網頁

2021年3月28日 星期日

C 語言新手十誡

C 語言新手十誡(The Ten Commandments for Newbie C Programmers)

  1. 本篇旨在提醒新手,避免初學常犯的錯誤(其實老手也常犯:-Q)。但不能取代完整的學習,請自己好好研讀一兩本 C 語言的好書,並多多實作練習。
  2. 強烈建議新手先看過此文再發問,你的問題極可能此文已經提出並解答了。
  3. 以下所舉的錯誤例子如果在你的電腦上印出和正確例子相同的結果,那只是不足為恃的一時僥倖。
  4. 不守十誡者,輕則執行結果的輸出數據錯誤,或是程式當掉,重則引爆核彈、毀滅地球(如果你的 C 程式是用來控制核彈發射器的話)。

  • 1. 不可以使用尚未給予適當初值的變數
  • 2. 不能存取超過陣列既定範圍的空間
  • 3. 不可以提取不知指向何方的指標
  • 4. 不要試圖用 char* 去更改一個"字串常數"
  • 5. 不能在函式中回傳一個指向區域性自動變數的指標
  • 6. 不可以只做 malloc(), 而不做相應的 free()
  • 7. 在數值運算、賦值或比較中不可以隨意混用不同型別的數值
  • 8. 在一個運算式中,不能對一個基本型態的變數修改其值超過一次以上
  • 9. 在 Macro 定義中, 務必為它的參數個別加上括號
  • A(10). 不可以在 stack 設置過大的變數
  • B(11). 使用浮點數精確度造成的誤差問題
  • C(12). 不要猜想二維陣列可以用 pointer to pointer 來傳遞
  • D(13). 函式內 new 出來的空間記得要讓主程式的指標接住

01. 你不可以使用尚未給予適當初值的變數
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
    錯誤例子:
    int accumulate(int max)    /* 從 1 累加到 max,傳回結果 */
    {
        int sum;    /* 未給予初值的區域變數,其內容值是垃圾 */
        int num;
        for (num = 1; num <= max; num++) {  sum += num;  }
        return sum;
    }

    正確例子:
    int accumulate(int max)
    {
        int sum = 0;    /* 正確的賦予適當的初值 */
        int num;
        for (num = 1; num <= max; num++) {  sum += num;  }
        return sum;
    }

02. 你不可以存取超過陣列既定範圍的空間
1
2
3
4
5
6
7
8
9
    錯誤例子:
    int str[5];
    int i;
    for (i = 0 ; i <= 5 ; i++) str[i] = i;

    正確例子:
    int str[5];
    int i;
    for (i = 0; i < 5; i++) str[i] = i;
  • 說明:宣告陣列時,所給的陣列元素個數值如果是 N, 那麼我們在後面
  • 透過 [索引值] 存取其元素時,所能使用的索引值範圍是從 0 到 N-1
  • C/C++ 為了執行效率,並不會自動檢查陣列索引值是否超過陣列邊界,我們要自己來確保不會越界。一旦越界,操作的不再是合法的空間,將導致無法預期的後果。

03. 你不可以提取(dereference)不知指向何方的指標(包含 null 指標)。
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
    錯誤例子:
    char *pc1;      /* 未給予初值,不知指向何方 */
    char *pc2 = 0;  /* pc2 起始化為 null pointer */
    *pc1 = 'a';     /* 將 'a' 寫到不知何方,錯誤 */
    *pc2 = 'b';     /* 將 'b' 寫到「位址0」,錯誤 */

    正確例子:
    char c;          /* c 的內容尚未起始化 */
    char *pc1 = &c;  /* pc1 指向字元變數 c */
    *pc1 = 'a';      /* c 的內容變為 'a' */

    /* 動態分配 10 個 char(其值未定),並將第一個char的位址賦值給 pc2 */
    char *pc2 = (char *) malloc(10);
    pc2[0] = 'b';    /* 動態配置來的第 0 個字元,內容變為 'b'
    free(pc2);
  • 說明:指標變數必需先指向某個可以合法操作的空間,才能進行操作。
    ( 使用者記得要檢查 malloc 回傳是否為 NULL,礙於篇幅本文假定使用上皆合法,也有正確歸還記憶體 )
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
    錯誤例子:
    char *name;   /* name 尚未指向有效的空間 */
    printf("Your name, please: ");
    gets(name);   /* 您確定要寫入的那塊空間合法嗎??? */
    printf("Hello, %s\n", name);

