Github user rednaxelafx commented on a diff in the pull request:
https://github.com/apache/spark/pull/20224#discussion_r163947756
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/WholeStageCodegenExec.scala
---
@@ -471,19 +531,21 @@ case class WholeStageCodegenExec(child: SparkPlan)
extends UnaryExecNode with Co
}
""", inlineToOuterClass = true)
+ val className = generatedClassName()
+
val source = s"""
public Object generate(Object[] references) {
- return new GeneratedIterator(references);
+ return new $className(references);
}
${ctx.registerComment(s"""Codegend pipeline
for\n${child.treeString.trim}""")}
- final class GeneratedIterator extends
org.apache.spark.sql.execution.BufferedRowIterator {
+ final class $className extends
${classOf[BufferedRowIterator].getName} {
--- End diff --
Yes, that's a great idea that I missed! Thanks for your comments, @kiszk !
We can get that effect in two ways:
1. Change the current line `${ctx.registerComment(s"""Codegend pipeline
for\n${child.treeString.trim}""")}` from `child.treeString` to
this.treeString`, which will include the codegen stage ID through the
`treeString`, just like the explain output.
2. Simply add `$codegenStageId` into the `Codegend pipeline for` line.
3. Do both above. (Did I say two...?)
Which do you prefer?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
