shift operators.. again

[Andres Gonzalez]

just when I'm starting to feel confortable with shifting, there's my friend Dan with his excellent questions... :roll:

class K {
  public static void main (String[] s) {
    byte b = 127;
    b <<= 2;
    System.out.println(b);
  }
}

What is the result of attempting to compile and run the above program?
a. Prints: -4
b. Prints: -3
c. Prints: -2
d. Prints: 0
e. Prints: 1
f. Prints: 127
g. Prints: 508
h. Runtime Exception
i. Compiler Error
j. None of the Above
It prints -4. I reached til 508, but now this 508 int value has to be chopped off to fit in a byte, right? How do you guys do that?..

[Alexan Kahkejian]

Hi Andres
You cannot consider this way, you must consider it as 8 bit number and the digits that go out from the left are lost. Then if the leftmost bit is 1 that means it is a negative number so you must calculate the two's complement of the number.

 
decimal   |hex  |binary
---------------------------------
1277F0111 1111// original
254FE1111 1110// afer one place shift
252FC1111 1100// after two place shift
3030000 0011// 1'st complement
4040000 0100// 2's  complement
 

Hope that Helps
[ June 29, 2003: Message edited by: Alexan Kahkejian ]

[Andres Gonzalez]

Thanks Alexan. What I usually do is the following:
x >> n = x / 2^n
x << n = x * 2^n
and that's it. Using this formula I got 508. So from this number, I didn't know how to "make it fit" into a byte.

[Marlene Miller]

b <<= 2 is the same as b = (byte)(b << 2)
except that b is evaluated only once.
Let�s evaluate this expression: (byte)(b << 2)
First the left operand b is promoted to type int.
01111111 is promoted to 0000 0000 0000 0000 0000 0000 0111 1111
Then the left shift operation << is performed.
0000 0000 0000 0000 0000 0001 1111 1100 (== 508)
Then the result is converted from type int to type byte by
discarding all but the 8 lowest order bits.
11111100 (== -4)
[ June 29, 2003: Message edited by: Marlene Miller ]

[Marlene Miller]

b = b << 2; //compiler error, must be cast to type byte
b <<= 2; //implicit cast to type byte
In both of these examples, the value of b is promoted to an int before << is performed.

[Prasad Kuppa]

I think we need to do like this.. keep subtracting 128 from 508 till you get a nearest value to -128 in range (-128 & 127). Can anyone confirm whether this is correct way?
Thanks,
Prasad