but then, aint there little point in doing this? i might as well use the whole integer then, instead of masking it off to simulate an unsigned 8-bit, if theres no gain in memory use, right?

Zahlman

257

Zahlman

Nov 2, 2004 at 4:26 AM

In some cases it will be possible to keep the value stored in a normal-sized variable (as signed; so a value in the top half of the range will instead be stored as a negative value) and just do the masking when you need to do a calculation - then store the result back if needed. Example:

byte numberOfEggs = (byte)144; // actually -112// ...// Tell the user how many dozens of eggs there are.System.out.println("You have " + (numberOfEggs & 255) / 12 + "dozen eggs.");// ...// Oh noes! Teh haxorz haev st0len half the eggs!!!11numberOfEggs = (byte)((numberOfEggs & 255) / 2);// The byte cast is necessary even though the value is < 128,// because Java promotes everything to an int when evaluating// the arithmetic expression.

Of course, optimization (according to the specific way in which you are using the variables) is left as an exercise ;)

UeberBobo

Author

UeberBobo

Nov 2, 2004 at 4:38 PM

zahlman: thanks that cool. could you explain how it works though?

to me it feels like (byte & 255) would just stay the same as it is...
say we take the number -236 (or whatever):

    10010100 &  11111111-------------    10010100

so why does makes java decide its positive? is it the typecast to integer or what is it?

Zahlman

257

Zahlman

Nov 2, 2004 at 9:21 PM

As I alluded to, all the intermediate calculations are done in integers. So if you need to use unsigned ints, you'll have to add some (long) casts; and if you need to use unsigned longs, you're SOL :)

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.