Hi Friends, I have some doubt with this ImplicitExplicitCast thing example on javaranch journal:
Here,I don't understand how come we lose our precious sign-bit after chopping off first 24-bit.The 8th bit--won't that be our sign-bit??If this,is wrong what should be the correct answer then -7?? Thanks in Advance.
The statement b1 >>>= 1; is a compound statement that is equivalent to b1 = (byte)(b1 >>> 1);
The part in the parentheses shows a byte variable getting bitshifted to the right with the unsigned right shift operator. This means that b1 will be promoted to an int so its bit pattern will be: 11111111111111111111111111110011 which is -13.
Then it will be shifted right one place, with a leading zero added on to the left side, because that's how >>> works: 01111111111111111111111111111001 which is positive 2147483641 because the 0 on the left end is the sign bit for this 32-bit int value. The number is positive because the leftmost bit is a 0.
So (b1 >>> 1) evalutes to 2147483641.
The statement can now be rewritten: b1 = (byte)(2147483641);
The next step is to cast this value explicitly to a byte. Since its current bit pattern is 01111111111111111111111111111001 casting it to a byte will discard the high-order 24 bits (the ones on the left). Here's that value with x's for the bits that will be discarded: xxxxxxxxxxxxxxxxxxxxxxxx11111001 After discarding those bits, the resulting byte has the bit pattern 11111001 which is now negative because the leftmost bit is 1. This value is -7.
The statement can now be rewritten: b1 = -7;
If b1 had been an int or long instead of a byte, then you could assume that the result of b1 >>>= 1; would be positive, since >>> puts a zero on the left. However, because of the cast to byte, there's no certainty that the result will be positive or negative until you work through the problem, since the sign bit of the int gets discarded, and whatever bit happens to be in the 8th place from the right will be the sign bit of the byte that results from the cast to byte. In this case that bit is a 1, so the byte is negative.
[ May 05, 2005: Message edited by: Joe Sanowitz ] [ May 05, 2005: Message edited by: Joe Sanowitz ]
SCJA 1.0 (98%), SCJP 1.4 (98%)
Joined: Jan 16, 2005
Thanks Animesh and Joe, Animesh i am trying to understand what Joe is trying to explain to me.Thanks Joe,after reading your post all my doubts got cleared....the only mistake i was making was i was taking the "unsigned right shift operator"...as a "signed right shift operator"..and then wondering what the author is tryin to prove. Anywas,thanks for your help!!!