Question

unsigned value( unsigned n, unsigned low, unsigned high ){
    if( !(low <= high && high <= 32) )  exit(EXIT_FAILURE);
    if( low == 0 && high == 32 )
        return n;
    else
        return n >> low & (1U << (high-low)) - 1;
}

imagine we had the following as n=11100011, low=2, and high=7.

by the time we reached the return statement I’d have this

00111000 & (00100000 – 00000001)

this would be

00111000 & 00011111

which would equal

00011000

but thats not right is it? Thats 24 while I want 00111000 which is 56

what am I doing wrong here? where did I screw up?

Answer 1

You need to left-shift one more position before you subtract 1. That will extend the string of 1 bits one more position to the left, which is where the high bit resides.

return (n >> low) & ((1U << (high-low+1)) - 1);

Imagine the extreme case, where low = 0 and high = 32. Ignore the overflow, which is an artifact of the size of long, and do the calculation:

(n >> 0) & ((1U << 33) - 1)

The term on the left is just n, and the term on the right is a string of 32 1 bits.