Github user cloud-fan commented on the issue: https://github.com/apache/spark/pull/19601 So for primitive types, we encode and compress them to binary. When reading cached data, they are decoded to primitive array and can be put in `OnHeadColumnVector` directly. For primitive type array, we treat it as binary. So when decoding it, we get a byte[] and need more effort to convert it to primitive type and put in `OnHeadColumnVector`. Can we change how we encode array type like Arrow did?
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org