Skip to content

[SPARK-33971][SQL] Eliminate distinct from more aggregates #30999

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
tanelk wants to merge 7 commits into from
Closed

[SPARK-33971][SQL] Eliminate distinct from more aggregates #30999

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
c57f4e8
Eliminate distinct from more aggregates
tanelk Jan 3, 2021
0057d57
Eliminate distinct from more aggregates
tanelk Jan 3, 2021
ef3c7f5
Merge branch 'SPARK-33971_eliminate_distinct' of git+ssh://github.com…
tanelk Jan 3, 2021
d1cd01d
Eliminate distinct from more aggregates
tanelk Jan 3, 2021
5f9df99
Remove HLL++
tanelk Jan 9, 2021
4b9d04b
Refactor test
tanelk Jan 9, 2021
f95fb8d
Remove first and last
tanelk Jan 26, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 14 additions & 5 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -349,11 +349,20 @@ abstract class Optimizer(catalogManager: CatalogManager)
*/
object EliminateDistinct extends Rule[LogicalPlan] {
override def apply(plan: LogicalPlan): LogicalPlan = plan transformExpressions {
case ae: AggregateExpression if ae.isDistinct =>
ae.aggregateFunction match {
case _: Max | _: Min => ae.copy(isDistinct = false)
case _ => ae
}
case ae: AggregateExpression if ae.isDistinct && isDuplicateAgnostic(ae.aggregateFunction) =>
ae.copy(isDistinct = false)
}

private def isDuplicateAgnostic(af: AggregateFunction): Boolean = af match {
case _: Max => true
case _: Min => true
case _: BitAndAgg => true
case _: BitOrAgg => true
case _: First => true

@zhengruifeng zhengruifeng Jan 11, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

First and Last are non-deterministic, does this matter?

@tanelk tanelk Jan 11, 2021

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In another PR I try to argue that they should be deterministic: #29810
TLDR: An analogous aggregator would be sum on float and double datatype - its result does depend on the order of its inputs, but is deterministic.

@zhengruifeng zhengruifeng Jan 11, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If we finally make First and Last deterministic, then I guess they need to be removed from EliminateDistinct?

@cloud-fan cloud-fan Jan 13, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let's exclude First/Last here before #29810 is merged

@tanelk tanelk Jan 26, 2021

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done

case _: Last => true
case _: HyperLogLogPlusPlus => true
Comment thread
tanelk marked this conversation as resolved.
Outdated
case _: CollectSet => true
case _ => false
}
}

Expand Down
43 changes: 23 additions & 20 deletions ...alyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/EliminateDistinctSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,8 @@ package org.apache.spark.sql.catalyst.optimizer

import org.apache.spark.sql.catalyst.dsl.expressions._
import org.apache.spark.sql.catalyst.dsl.plans._
import org.apache.spark.sql.catalyst.expressions.Expression
import org.apache.spark.sql.catalyst.expressions.aggregate._
import org.apache.spark.sql.catalyst.plans.PlanTest
import org.apache.spark.sql.catalyst.plans.logical.{LocalRelation, LogicalPlan}
import org.apache.spark.sql.catalyst.rules.RuleExecutor
Expand All @@ -32,25 +34,26 @@ class EliminateDistinctSuite extends PlanTest {

val testRelation = LocalRelation('a.int)

test("Eliminate Distinct in Max") {
val query = testRelation
.select(maxDistinct('a).as('result))
.analyze
val answer = testRelation
.select(max('a).as('result))
.analyze
assert(query != answer)
comparePlans(Optimize.execute(query), answer)
}

test("Eliminate Distinct in Min") {
val query = testRelation
.select(minDistinct('a).as('result))
.analyze
val answer = testRelation
.select(min('a).as('result))
.analyze
assert(query != answer)
comparePlans(Optimize.execute(query), answer)
Seq(
("max", Max(_)),
("min", Min(_)),
Comment thread
tanelk marked this conversation as resolved.
Outdated
("approx_count_distinct", HyperLogLogPlusPlus(_: Expression)),
("first", First(_, ignoreNulls = true)),
("last", Last(_, ignoreNulls = true)),
("bit_and", BitAndAgg(_)),
("bit_or", BitOrAgg(_)),
("collect_set", CollectSet(_: Expression))
).foreach {
case (name, agg) =>
test(s"Eliminate Distinct in $name") {
val query = testRelation
.select(agg('a).toAggregateExpression(isDistinct = true).as('result))
.analyze
val answer = testRelation
.select(agg('a).toAggregateExpression(isDistinct = false).as('result))
.analyze
assert(query != answer)
comparePlans(Optimize.execute(query), answer)
}
}
}