Skip to content

[SPARK-13242] [SQL] Fall back to interpreting complex when expressions #11243

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
joehalliwell wants to merge 6 commits into from
Closed

[SPARK-13242] [SQL] Fall back to interpreting complex when expressions #11243

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
5849e45
Add test case
joehalliwell Feb 9, 2016
9b5736b
Fall back to interpreting complex when expressions
joehalliwell Feb 17, 2016
de601db
Add comment
joehalliwell Feb 17, 2016
6d2e9ff
Fix code style
joehalliwell Feb 17, 2016
7d52516
Reduce source code threshold to 10K
joehalliwell Feb 17, 2016
4866a8d
Drop source code threshold to 1K
joehalliwell Feb 17, 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
12 changes: 9 additions & 3 deletions ...yst/src/main/scala/org/apache/spark/sql/catalyst/expressions/conditionalExpressions.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ case class If(predicate: Expression, trueValue: Expression, falseValue: Expressi
* @param elseValue optional value for the else branch
*/
case class CaseWhen(branches: Seq[(Expression, Expression)], elseValue: Option[Expression] = None)
extends Expression {
extends Expression with CodegenFallback {

override def children: Seq[Expression] = branches.flatMap(b => b._1 :: b._2 :: Nil) ++ elseValue

Expand Down Expand Up @@ -155,6 +155,7 @@ case class CaseWhen(branches: Seq[(Expression, Expression)], elseValue: Option[E
// }
// }
// }

val cases = branches.map { case (condExpr, valueExpr) =>
val cond = condExpr.gen(ctx)
val res = valueExpr.gen(ctx)
Expand Down Expand Up @@ -182,11 +183,16 @@ case class CaseWhen(branches: Seq[(Expression, Expression)], elseValue: Option[E

generatedCode += "}\n" * cases.size

s"""
if (generatedCode.size > 64 * 1000) {
Copy link
Contributor

@davies davies Feb 17, 2016

Choose a reason for hiding this comment

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

64k is too large, JIT will stop to work after 8K bytes code. 1K?

Copy link
Author

@joehalliwell joehalliwell Feb 17, 2016

Choose a reason for hiding this comment

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

I guess you mean 10K? That makes sense. Should we modify the corresponding check in CodegenContext.splitExpressions?

super.genCode(ctx, ev)
}
else {
Copy link
Contributor

@davies davies Feb 17, 2016

Choose a reason for hiding this comment

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

} else {

Copy link
Author

@joehalliwell joehalliwell Feb 17, 2016

Choose a reason for hiding this comment

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

Thanks

s"""
boolean ${ev.isNull} = true;
${ctx.javaType(dataType)} ${ev.value} = ${ctx.defaultValue(dataType)};
$generatedCode
"""
"""
}
}

override def toString: String = {
Expand Down
22 changes: 22 additions & 0 deletions ...talyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/CodeGenerationSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,28 @@ class CodeGenerationSuite extends SparkFunSuite with ExpressionEvalHelper {
}
}


test("SPARK-13242: complicated when expressions") {
val cases = 50
val clauses = 20

// Generate an individual case
def generateCase(n: Int): (Expression, Expression) = {
val condition = (1 to clauses)
.map(c => EqualTo(BoundReference(0, StringType, false), Literal(s"$c:$n")))
.reduceLeft[Expression]((l, r) => Or(l, r))
(condition, Literal(n))
}

val expression = CaseWhen((1 to cases).map(generateCase(_)))

val plan = GenerateMutableProjection.generate(Seq(expression))()
val input = new GenericMutableRow(Array[Any](UTF8String.fromString(s"${clauses}:${cases}")))
val actual = plan(input).toSeq(Seq(expression.dataType))

assert(actual(0) == cases)
}

test("test generated safe and unsafe projection") {
val schema = new StructType(Array(
StructField("a", StringType, true),
Expand Down