    正確例子:
    /* 如果編譯期就能決定字串的最大空間,那就不要宣告成 char* 改用 char[] */

    char name[21];   /* 可讀入字串最長 20 個字元,保留一格空間放 '\0' */
    printf("Your name, please: ");
    gets(name);
    printf("Hello, %s\n", name);
正確例子(2):
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
    /* 若是在執行時期才能決定字串的最大空間,則需利用 malloc() 函式來動態分配空間 */

    size_t length;
    char *name;
    printf("請輸入字串的最大長度(含null字元): ");
    scanf("%u", &length);

    name = (char *)malloc(length);
    printf("Your name, please: ");
    scanf("%s", name);
    printf("Hello, %s\n", name);

    /* 最後記得 free() 掉 malloc() 所分配的空間 */
    free(name);

04. 你不可以試圖用 char* 去更改一個"字串常數"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
    錯誤例子:
    char* pc = "john";   /* pc 現在指著一個字串常數 */
    *pc = 'J';   /* 但是 pc 沒有權利去更改這個常數! */



    正確例子:
    char pc[] = "john";  /* pc 現在是個合法的陣列,裡面住著字串 john */
                         /* 也就是 pc[0]='j', pc[1]='o', pc[2]='h',                                              pc[3]='n', pc[4]='\0'  */
    *pc = 'J';
    pc[2] = 'H';
  • 說明:字串常數的內容是"唯讀"的。您有使用權,但是沒有更改的權利。
  • 若您希望使用可以更改的字串,那您應該將其放在合法空間
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
    錯誤例子:
    char *s1 = "Hello, ";
    char *s2 = "world!";
    /* strcat() 不會另行配置空間,只會將資料附加到 s1 所指唯讀字串的後面,       造成寫入到程式無權碰觸的記憶體空間 */
    strcat(s1, s2);



    正確例子(2)
    /* s1 宣告成陣列,並保留足夠空間存放後續要附加的內容 */
    char s1[20] = "Hello, ";
    char *s2 = "world!";
    /* 因為 strcat() 的返回值等於第一個參數值,所以 s3 就不需要了 */
    strcat(s1, s2);

05. 你不可以在函式中回傳一個指向區域性自動變數的指標。否則,會得到垃圾值
[感謝 gocpp 網友提供程式例子]
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
    錯誤例子:
    char *getstr(char *name)
    {
        char buf[30] = "hello, "; /*將字串常數"hello, "的內容複製到buf陣列*/
        strcat(buf, name);
        return buf;
    }

    說明:區域性自動變數,將會在離開該區域時(本例中就是從getstr函式返回時)
    被消滅,因此呼叫端得到的指標所指的字串內容就失效了。

    正確例子:
    void getstr(char buf[], int buflen, char const *name)
    {
        char const s[] = "hello, ";
        strcpy(buf, s);
        strcat(buf, name);
    }

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
    正確例子:
    int* foo()
    {
        int* pInteger = (int*) malloc( 10*sizeof(int) );
        return pInteger;
    }

    int main()
    {
        int* pFromfoo = foo();
    }
  • 說明:上例雖然回傳了函式中的指標,但由於指標內容所指的位址並非區域變數,
  • 而是用動態的方式抓取而得,換句話說這塊空間是長在 heap 而非 stack
  • 又因 heap 空間並不會自動回收,因此這塊空間在離開函式後,依然有效
  • (但是這個例子可能會因為 programmer 的疏忽,忘記 free 而造成memory leak)
[針對字串操作,C++提供了更方便安全更直觀的 string class, 能用就盡量用]


1
2
3
4
5
6
7
8
9
    正確例子:

    #include <string>    /* 並非 #include <cstring> */
    using std::string;

    string getstr(string const &name)
    {
        return string("hello, ") += name;
    }

