Skip to content

[SPARK-8443][SQL] Split GenerateMutableProjection Codegen due to JVM Code Size Limits #7076

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
saurfang wants to merge 4 commits into from
Closed

[SPARK-8443][SQL] Split GenerateMutableProjection Codegen due to JVM Code Size Limits #7076

Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
9405680
[SPARK-8443][SQL] split projection code by size limit
saurfang Jun 18, 2015
1b5aa7e
[SPARK-8443][SQL] inline execution if one block only
saurfang Jul 18, 2015
adef95a
[SPARK-8443][SQL] Use safer factor and rewrite splitting code
saurfang Jul 18, 2015
b7a7635
[SPARK-8443][SQL] Execute and verify split projections in test
saurfang Jul 19, 2015
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
Prev Previous commit
Next Next commit
[SPARK-8443][SQL] inline execution if one block only
  • Loading branch information
@saurfang
saurfang committed Jul 18, 2015
commit 1b5aa7e3c4fd9f0ffd2a55e0b26a1fdbda446005
33 changes: 19 additions & 14 deletions ...n/scala/org/apache/spark/sql/catalyst/expressions/codegen/GenerateMutableProjection.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ object GenerateMutableProjection extends CodeGenerator[Seq[Expression], () => Mu
${ctx.setColumn("mutableRow", e.dataType, i, evaluationCode.primitive)};
"""
}
val partitionedProjectionCode = projectionCode.foldLeft(List.empty[String]) {
val projectionBlocks = projectionCode.foldLeft(List.empty[String]) {
Copy link
Contributor

@rxin rxin Jul 18, 2015

Choose a reason for hiding this comment

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

would be great if you can rewrite the code to make it slightly less functional here. I'd actually prefer a more imperative version here

(acc, code) =>
acc match {
case Nil => List(code)
Expand All @@ -58,19 +58,24 @@ object GenerateMutableProjection extends CodeGenerator[Seq[Expression], () => Mu
code::acc
}
}
}
.zipWithIndex
.map {
case (body, i) =>
s"""
private void apply$i(InternalRow i) {
$body
}
"""
}
val projectionCalls = ((partitionedProjectionCode.length - 1) to 0 by -1)
.map(i => s"apply$i(i);")
.mkString("\n")
val (projectionFuns, projectionCalls) = {
// inline execution if codesize limit was not broken
if (projectionBlocks.length == 1) {
("", projectionBlocks.head)
} else {
(
projectionBlocks.zipWithIndex.map { case (body, i) =>
s"""
|private void apply$i(InternalRow i) {
| $body
|}
""".stripMargin
}.mkString,
Copy link
Contributor

@JoshRosen JoshRosen Jul 19, 2015

Choose a reason for hiding this comment

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

Similarly, should this be mkString("\n") or is there implicitly a newline somewhere that I'm overlooking?

Copy link
Contributor Author

@saurfang saurfang Jul 19, 2015

Choose a reason for hiding this comment

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

Similar to above. The lines of s""" and """ seem already added newlines for readability.

Copy link
Contributor

@JoshRosen JoshRosen Jul 19, 2015

Choose a reason for hiding this comment

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

SGTM.

((projectionBlocks.length - 1) to 0 by -1).map(i => s"apply$i(i);").mkString("\n")
)
}
}
val mutableStates = ctx.mutableStates.map { case (javaType, variableName, initialValue) =>
s"private $javaType $variableName = $initialValue;"
}.mkString("\n ")
Expand Down Expand Up @@ -100,7 +105,7 @@ object GenerateMutableProjection extends CodeGenerator[Seq[Expression], () => Mu
return (InternalRow) mutableRow;
}

${partitionedProjectionCode.mkString("\n")}
$projectionFuns

public Object apply(Object _i) {
InternalRow i = (InternalRow) _i;
Expand Down