why BigDecimal doesn't work as I expected?

Question

This is the code and the result in the comments:

I was thinking the numB2 should be 16.7500, numB3 = 16.7500, numB4=16.7 and numB5=16.75. But they are not as I expected. Why?

public void testBigDecimal()
{
    float num = 16.7499924f;

    int digitCount = 4;
    MathContext mContext = new MathContext(digitCount, RoundingMode.UP);
    BigDecimal numB = new BigDecimal(num, mContext);
    System.out.println("16.7499924f with MathContext(precision) of 4 =" + numB.floatValue());// 16.75
    numB.setScale(4, RoundingMode.UP);
    System.out
            .println("16.7499924f with MathContext(precision) of 4 and after setScale of 4 =" + numB.floatValue());// 16.75

    BigDecimal numB2 = new BigDecimal(num);
    System.out.println("16.7499924f without MathContext(precision =" + numB2.floatValue());// 16.749992
    numB2.setScale(4, RoundingMode.UP);
    System.out.println("16.7499924f without MathContext(precision) and after setScale of 4 =" + numB2.floatValue());// 16.749992

    BigDecimal numB3 = new BigDecimal(num);
    numB3.setScale(4, RoundingMode.UP);
    System.out.println("16.7499924f without MathContext(precision) but with setScale of 4 =" + numB3.floatValue());// 16.749992

    BigDecimal numB4 = new BigDecimal(num);
    numB4.setScale(1, RoundingMode.UP);
    System.out.println("16.7499924f without MathContext(precision) but with setScale of 1 =" + numB4.floatValue());// 16.749992

    BigDecimal numB5 = new BigDecimal(num);
    numB5.setScale(2, RoundingMode.UP);
    System.out.println("16.7499924f without MathContext(precision) but with setScale of 2 =" + numB5.floatValue());// 16.749992

}

Answer 1

A BigDecimal is immutable. You're ignoring the result of setScale(). So all you have there are basically noops.

Also, use a string to create your BigDecimal. And don't print it as a float value. You're losing precision when doing that, which is the whole point of using BigDecimal in the first place:

String num = "16.7499924";

int digitCount = 4;
MathContext mContext = new MathContext(digitCount, RoundingMode.UP);
BigDecimal numB = new BigDecimal(num, mContext);
System.out.println("16.7499924f with MathContext(precision) of 4 =" + numB.toPlainString());
numB = numB.setScale(4, RoundingMode.UP);
System.out.println("16.7499924f with MathContext(precision) of 4 and after setScale of 4 =" + numB.toPlainString());

Answer 2

As a BigDecimal is immutable, setScale(4, RoundingMode.UP); returns a new BigDecimal with the scale operation applied, it does not affect the original variable. So substitute the numB.setScale(4, RoundingMode.UP); calls by numB2 = numB2.setScale(4, RoundingMode.UP); and omit BigDecimal numB2 = new BigDecimal(num); as it is useless, since you overwrite it in the next step anyways.

Also, you are using a BigDecimal , but then you convert it to float . This may be imprecise, as the documentation states for floatValue() :

Note that even when the return value is finite, this conversion can lose information about the precision of the BigDecimal value.