Github user cloud-fan commented on the issue: https://github.com/apache/spark/pull/19586 looking at the `SerializationStream` interface, I think it's designed for read/write objects of different classes, so your optimization should not be applied there. Instead, I think we should introduce `SerializerInstance#serializeStreamForClass[T]`, which returns `ClassSpecificSerializationStream[T]` that is designed for writing objects of same class.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org