Value of zero is considered negative if produced by adding to a negative number

[dlongley]

Value of zero is considered negative if produced by adding to a negative number

[MikeMcl]

Should, according to who?

I never claimed the library conformed to IEEE 754 in all aspects, and in mathematics zero is unsigned anyway.

If -0 + 0 equals minus zero, it seems reasonable that -1 + 1 should do so also.

I did however state in the API document that "The treatment of ±0, ±Infinity and NaN is consistent with how JavaScript treats these values", and in JavaScript, -1 + 1 equals positive zero, so I agree that either the document or the behaviour was incorrect.

According to IEEE 754 (2008) 6.3, whether -1 + 1 should equal positive or negative zero depends on the current rounding mode.

When the sum of two operands with opposite signs (or the difference of two operands with like signs) is
exactly zero, the sign of that sum (or difference) shall be +0 in all rounding-direction attributes except
roundTowardNegative; under that attribute, the sign of an exact zero sum (or difference) shall be −0.
However, x + x = x −(−x) retains the same sign as x even when x is zero.

I have updated the library so that it now conforms fully with IEEE 754 in this respect. I.e.

// Round towards negative Infinity
BigNumber.config({ ROUNDING_MODE: BigNumber.ROUND_FLOOR });
// -1 +  1 = -0
//  1 + -1 = -0
// -0 +  0 = -0
//  0 + -0 = -0

// Any rounding mode but ROUND_FLOOR
BigNumber.config({ ROUNDING_MODE: 4 });    
// -1 +  1 =  0
//  1 + -1 =  0
// -0 +  0 =  0
//  0 + -0 =  0

// Any rounding mode
// -0 + -0 = -0
//  0 +  0 =  0

This means that if ROUND_FLOOR is selected, the sign of the result of -1 + 1 will continue to differ from JavaScript numbers in this regard.

Anyway, thank you for the feedback and for spotting this. I am disappointed that I hadn't noticed this earlier.
I'll have to make changes to big.js also, so you have made a significant contribution to both libraries that is much appreciated.

Fixed in v1.2.1.

[dlongley]

I follow and understand the above reasoning. Thanks for the patch!