Thanks for your bug report. It looks like the beta has extended float precision to the point where rounding fails for some numbers. This is due to the fact that
0.1 does not actually exist in floating point. The next beta will decrease the precision to 12 digits, which is still double that of v7.45.
