The Dotted Decimal Notation
An IP address is a 32-bit unsigned integer. It's usually written as
four numbers separated by dots, as in: 126.96.36.199. This format is
knows as the "dotted decimal notation". Applications use struct in_addr
to store the binary representation of an IP address. Struct in_addr is
declared in the header as follows:
unsigned char s_b1, s_b2, s_b3, s_b4;
unsigned short s_w1, s_w2;
unsigned long S_addr;
The struct contains a union that wraps three different views of the
same data variable: as four octets, as two unsigned short integers, or
as a single unsigned long integer.
in_ddr to String Conversion
To obtain a string representation of an in_addr object, use the
inet_ntoa() function that has the following prototype:
char * inet_ntoa(struct in_addr addr);
This function takes an in_addr object as an argument and returns a
pointer to a static string containing its dotted decimal form. You
should copy the result to a private buffer because each call to
inet_ntoa() will assign a new value to the string.
String to in_addr Conversion
The deprecated function inet_addr() performs the reverse, namely
converting a dotted decimal string to its binary representation.
inet_addr() has the following prototype:
unsigned long inet_addr(const char *cp);
This function returns the unsigned long representation of the dotted
decimal string cp. There are two problems with this function, though.
First, it returns unsigned long rather than an in_addr object. This
forces programmers to use ugly casts. The second problem is that you
have no way to distinguish between a return value of -1 (indicating an
error) or the address 255.255.255.255 since both these values have an
identical binary representation. You may still find occurrences of
inet_addr() in legacy code. However, in new applications you should use
int inet_aton(const char *cp, struct in_addr *paddr);
This function converts the string cp to an in_addr object and writes
the result to paddr. Note that unlike most library functions, inet_aton
() returns 0 to indicate an error, and a nonzero value to indicate
success. Next week I will show how to translate an IP address to a
hostname and vice versa.