Github user davies commented on a diff in the pull request: https://github.com/apache/spark/pull/12067#discussion_r59617422 --- Diff: sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/TypedAggregateExpression.scala --- @@ -19,133 +19,153 @@ package org.apache.spark.sql.execution.aggregate import scala.language.existentials -import org.apache.spark.internal.Logging import org.apache.spark.sql.Encoder -import org.apache.spark.sql.catalyst.InternalRow -import org.apache.spark.sql.catalyst.encoders.{encoderFor, ExpressionEncoder, OuterScopes} +import org.apache.spark.sql.catalyst.analysis.{UnresolvedAttribute, UnresolvedDeserializer, UnresolvedExtractValue} +import org.apache.spark.sql.catalyst.encoders.encoderFor import org.apache.spark.sql.catalyst.expressions._ -import org.apache.spark.sql.catalyst.expressions.aggregate.ImperativeAggregate +import org.apache.spark.sql.catalyst.expressions.aggregate.DeclarativeAggregate import org.apache.spark.sql.expressions.Aggregator import org.apache.spark.sql.types._ object TypedAggregateExpression { - def apply[A, B : Encoder, C : Encoder]( - aggregator: Aggregator[A, B, C]): TypedAggregateExpression = { + def apply[BUF : Encoder, OUT : Encoder]( + aggregator: Aggregator[_, BUF, OUT]): TypedAggregateExpression = { + val bufferEncoder = encoderFor[BUF] + // We will insert the deserializer and function call expression at the bottom of each serializer + // expression while executing `TypedAggregateExpression`, which means multiply serializer + // expressions will all evaluate the same sub-expression at bottom. To avoid the re-evaluating, + // here we always use one single serializer expression to serialize the buffer object into a + // single-field row, no matter whether the encoder is flat or not. We also need to update the + // deserializer to read in all fields from that single-field row. + // TODO: remove this trick after we have better integration of subexpression elimination and --- End diff -- Can we hold this pr a little bit? let me think of how to do subexpression elimination in aggregate.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org