06. 你不可以只做 malloc(), 而不做相應的 free(). 否則會造成記憶體漏失
  • 但若不是用 malloc() 所得到的記憶體,則不可以 free()。
  • 已經 free()了所指記憶體的指標,在它指向另一塊有效的動態分配得來的空間之前,不可以再被 free(),也不可以提取(dereference)這個指標。
  • [C++] 你不可以只做 new, 而不做相應的 delete
  • 注:new 與 delete 對應,new[] 與 delete[] 對應,不可混用
    (切記,做了幾次 new,就必須做幾次 delete)
  • 小技巧: 可在 delete 之後將指標指到 0,由於 delete 本身會先做檢查,因此可以避免掉多次 delete 的錯誤
1
2
3
4
5
   正確例子:
   int *ptr = new int(99);
   delete ptr;
   ptr = NULL;
   delete ptr;   /* delete 只會處理指向非 NULL 的指標 */

07. 你不可以在數值運算、賦值或比較中隨意混用不同型別的數值,而不謹慎考
  • 慮數值型別轉換可能帶來的「意外驚喜」(錯愕)。必須隨時注意數值運算的結果,其範圍是否會超出變數的型別
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
    錯誤例子:
    unsigned int sum = 2000000000 + 2000000000;  /* 超出 int 存放範圍 */
    unsigned int sum = (unsigned int) (2000000000 + 2000000000);
    double f = 10 / 3;


    正確例子:
    /* 全部都用 unsigned int, 注意數字後面的 u, 大寫 U 也成 */
    unsigned int sum = 2000000000u + 2000000000u;

    /* 或是用顯式的轉型 */
    unsigned int sum = (unsigned int) 2000000000 + 2000000000;

    double f = 10.0 / 3.0;


1
2
3
4
    錯誤例子:
    unsigned int a = 0;
    int b[10];
    for(int i = 9 ; i >= a ; i--) {  b[i] = 0;  }
  • 說明:由於 int 與 unsigned 共同運算的時候,會提升 int 為 unsigned,因此迴圈條件永遠滿足,與預期行為不符
1
2
3
4
5
6
    錯誤例子:                                 (感謝 sekya 網友提供)
    unsigned char a = 0x80;   /* no problem */
    char b = 0x80;    /* implementation-defined result */
    if( b == 0x80 ) {        /* 不一定恒真 */
        printf( "b ok\n" );
    }
  • 說明:語言並未規定 char 天生為 unsigned 或 signed,因此將 0x80 放入char 型態的變數,將會視各家編譯器不同作法而有不同結果

08. 你不可以在一個運算式(expression)中,對一個基本型態的變數修改其值超過一次以上。否則,將導致未定義的行為(undefined behavior)
1
2
3
4
5
6
7
8
    錯誤例子:
    int i = 7;
    int j = ++i + i++;

    正確例子:
    int i = 7;
    int j = ++i;
    j += i++;
  • 你也不可以在一個運算式(expression)中,對一個基本型態的變數修改其值,而且還在同一個式子的其他地方為了其他目的而存取該變數的值。(其他目的,是指不是為了計算這個變數的新值的目的)。否則,將導致未定義的行為。
1
2
    錯誤例子:
    x = x++;

1
 2
 3
 4
 5
 6
 7
 8
 9
10
    錯誤例子:
    int arr[5];
    int i = 0;
    arr[i] = i++;

    正確例子:
    int arr[5];
    int i = 0;
    arr[i] = i;
    i++;

1
2
3
4
5
6
7
8
    錯誤例子:
    int i = 10;
    cout << i << "==" << i++;

    正確例子:
    int i = 10;
    cout << i << "==";
    cout << i++;


 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
    錯誤例子:
    int Integer=10;
    printf( "%d %d %d", Integer++, Integer++, Integer++ );


    錯誤例子:
    void foo(int a, int b) { ... }
    int main() {
        int i=0;
        foo(i++, i++);
    }

  • 說明: C/C++ 並沒有強制規定參數會由哪個方向開始處理(不像Java是由左到右),因此可能會造成與預期不符的情況

09. 在 Macro 定義中, 務必為它的參數個別加上括號

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
    錯誤例子:
    #include <stdio.h>
    #define SQUARE(x)    (x * x)
    int main()
    {
        printf("%d\n", SQUARE(10-5));
        return 0;
    }

