Issue #4084: Fix max, min, max_mag and min_mag Decimal methods to
give correct results in the case where one argument is a quiet NaN
and the other is a finite number that requires rounding.
Thanks Mark Dickinson.
3 files changed