cpp_bin_float: Anal fixation. Part 1. Sign of zero result of summation.

[Michael Shatz]

IEEE-754 has one simple rule - subtraction of two equal finite numbers results in +0. From here follows that there are only two cases in which subtraction/addition of zeros results in -0:

-0 + -0 and -0 - +0. All other combinations results in +0.

Also from here follows that subtraction/addition of non-zeros with equal absolute value never results in -0.

I see no good reasons why boost::cpp_bin_float should not follow the same conventions. But right now it does not.

[Michael Shatz]

demonstrates non-IEEE sign of zero of cpp_bin_float addition/subtraction

[John Maddock]

Fixed in ​https://github.com/boostorg/multiprecision/commit/5a5612c0a9c2de934a761b7e090b7c15466b2988, thanks for the report!