    正確例子:
    #include <stdio.h>
    #define SQUARE(x)    ((x) * (x))
    int main()
    {
        printf("%d\n", SQUARE(10-5));
        return 0;
    }
  • 說明:如果是用 C++, 請多多利用 inline function 來取代上述的 macro,以免除 macro 定義的種種危險性。如:
    inline int square(int x) { return x * x; }
  • macro 定義出的「偽函式」至少缺乏下列數項函式本有的能力:
    • (1) 無法進行參數型別的檢查。
    • (2) 無法遞迴呼叫。
    • (3) 無法用 & 加在 macro name 之前,取得函式位址。
    • (4) 呼叫時往往不能使用具有 side effect 的引數。例如:
1
2
3
4
5
6
7
8
    錯誤例子:                                      (感謝 yaca 網友提供)
    #define MACRO(x)     (((x) * (x)) - ((x) * (x)))
    int main()
    {
        int x = 3;
        printf("%d\n", MACRO(++x));
        return 0;
    }
    • 備註:副作用(side effect)
    • 表达式有两种功能:每个表达式都产生一个值( value ),同时可能包含副作用( side effect )。副作用是指改变了某些变量的值。
      • 1:20 //这个表达式的值是20;它没有副作用,因为它没有改变任何变量的值。
      • 2:x=5 // 这个表达式的值是5;它有一个副作用,因为它改变了变量x的值。
      • 3:x=y++ // 这个表达示有两个副作用,因为改变了两个变量的值。
      • 4:x=x++ // 这个表达式也有两个副作用,因为变量x的值发生了两次改变。

10. 不可在 stack 設置過大的變數,否則會造成 stack overflow
(感謝 VictorTom 版友幫忙)
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
    錯誤例子:
    int array[10000000];       // 僅舉例說明

   /* 說明:由於編譯器會自行決定 stack 的上限,某些預設是數 KB 或數十 KB
    當變數所需的空間過大時,很容易造成 stack overflow,程式亦隨之
    當掉,若真正需要如此大的空間,那麼建議配置在 heap 上,或是採用
    static / globla variable,亦或是改變編譯器的設定
    使用 heap 時,雖然整個 process 可用的空間是有限的,但採用動態抓取
    的方式,new 無法配置時會丟出 std::bad_alloc 例外,malloc 無法配置
    時會回傳 null,不會影響到正常使用下的程式功能    */
    正確例子:
    int *array = (int*) malloc( 10000000*sizeof(int) );
  • 說明:由於此時 stack 上只需配置一個 int* 的空間,可避免 stack overflow
  • 更多說明請參考精華區 z-10-13

11. 使用浮點數千萬要注意精確度所造成的誤差問題

根據 IEEE 754 的規範,又電腦中是用有限的二進位儲存數字,因此常有可
能因為精確度而造成誤差,例如加減乘除,等號大小判斷,分配律等數學上
常用到的操作,很有可能因此而出錯(不成立)

更詳細的說明可以參考精華區 z-8-11
或參考冼鏡光老師所發表的一文 "使用浮點數最最基本的觀念"
http://blog.dcview.com/article.php?a=VmgBZFE5AzI%3D

12. 不要猜想二維陣列可以用 pointer to pointer 來傳遞
(感謝 loveme00835 legnaleurc 版友的幫忙)

首先必須有個觀念,C 語言中陣列是無法直接拿來傳遞的!
不過這時候會有人跳出來反駁:
1
2
3
4
    void pass1DArray( int array[] );

    int a[10];
    pass1DArray( a );   /* 可以合法編譯,而且執行結果正確!! */
事實上,編譯器會這麼看待
1
2
3
4
    void pass1DArray( int *array );

    int a[10];
    pass1DArray( &a[0] );
  • 我們可以順便看出來,array 變數本身可以 decay 成記憶體起頭的位置
  • 因此我們可以 int *p = a; 這種方式,拿指標去接陣列。
  • 也因為上述的例子,許多人以為那二維陣列是不是也可以改成 int **
1
2
3
4
5
6
7
    錯誤例子:
    void pass2DArray( int **array );

    int a[5][10];
    pass2DArray( a );
    /* 這時候編譯器就會報錯啦 */
    /* expected ‘int **’ but argument is of type ‘int (*)[10]’*/

