跳至主要内容

CH10 Self-defined Data Type

備註

本文為 2021-Fall 學期旁聽台大資管系孔令傑教授開授的 Programming Design 所記錄的課程筆記。課程內容程式碼可以參閱我的 Github repo: C++ Programming-Design-2021-Fall

struct​

struct definition​

We want to group different data types into a single type. In C, we do so by using struct (abbreviation of structure).

  • We do so when an item naturally consists of multiple attributes.
  • We do so to make the program easier to read and maintain. exp:
struct Point 
{
    int x;
    int y; 
};

Point vector(Point A, Point B)
// Point as parameters 
{
    Point vecXY;
    vecXY.x = B.x - A.x;
    vecXY.y = B.y - A.y;
    return vecXY; // return a Point 
} 

int main() {
    Point a = {0, 0}, b = {10, 20};
    Point vecAB = vector(a, b);
    cout << vecAB.x << " ";
    cout << vecAB.y << "\n";
    return 0; 
}
  • Declare variables with the self - defined type name.
  • Assign values to both attributes by grouping values by curly brackets.
  • Access attributes through the dot operator.
// 定義
struct struct_name 
{
    type1 field 1;
    type2 field 2;
    type3 field 3;
    // more fields 
};

// 宣告
struct_name variable_name;

// 存取
struct_variable.attribute_name

Partial assignments are allowed (with unassigned attributes set to 0).

Point A = {0, 0, -8}; 
Point B; 
B = {10, 20, 5}; 
C = {5, 0}; 
D = {2}

struct and functions​

You may pass a struct variable as an argument into a function.
You may return a struct variable from a function, too.
參考10_1.cpp

Memory allocation for struct​

參考10_2.cpp


typedef​

typedef is the abbreviation of “type definition”. It allows us to create a new data type from another data type.

typedef <old type> <new type>;

So we do not really create any new type. (只是幫old type取一個別名)

Example with typedef​

typedef double Dollar; // define Dollar as double D
ollar nt; // declare a variable as Dollar 
Dollar us; 
cin >> us; nt = us * 29; 
cout << nt << "\n";
  • The self-defined type can be used only in the block (if you declare it in any block). The same rule applies to struct.

typedef from struct​

We may combine typedef and struct.

// define Vector from Point 
typedef Point Vector;

Point a = {0, 0}; 
Point b = {10, 20}; 
Vector vecAB = vector(a, b);

Example: <ctime>​

很多C++的standard library 都會用typedef來提供函式功能，ex: function clock(), (defined in <ctime>) 回傳程式開始執行後經過了幾個system clock ticks 參考10_3.cpp

clock() returns a type clock_t variable (for the number of ticks).

  • clock_t is actually a long int. In <ctime>, there is a statement:

typedef long int clock_t;

Why clock_t?

假如某天C++ standard library 想要更改clock()回傳的變數型態，若原本回傳為long int，就會遇到舊版與新版回傳變數形態不同而所有程式碼都都改的窘境

To print out the number of seconds instead of ticks:

cout << static_cast<double>(eTime - sTime) / CLOCKS_PER_SEC << "\n";

使用者就算不知道clock_t是什麼變數型態也可以直接使用，library就算改了clock_t的old type也不會影響到使用者。


struct with member functions​

A member-function implementation​

We may redefine Point to include a member function:

  • distOri() is a member function of Point.
  • No argument is needed.
struct Point 
{
int x;
int y;
double distOri()
    {
        return sqrt(pow(x, 2) + pow(y, 2)); 
    }
};

int main() 
{
Point a = {3, 4};
cout << a.distOri();
return 0; 
}

One may define a member function outside the struct statement.
必須使用 scope resolution operator ::，並且把 struct 的名字加在前面，以區別出 member-function 與global-function。

struct Point 
{
    int x;
    int y;
    double distOri(); 
};

double Point::distOri() // scope resolution 
{                       // is required
    return sqrt(pow(x, 2) + pow(y, 2)); 
}

Two different perspectives​

  • As a global function: I want to create a machine outside a point.
  • As a member function: I want to attach an operation on a point.
    The second way also enhances modularity(讓每個模塊功能區分清楚，成為獨立自主的個體，也較容易維護)

參考10_4.cpp


Randomization​

rand()​

It “randomly” returns an integer between 0 and RAND_MAX (in <cstdlib>, typically 32767). Actually rand() returns a “pseudo-random” integer.

  • They just look like random numbers. But they are not really random.
  • There is a formula to produce each number.

參考10_5.cpp

srand()​

We use srand to determine the seed.

void srand(unsigned int);

  • A seed can be generated based on the input number.
  • The sequence will become different.

參考10_6.cpp

In many cases, we use time(nullptr) to be the argument of srand()

  • The function time(0), defined in <ctime>, returns the number of seconds that have past since 0:0:0, Jan, 1st, 1970.

time_t time(time_t* timer);

參考10_7.cpp

Random numbers in a range​

If you want to produce random numbers in a specific range, use %.

參考10_8.cpp


Reference​