Nowadays multiplying or adding BigDecimal with Float results in BigDecimal. This is the case for at least Ruby >= 2.3.
Ruby comes with a class
BigDecimal
Show archive.org snapshot
which you can use for arbitrary precision arithmetic. You should use BigDecimal instead of Float whenever you care about rounding errors, e.g. whenever you are dealing with money.
You should remember these two rules when working with BigDecimal values:
- When you add or multiply a
BigDecimalwith anotherBigDecimal, the result will be a newBigDecimalwith sufficient precision to represent the result. No rounding or clipping should occur in that operation. - When you add or multiply a
BigDecimalwith aFloat, the result will be aFloat. That means you just transitioned back into the land of random rounding errors.
Don't screw up your clean BigDecimal values by thoughtlessly multiplying them with a Float. For instance, this is a stupid idea:
VAT_RATE = 1.19
net = BigDecimal('12.3413')
gross = net * VAT_RATE # oops, gross is now a Float
Do this instead:
VAT_RATE = BigDecimal('1.19')
net = BigDecimal('12.3413')
gross = net * VAT_RATE # gross is now a BigDecimal
ActiveRecord models represent MySQL's DECIMAL columns as BigDecimal attributes. Keep in mind that values assigned to such attributes might be clipped or rounded when the record is saved to the database.