cpp_dec_float: assigning double(-infinity) drops the negative

[Jason Rhinelander <jason@…>]

Initializing (or assigning) negative infinity to a cpp_dec_float assigns positive infinity instead of negative infinity.

The problem seems to be that cpp_dec_float.hpp, in operator=(long double a), does:

   if((boost::math::isinf)(a))
      return *this = inf();

but I think that should instead be (I have not tested this):

    if((boost::math::isinf)(a)) {
       *this = inf();
       if (a < 0) negate();
       return *this;
    }

[John Maddock]

Thanks, that exposed a number of related issues, fixed in ​https://github.com/boostorg/multiprecision/commit/aeab24b4fbb5738740d40beb1c55aad039ca1ac1