  • 在一維陣列中,指標的移動操作,會剛好覆蓋到陣列的範圍,例如,宣告了一個 a[10],那我可以把 a 當成指標來操作 *a 至 *(a+9)
  • 因此我們可以得到一個概念,在操作的時候,可以 decay 成指標來使用,也就是我可以把一個陣列當成一個指標來使用 (again, 陣列!=指標)
  • 但是多維陣列中,無法如此使用,事實上這也很直觀,試圖拿一個pointer to pointer to int 來操作一個 int 二維陣列,這是不合理的!
  • 儘管我們無法將二維陣列直接 decay 成兩個指標,但是我們可以換個角度想,二維陣列可以看成 "外層大的一維陣列,每一維內層各又包含著一維陣列"
  • 如果想通了這一點,我們可以仿造之前的規則,把外層大的一維陣列 decay 成指標,該指標指向內層的一維陣列
1
2
3
4
    void pass2DArray( int (*array) [10] );  // array 是個指標,指向 int [10]

    int a[5][10];
    pass2DArray( a );

  • 這時候就很好理解了,函數 pass2DArray 內的 array[0] 會代表什麼呢?答案是它代表著 a[0] 外層的那一維陣列,裡面包含著內層 [0]~[9],也因此 array[0][2] 就會對應到 a[0][2],array[4][9] 對應到 a[4][9]
  • 結論就是,只有最外層的那一維陣列可以 decay 成指標,其他維陣列都要明確的指出陣列大小,這樣多維陣列的傳遞就不會有問題了
  • 也因為剛剛的例子,我們可以清楚的知道在傳遞陣列時,實際行為是在傳遞指標,也因此如果我們想用 sizeof 來求得陣列元素個數,那是不可行的
1
2
3
4
    錯誤例子:
    void print1DArraySize( int* arr ) {
        printf("%u", sizeof(arr)/sizeof(arr[0])); /* sizeof(arr) 只是 */
    }      
  • 受此限制,我們必須手動傳入大小
1
    void print1DArraySize( int* arr, size_t arrSize );
  • 註:typedef unsigned int size_t 與平台無關的,表示0-MAXINT的無號整數
  • C++ 提供 reference 的機制,使得我們不需再這麼麻煩,可以直接傳遞陣列的 reference 給函數,大小也可以直接求出
1
2
3
4
    正確例子:
    void print1DArraySize( int (&array)[10] ) {    // 傳遞 reference
        cout << sizeof(array) / sizeof(int);   // 正確取得陣列元素個數
    }

13. 函式內 new 出來的空間記得要讓主程式的指標接住
  • 對指標不熟悉的使用者會以為以下的程式碼是符合預期的
1
2
3
4
5
6
7
8
    void newArray(int* local, int size) {
        local = (int*) malloc( size * sizeof(int) );
    }

    int main() {
        int* ptr;
        newArray(ptr, 10);
    }
  • 接著就會找了很久的 bug,最後仍然搞不懂為什麼 ptr 沒有指向剛剛拿到的合法空間,讓我們再回顧一次
  • 1. int* ptr;ptr 指向未知的空間
  • 2. 呼叫函式 newArray,ptr和local指向未知的空間
  • 3. malloc 取得合法空間ptr指向未知的空間 ,為local
  • 4. 離開函ptr 只向未知空間
  • 也許有人會想問,指標不是傳址嗎?精確來講,指標也是傳值,只不過該值是一個位址 (ex: 0xfefefefe)
    • local 接到了 ptr 指向的那個位置,接著函式內 local 要到了新的位置
    • 但是 ptr 指向的位置還是沒變的,因此離開函式後就好像事什麼都沒發生
      ( 嚴格說起來還發生了 memory leak )
  • 以下是一種解決辦法
1
2
3
4
5
6
7
8
    int* createNewArray(int size) {
        return (int*) malloc( size * sizeof(int) );
    }

    int main() {
        int* ptr;
        ptr = createNewArray(10);
    }
  • 改成這樣亦可 ( 為何用 int** 就可以?想想他會傳什麼過去給local )
1
2
3
4
5
6
7
8
    void createNewArray(int** local, int size) {
        *local = (int*) malloc( size * sizeof(int) );
    }

    int main() {
        int *ptr;
        createNewArray(&ptr, 10);
    }
  • 如果是 C++,別忘了可以善用 Reference
1
2
3
    void newArray(int*& local, int size) {
        local = new int[size];
    }

