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
