Skip to content

[SPARK-15647] [SQL] Fix Boundary Cases in OptimizeCodegen Rule #13392

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
gatorsmile wants to merge 9 commits into from
Closed

[SPARK-15647] [SQL] Fix Boundary Cases in OptimizeCodegen Rule #13392

Show file tree
Hide file tree
Changes from 7 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
40b70f3
fix the boundary case.
gatorsmile May 29, 2016
aaf5f2e
remove the useless comments.
gatorsmile May 29, 2016
f351c10
style fix.
gatorsmile May 29, 2016
ecc4318
fix
gatorsmile May 30, 2016
db67f8c
revert
gatorsmile May 30, 2016
414e116
address comments
gatorsmile May 31, 2016
4306c4f
address comments
gatorsmile May 31, 2016
b2849e8
address comments
gatorsmile May 31, 2016
9830e31
address comments
gatorsmile May 31, 2016
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
8 changes: 6 additions & 2 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 @@ -937,8 +937,12 @@ object SimplifyConditionals extends Rule[LogicalPlan] with PredicateHelper {
*/
case class OptimizeCodegen(conf: CatalystConf) extends Rule[LogicalPlan] {
def apply(plan: LogicalPlan): LogicalPlan = plan transformAllExpressions {
case e @ CaseWhen(branches, _) if branches.size < conf.maxCaseBranchesForCodegen =>
e.toCodegen()
case e: CaseWhen if canCodeGen(e) => e.toCodegen()
Copy link
Contributor

@jaceklaskowski jaceklaskowski May 31, 2016

Choose a reason for hiding this comment

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

Sorry for nitpicking, but could you use canCodegen instead (to follow the name of the method to call)? Thanks!

Copy link
Member Author

@gatorsmile gatorsmile May 31, 2016

Choose a reason for hiding this comment

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

Sure, let me fix it. Thanks!

}

private def canCodeGen(e: CaseWhen): Boolean = {
val numBranches = e.branches.size + e.elseValue.size
numBranches <= conf.maxCaseBranchesForCodegen
}
}

Expand Down
38 changes: 38 additions & 0 deletions sql/core/src/test/scala/org/apache/spark/sql/internal/SQLConfSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@
package org.apache.spark.sql.internal

import org.apache.spark.sql.{QueryTest, Row, SparkSession, SQLContext}
import org.apache.spark.sql.execution.WholeStageCodegenExec
import org.apache.spark.sql.test.{SharedSQLContext, TestSQLContext}

class SQLConfSuite extends QueryTest with SharedSQLContext {
Expand Down Expand Up @@ -219,4 +220,41 @@ class SQLConfSuite extends QueryTest with SharedSQLContext {
}
}

test("MAX_CASES_BRANCHES") {
import testImplicits._

val original = spark.conf.get(SQLConf.MAX_CASES_BRANCHES)
try {
withTable("tab1") {
spark
.range(10)
.select('id as 'a, 'id as 'b, 'id as 'c, 'id as 'd)
Copy link
Contributor

@cloud-fan cloud-fan May 31, 2016

Choose a reason for hiding this comment

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

I think we only need a? or just spark.range(10).write.saveAsTable, then we can use id in the case when

Copy link
Member Author

@gatorsmile gatorsmile May 31, 2016

Choose a reason for hiding this comment

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

Sure, will do it. Thanks!

.write
.saveAsTable("tab1")

val sql_one_branch_caseWhen = "SELECT CASE WHEN a = 1 THEN 1 END FROM tab1"
val sql_two_branch_caseWhen = "SELECT CASE WHEN a = 1 THEN 1 ELSE 0 END FROM tab1"

spark.conf.set(SQLConf.MAX_CASES_BRANCHES.key, "0")
Copy link
Contributor

@cloud-fan cloud-fan May 31, 2016

Choose a reason for hiding this comment

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

how about:

withTable {
  spark.range(10)....
  val oneBranchCaseWhen = ....
  val twoBranchCaseWhen = ...
  withConf {
    assert(...)
  }
  withConf {
    assert(...)
  }
  withConf {
    assert(...)
  }
}

Copy link
Member Author

@gatorsmile gatorsmile May 31, 2016

Choose a reason for hiding this comment

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

Yeah, will do it.

assert(!sql(sql_one_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])
assert(!sql(sql_two_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])

spark.conf.set(SQLConf.MAX_CASES_BRANCHES.key, "1")
assert(sql(sql_one_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])
assert(!sql(sql_two_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])

spark.conf.set(SQLConf.MAX_CASES_BRANCHES.key, "2")
assert(sql(sql_one_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])
assert(sql(sql_two_branch_caseWhen)
.queryExecution.executedPlan.isInstanceOf[WholeStageCodegenExec])
}
} finally {
spark.conf.set(SQLConf.MAX_CASES_BRANCHES.key, s"$original")
}
}
}