An issue was found where by precise integers >= scale 10 were being scaled back too aggressively causing numbers at an integer bound for the mantissa to be rounded toward the integer itself.
It was effectively "skipping" a power of 10 during comparison - which when the mantissa was a power of 10 itself, would cause inaccurate results.
This fixes #604
An issue was found where by precise integers >= scale 10 were being scaled back too aggressively causing numbers at an integer bound for the mantissa to be rounded toward the integer itself. It was effectively "skipping" a power of 10 during comparison - which when the mantissa was a power of 10 itself, would cause inaccurate results.