Github user HyukjinKwon commented on the issue: https://github.com/apache/spark/pull/19234 I think it was changed back in https://github.com/apache/spark/pull/7363. I think that avoids precision loss from Python's `float` limitation: ```python >>> decimal.Decimal(1 / 1e6) Decimal('9.99999999999999954748111825886258685613938723690807819366455078125E-7') >>> decimal.Decimal(1 / 100000) Decimal('0') >>> decimal.Decimal(1 / 100000.) Decimal('0.000010000000000000000818030539140313095458623138256371021270751953125') ``` Also, I think `//` is used for Python 3 behaviour: ```python >>> 5 / 2 2.5 >>> 5 // 2 2 ```
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org