Journal Article - The SCJP Tip Line - Bit Shifting

[Dirk Schreckmann]

The just-released June 2004 edition of The JavaRanch Journal includes an article by Corey McGlone, "The SCJP Tip Line - Bit Shifting".

Please use this thread to comment on and discuss the article.

[Billy Tobon]

Originally posted by Dirk Schreckmann:
The just-released June 2004 edition of The JavaRanch Journal includes an article by Corey McGlone, "The SCJP Tip Line - Bit Shifting".

Please use this thread to comment on and discuss the article.

Hi Dirk...
I think, there is a mistake in the article:
You wrote:

b2 = (byte)(b2 >>> 2);
b2, which was 00001101, becomes 00000110 (which is 6 in decimal)

But, Its must return 00000011 (which is 3 in decimal) or I'm confused?

Thanks...

[Xu Yani]

yes,it's true. I have compiled the source code. And the result should be 3.

[Michael J. Makunas]

Can anyone post any real-world examples of using bit-shifting? It's always been something I've found neat and interesting from a computer science standpoint but never had a good reason to use it.

[Marlene Miller]

Hi Michael,

>>Can anyone post any real-world examples of using bit-shifting?

An Internet address is a 32-bit number. The external representation is written as 4 unsigned bytes, ranging from 0 to 255, separated by periods.

java.sun.com/128.11.159.83
java.sun.com/209.249.116.141
java.sun.com/209.249.116.142
java.sun.com/209.249.116.143

In the source code of java.net.InetAddress, the Internet address is stored in a field of type int. When we call the toString() method, the 32-bit value is converted to 4 byte values, which are then converted to a String.

String hostName;
int address;
 
   public String toString() {
   return ((hostName != null) ? hostName : "") 
       + "/" + getHostAddress();
   }
 
   public String getHostAddress() {
   return numericToTextFormat(getAddress());
   }
 
   public byte[] getAddress() {
   byte[] addr = new byte[INADDRSZ];
 
   addr[0] = (byte) ((address >>> 24) & 0xFF);
   addr[1] = (byte) ((address >>> 16) & 0xFF);
   addr[2] = (byte) ((address >>> 8) & 0xFF);
   addr[3] = (byte) (address & 0xFF);
   return addr;
   }

----

A good source of examples where you will find bit shifting is sending data over the network. Data is written as a stream of bytes. 32-bit int values and 16-bit Unicode characters are broken up into bytes by shifting 8 bits at a time.

In paeleolithic times, programmers shifted 1 instead of multiplying by 2 because shifting takes few clock cycles than multiplying. In C code, you would see a lot of x << 1 all over a program.
[ June 19, 2004: Message edited by: Marlene Miller ]

[Marlene Miller]

Another example of bit shifting...

I guess there are 5 different class of Internet address. The 32-bit address is partitioned into fields.

Class A: 0, netid 7-bits, hosted 24-bits
Class B: 10, netid 14-bits, hostid 16-bit
Class C: 110, netid 21-bits, hosted 8-bits
Class D: 1110, multicast group ID 28-bits
Class E: 11110, reserved for future use 27-bits

To extract the values of the fields, you will probably do some shifting.

I guess there are other ways to look at the 32-bit Internet address. Here is some code from java.net.InetAddress.

/**
 * Utility routine to check if the multicast address has global scope.
 *
 * @return a <code>boolean</code> indicating if the address has 
 *         is a multicast address of global scope, false if it is not 
 *         of global scope or it is not a multicast address
 * @since 1.4
 */
public boolean isMCGlobal() {
   // 224.0.1.0 to 238.255.255.255
    byte[] byteAddr = getAddress();
    return ((byteAddr[0] & 0xff) >= 224 &&
            (byteAddr[0] & 0xff) <= 238 ) &&
            !((byteAddr[0] & 0xff) == 224 &&
               byteAddr[1] == 0 &&
               byteAddr[2] == 0);
}

----

If you want some more ideas, search the source code of the Java APIs, which comes with the Java SDK. My search found about 900 lines of code.