 

資料來源:

https://www.ptt.cc/bbs/C_and_CPP/M.1268491365.A.EF7.html


2021年3月15日 星期一

Unix System Programming (1092)-exer2

http://erdos.csie.ncnu.edu.tw/~klim/unix-p/usp-1092.html

exercise 2: probing argv[] and environ[]

    Write a program which prints the values of argc, argv[], and environ[]. Also their addresses are printed. Based on the printed information, you can figure out the memory layout of argc, argv[], and environ[].
For example:

$ ./a.out klim is not milk
 argc is 5
 argv[0]: ./a.out
 argv[1]: klim
 argv[2]: is
 argv[3]: not
 argv[4]: milk
 environ[0]: CPLUS_INCLUDE_PATH=/usr/lib64/qt/include
 environ[1]: NNTPSERVER=news.ncnu.edu.tw
 .......
 environ[33]: INFOPATH=/usr/info:/usr/share/texmf/info
 environ[34]: G_BROKEN_FILENAMES=1
 environ[35]: _=./a.out
 environ[36]: OLDPWD=/home/klim/course/usp-1092

 argc      is resided at 0x7ffdddad11fc
 argv      is resided at 0x7ffdddad11f0
 environ   is resided at 0x601040
 argv[]    is resided at 0x7ffdddad12f8
 environ[] is resided at 0x7ffdddad1328
 value of argv[ 0] is 0x7ffdddad24c3
 value of argv[ 1] is 0x7ffdddad24cb
 value of argv[ 2] is 0x7ffdddad24d0
 value of argv[ 3] is 0x7ffdddad24d3
 value of argv[ 4] is 0x7ffdddad24d7
 value of argv[ 5] is (nil)
 value of env [ 0] is 0x7ffdddad24dc
 value of env [ 1] is 0x7ffdddad2505
 value of env [ 2] is 0x7ffdddad2521
 ..........
 value of env [35] is 0x7ffdddad2fc4
 value of env [36] is 0x7ffdddad2fce
 $


    If there are too many environment strings to display, you can try to use ``env'' to get a cleaner environment. For example:

$ env -i A=1 hello=ohio ./a.out see you again
argc is 4
argv[0]: ./a.out
argv[1]: see
argv[2]: you
argv[3]: again
environ[0]: A=1
environ[1]: hello=ohio

argc      is resided at 0x7fff4b9083dc
argv      is resided at 0x7fff4b9083d0
environ   is resided at 0x601040
argv[]    is resided at 0x7fff4b9084d8
environ[] is resided at 0x7fff4b908500
value of argv[ 0] is 0x7fff4b908fcb
value of argv[ 1] is 0x7fff4b908fd3
value of argv[ 2] is 0x7fff4b908fd7
value of argv[ 3] is 0x7fff4b908fdb
value of argv[ 4] is (nil)
value of env [ 0] is 0x7fff4b908fe1
value of env [ 1] is 0x7fff4b908fe5
$

My program:
https://ideone.com/xz3wfv


2021年3月3日 星期三

Unix System Programming (1092)-exer1

exercise 1: a simple C program
  • Write a C program which reads strings from stdin and if a string is consisted of digits, it is treated as a number. Sum all the numbers and write sum to stdout.
  • The reading is terminated by end of file, that is, a ctrl-D.
  • The length of each string does not exceed 16.
  • You may assume the numbers and the sum can be handled with int.
  • For converting a string to a number, you may consult strtol().
  • For example:
    $ gcc ex1-sum-stdin.c 
    $ ./a.out 
    I have 1000
    You have 200
    I give you 50
    Now You have 250.   <---- ctrl-D pressed
    1250                <---- only 1000, 200, 50 are summed
    $ ./a.out 
    100 + 23 = 123
    $100 - 23 = 77.     <---- ctrl-D pressed
    269                 <---- 100, 23, 123, 23 are summed
     
    My program:
    https://ideone.com/V3fc8b
    https://ideone.com/hFVjyY