byte[] for IP address storage????

[Bob Ruth]

In the InetAddress class there are several methods that retrieve IP addresses. The address is stored in a byte array, 4 bytes for a IP4 address.

How does that work?

I tried using getByAddress() where you pass both the string and the byte array. Problem is, a byte is a signed value so any attempt to stick a value > 127 in an address bytes draws a compiler error .... loss of precision. When I call getByName() on a known address string, the address that comes back in the byte array WILL print values over 127. Like.....241 shows up as 241.

Why would they have ever used a byte array to hold an 8 bit unsigned value? How is it that the API calls can store numbers > 127 into the byte array but I can't?

I have GOT to be missing something here.... I just don't know what it is.....

Help???

[Joe Ess]

Originally posted by Bob Ruth:
The address is stored in a byte array, 4 bytes for a IP4 address.

I'm looking at the source for InetAddress in JDK 1.6.0_05 and the internal storage is an int:

 
public
class InetAddress implements java.io.Serializable {
. . .
    /**
     * Holds a 32-bit IPv4 address.
     *
     * @serial
     */
    int address;   
. . .

There are getters and setters for 4-byte arrays.

Problem is, a byte is a signed value so any attempt to stick a value > 127 in an address bytes draws a compiler error .... loss of precision.

This is called a "narrowing conversion". You have to reassure the compiler that you know what you are doing and the data will fit in a byte with a cast:

    
  byte b[] = new byte[4];
  b[0] = (byte)190;
  . . .

[Bob Ruth]

Thank you, sir! I knew that I had to be missing something elementary and I just didn't think of that in the light of a conversion. Thanks!