Incorrect `BigDecimal` fraction representation (2.9 -> 2.10 change)

[wlukowicz]

According to the spec, decimal fractions (tag 4) of value m * 10^e should be represented as a tagged array containing the exponent e and the mantissa m. However, Jackson writes BigDecimals as tagged arrays containing their scale and unscaled value. The difference being that the value of BigDecimal is unscaled value * 10^-scale, which means the scale is the opposite of the exponent e.

The spec gives an example how 273.15 should be represented:

C4             -- Tag 4
   82          -- Array of length 2
      21       -- -2
      19 6AB3  -- 27315

Here is the actual representation:

C4             -- Tag 4
   82          -- Array of length 2
      02       -- 2
      19 6AB3  -- 27315

[cowtowncoder]

Ok. Sounds like there is a problem, and looks like information included should be enough to reproduce the problem as a unit test.
Thank you for reporting this.

[cowtowncoder]

Hmmh. Ok, yes. Looks like I interpreted exponent the wrong way, but for better or worse it is consistently so, meaning that Jackson can read content it writes.

The challenge here is backwards compatibility. I think that I will fix this for 2.10, instead of 2.9 patch, since this does change working quite drastically.

[wlukowicz]

Looks good, thanks for fixing.