Setting scale to a negative number with BigDecimal

Question

Does it ever make sense to set the scale to a negative number like this?

BigDecimal result;
    .
    .
    .
result = result.setScale(-1, RoundingMode.HALF_UP)

What happens if you do?

Answer 1

Yes, it makes sense to set the scale to a negative number in certain situations. That just means that the number is rounded to the nearest 10 to the -scale number, or just 10 in this case.

BigDecimal bd = new BigDecimal(1094);
bd = bd.setScale(-1, RoundingMode.HALF_UP);
System.out.println(bd);
System.out.println(bd.doubleValue());

Prints

1.09E+3
1090.0

The 1.09E+3 is equivalent to the double value 1090.0.

This is useful when the scientific measurement for a quantity has less significant digits than the total number of digits necessary to represent the number normally. The precision of such a number is less than the size of the number.

If Earth's distance from the Sun is given as 92,960,000 miles, and if this value is only 5 significant digits, then that is equivalent to

BigDecimal bd = new BigDecimal(92960000.0);
bd = bd.setScale(-3, RoundingMode.HALF_UP);// 3 insignificant/imprecise "0" digits
System.out.println(bd);

Which prints

9.2960E+7

The number is rounded to the nearest 1,000.