MathProcessor narrowing types

[dohrayme]

MVEL 2.0.19

If I have a function:

def abs(x) { (x >= 0) ? x : -x }

and use it thus: abs(Integer / BigDecimal)

where the egress type is going to be (is fixed to be) a BigDecimal, a ClassCast exception is thrown during the abs function negation step. Investigating I see a BigDecimal signer is being used but the division operation result has had its type narrowed to be a Double, see MathsProcessor::_doOperations it always passes true to doBigDecimalArithmetic which results in the type being narrowed. Should this not be conditional on the operands and return type?

[mikebrock]

Can you confirm this is still a problem in MVEL 2.1.Beta6, because there will be no 2.0.20. 2.1 is the next 2.x.x version.

On Aug 5, 2011, at 6:35 AM, dohrayme wrote:

MVEL 2.0.19

If I have a function:

def abs(x) { (x >= 0) ? x : -x }

and use it thus: abs(Integer / BigDecimal)

where the egress type is going to be (is fixed to be) a BigDecimal, a ClassCast exception is thrown during the abs function negation step. Investigating I see a BigDecimal signer is being used but the division operation result has had its type narrowed to be a Double, see MathsProcessor::_doOperations it always passes true to doBigDecimalArithmetic which results in the type being narrowed. Should this not be conditional on the operands and return type?

Reply to this email directly or view it on GitHub: https://github.com/mikebrock/mvel/issues/20