http://qs321.pair.com?node_id=1103182


in reply to Re^3: Math::FixedPrecision and bignum clash
in thread Math::FixedPrecision and bignum clash

Yes, thanks. I was wrong and it is using bankers rounding. Looks at the last digit being kept and if odd rounds down and even rounds up so long as first digit beyond the precision is >= 5. That is another bug in the code we're using here which does not want bankers rounding.