Addition in Java bits, bytes and shift

Question

Hi,
I have a byte array consisting of 8 bit values in each cell.
When I do an addition of two cells - which gives me a 16 bit result I use the following:
Integer a = (int) (arr[value] & 0xff);
a = a << 8;
Integer b = (int) (arr[secondValue] & 0xff);
Integer c = a + b;This seems to work fine. My question is how would I add 3 (24 bit result) or 4 (32 bit result) cells.
Would the following work: 24 bit result
Integer a = (int) (arr[value] & 0xff);
a = a << 16;
Integer b = (int) (arr[secondValue] & 0xff);
Integer c = (int) (arr[thirdValue] & 0xff);
Integer d = a + b + c;I am not sure if I have got the shift (<<16) usage correct or if it should be used in order to obtain the variables b or c.
All help appreciated.

Answer 1

khublaikhan wrote:
Just to confirm for 32 bit it would be:
// 32-bit
int a = (int)(arr[value] & 0xff);
int b = (int)(arr[secondValue] & 0xff);
int c = (int)(arr[thirdValue] & 0xff);
int d = (int)(arr[fourthValue] & 0xff);
int e = (a<<24) + (b<<16) + (c<<8) + d;
Integer eInt = new Integer(e);
Actually, the more I think about it, you may need to use longs instead of ints for 32-bit (not 16- or 24-bit though). It depends on what your data actually is. If you're expecting your 32-bit number to be an unsigned integer, then you'd better go with long. This is because in Java, ints utilize two's complement, so if the high-order byte you read in for a 32-bit value is very large, the resulting int will be interpreted by Java as a negative number instead of the large positive value you expect.
I'm probably not being very clear. Check out http://en.wikipedia.org/wiki/Two's_complement for the details.
In other words, if you wanted to get 32-bit values in this way, and the computed values are expected to be non-negative integers, use this:
// 32-bit
int a = (int)(arr[value] & 0xff);
int b = (int)(arr[secondValue] & 0xff);
int c = (int)(arr[thirdValue] & 0xff);
int d = (int)(arr[fourthValue] & 0xff);
long e = (a<<24) + (b<<16) + (c<<8) + d;
Long l = new Long(e);If you're actually reading in a 32-bit, two's complement value, then keep using int like you originally posted.