We sometimes see floating point errors in grade calculation. Some of the calculations are already using Double instead of Float, but other calculations only use Float, e.g. for collection grades and collection weights.
Change the calculations to use Double, so we don’t introduce precision errors.

Submitted by Elmer van Chastelet on 12 July 2021 at 09:34

This won’t fix the precision issue. It even makes it worse, because the grade traces (String) would expose the precision errors from the Double values, where a stringified Float would project the least number of decimals to distinguish the number from the next Float value.
The fix would be to limit the precision using BigDecimal (new issue)