing can also be used on pointer variables char s1[6] = “peach”, *p1; p1 = s1; cout << s1[3] << *(p1+3) << p1[3] << *(s1+3);  cccc

Passing Pointers and Arrays to Functions 1D arrays and pointers are interchangeable when passing arguments to functions Since the memory location is passed to the function it will modify the contents of the array: call by reference No copies of the array are made when passed to the function Be sure to know the boundaries of the array to avoid invalid memory access

Modifying Function Arguments Pointers allow C functions to modify their arguments call by value int i = 0; fun1(i); cout << i;  0 call by reference int i = 0; fun2(&i); cout << i;  7777 void fun1(int i) { i = 7777; } void fun2(int *i) { *i = 7777; }

Array Swapping int ar[2]={1,2}, br[2]={3,4}; int *a=ar,*b=br,*c; c = a; a = b; b = c; cout << a[0] << a[1] << “\n”; 34 cout << b[0] << b[1];  12 This is more efficient than swapping each element Make sure you swap any dimension information too

Initializing Pointers Always initialize your pointers to valid memory locations A pointer initialized incorrectly is called a wild pointer Reading data from a wild pointer will return garbage Writing data to a wild pointer could overwrite program, data or operating system memory, e.g. int x, *p; x = 10; *p = x;

Common Problems With Pointers int x, *p; x = 10; p = x; cout << *p;  improper usage char a[10], b[10]; char *p1, *p2; p1 = a; p2 = b; if (p1 < p2) … two arrays will not be stored in any particular order

Dynamic Memory Allocation static arrays allocate memory at the beginning of program execution char s1[10], *p1; p1 = s1; dynamic memory is allocated during program execution #include <stdlib.h> #include <malloc.h> char *p1; p1 = (char *)malloc(10*sizeof(char));

Dynamic Arrays Dynamic arrays are useful because the size of the array can be determined at run-time The size of the array can be made just small enough for the application The memory allocated can also be freed for other functions to use when it is no longer needed It must be freed before the program is finished using the function free(p1)

New and Delete Operators C++ provides operators for dynamic memory allocation p = new type [size]; // new operator for allocation delete p; // delete operator for freeing memory int *p1; char *p2; p1 = new int; p2 = new char [5]; *p1 = 8; strcpy(p2,”ball”); cout << *p1 << p2; delete p1; delete p2;  8ball

Comparison with C Allocation Routines New and delete have the following advantages over malloc and free: Automatically computes the size needed in bytes Automatically returns a pointer of a specified type Provides support for C++ features related to operator overloading, initialization, constructors, and destructors Note that both malloc and new can take longer to run than static memory allocation for large blocks of memory

Arrays of Pointers Arrays of pointers can be declared like any other type char *pa[2]; char p1[]="hello", p2[]="goodbye"; pa[0] = p1; pa[1] = p2; cout << "\n" << pa[0] << "\t" << pa[1]; cout << "\n" << *(pa[0]+1) << "\t" << *(pa[1]+1); cout << "\n" << pa[0][1] << "\t" << pa[1][1]; This is useful for developing multidimensional arrays

memory address variable 900 1000 1008 1004 100E ‘h’ 1009 ‘e’ 100A ‘l’ 100C ‘o’ 100D ‘\0’ ‘g’ 100F pa pa[0] pa[1]

Pointers to Pointers Since pointers are stored in memory we can declare pointers to them as well char **pp; char *p, s[]="MJ"; p = s; // point to the string pp = &p; // get the pointer to p cout << "\n" << p << "\t" << *pp; cout << "\n" << *(p+1) << "\t" << *(*pp+1); A pointer to a pointer is the same as an array of pointers

memory address variable 1000 1004 1001 1002 1003 1008 1005 1006 1007 1009 ‘J’ pp p , s *pp **pp

memory address variable 900 1000 1008 1004 100E ‘h’ 1009 ‘e’ 100A ‘l’ 100C ‘o’ 100D ‘\0’ ‘g’ 100F pp pa pp[0] pa[0] pp[1] pa[1]

Why Pointers ? Allows functions to modify their arguments Supports dynamic memory allocation functions Improves the efficiency of certain routines (swapping, …) Allows communication to other functions, programs, and hardware using shared memory and memory transfers Useful for passing functions to other functions Important for developing multidimensional dynamic arrays

Key Things to Remember A 1D array is the same as a pointer to the first element double s[2] = { 1.0, 2.0 }, *p; p = s; p[1] == s[1] == *(p+1) == *(s+1) == 2 We can dynamically allocate a 1D array using new double *p; int n = 10; p = new double [n]; p[1] = 1.0; delete p; We can also have a 1D array of pointers which is the same as a pointer to a pointer double *pa[10], **pp; pp = pa;

More on Static Multidimensional Arrays 2D static arrays are stored in a row by row format int a[2][3] = {1,2,3,4,5,6};  a = int *r0, *r1, i; r0 = a[0]; r1 = a[1]; for(i=0;i<6;i++) cout << r0[i] << “ ”;  1 2 3 4 5 6 for(i=0;i<3;i++) cout << r1[i] << “ ”;  4 5 6 1 2 3 4 5 6

Limitations of Static Multidimensional Arrays They cannot be passed to general functions since arrays of different sizes cannot be used interchangeably int a[4][4]; a[0][0] = 1; f1(a);  can’t convert int[4][4] to int[][3] int fun1(int a[3][3]) { return a[0][0]; } They also cannot be converted to a pointer to pointer Only use static multidimensional arrays for simple applications where the size is always fixed

Multidimensional Dynamic Arrays One way to construct a 2D dynamic array: First allocate a 1D dynamic array of pointers This will have a pointer to pointer type Then initialize each array element to a 1D dynamic array that stores each row  done ! Dynamic arrays of any size can be passed to a function since they all have pointer to pointer type

Row Swapping Rows of a dynamic array can be efficiently swapped by swapping the pointers to the rows int *pa[2], **m, r0[] = {1,2}, r1[] = {3,4}, *d; m = pa; m[0] = r0; m[1] = r1; cout << m[0][0] << m[0][1] << “\n” << m[1][0] << m[1][1]; d = m[0]; m[0] = m[1]; m[1] = d; cout << “\n\n”;

Pointers to Objects class string_class { public: char str[100]; void strcpy2(char *s); }; void string_class::strcpy2(char *s) { strcpy(str,s); } string_class s1, *sp; s1.strcpy2(“ya”); sp = &s1; cout << s1.str << (*sp).str << sp->str;  yayaya

Arrays of Objects string_class s1[3], *sp; s1[0].strcpy2(“one”); s1[1].strcpy2(“two”); s1[2].strcpy2(“three”); sp = s1; cout << sp->str << “\n“; sp++;  one cout << sp->str << “\n“; sp++; two cout << sp->str << “\n”; three cout << sp[0].str; three

Dynamically Allocated Objects string_class *sp1,*sp2; sp1 = new string_class(“apple”); // initialize object to “apple” sp2 = new string_class [3]; // can’t initialize dynamic arrays sp2[0].strcpy2(“peach”); cout << sp1->str << “ “ << sp2[0].str << “\n”; delete sp1; delete [3] sp2; // need to put [size] first for dyn object arrays string_class::string_class(char *s) { strcpy(str,s); cout << “construct\n”; } // for sp1 string_class::string_class() { cout << “c0\n”; } // for sp2