[GROOVY-3171] Handle floating conversion to BigDecimal and Double using 'as' gracefully.

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.5.7, 1.6-beta-2
• Fix Version/s: 1.6-rc-2
• Assignee: James P. White
• Reporter: James P. White
• Created: 26/Nov/08 00:03
• Updated: 22/Jan/09 21:34
• Resolved: 18/Dec/08 17:44

Description

The current implementation of DGM.toBigDecimal uses a new BigDecimal(Number.doubleValue()) to do the conversion. As the docs on BigDecimal(double) explain, that may result in a bunch of insignificant digits yielding a decimal number that is not what the user expects (or wants).

http://java.sun.com/javase/6/docs/api/java/lang/Double.html#toString(double)

That means we get this:

println (0.1f as BigDecimal)
==>
0.100000001490116119384765625

So the preferred method is to use new BigDecimal(Double.toString()) which will round off insignificant digits.

So that now we get this:

println (0.1f as BigDecimal)
==>
0.1

and most reassuringly this is now true (which fails in 1.5.x):

assert (0.1 == (0.1f as BigDecimal))