Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
1.0-beta-3, 1.0-beta-4
-
None
-
None
-
All
Description
Mathmatical operators such as +, -, ... do not behave as expected with java.math.BigDecimal. They are coverted to Integer for + and -.
For example:
bd = new java.math.BigDecimal("1.3");
print 1 + bd;
Shows 3, not 3.3 as expected.
From post by Steve Goetze:
> The DefaultGroovyMethods: plus(Number, Number), minus(Number,
> Number),
> multiply(Number, Number), and divide(Number, Number) have coercion
> techniques that don't work very well for BigDecimal objects
> (results are
> generally coerced to Integer types except for divide which returns
> Double). Would it be possible to change these methods to check for
> a left or right BigDecimal and preserve? For example:
>
> /**
> * Add two Numbers
> *
> * @param left a Number
> * @param right another Number to add
> * @return the addition of both Numbers
> */
> public static Number plus(Number left, Number right) {
> /** @todo maybe a double dispatch thing to handle new large
> numbers? */
> if (isBigDecimal(left) || isBigDecimal(right))
> else if (isBigInteger(left) || isBigInteger(right))
{ > return new Double(left.doubleValue() + > right.doubleValue()); > }> else if (isFloatingPoint(left) || isFloatingPoint(right)) {> return new Double(left.doubleValue() + > right.doubleValue());> }
> else if (isLong(left) || isLong(right))
{ > return new Long(left.longValue() + right.longValue()); > }> else
{ > return new Integer(left.intValue() + right.intValue()); > }> }
>
> where:
>
> public static boolean isBigDecimal(Number number)
>
> and:
>
> public static BigDecimal asBigDecimal(Number n)
>
>
> divide(Number, Number) could use a default rounding mode of
> BigDecimal.ROUND_UP.
>
> Also, I don't see that literal longs and doubles can be specified ala
> java with an "L" or "D". If this is something you would consider
> adding, it would also be nice to be able to specify literal
> BigDecimals
> and BigIntegers, perhaps with a "B" suffix.