Github user mgaido91 commented on a diff in the pull request:
https://github.com/apache/spark/pull/19937#discussion_r155927167
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/SortMergeJoinExec.scala
---
@@ -617,6 +619,7 @@ case class SortMergeJoinExec(
s"""
|while (findNextInnerJoinRows($leftInput, $rightInput)) {
+ | ${leftVarDecl.mkString("\n")}
--- End diff --
I am not sure about the amount of overhead introduced honestly. But I think
that there might be, even if low. Therefore my suggestion was to avoid the
overhead, since IMHO it is feasible and it reflects the previous situation.
Honestly I don't think that the readability of the generated code is a big
point, because I think that the generated code is already nearly impossible to
be read as of now (without reading and knowing the code which generates it).
But maybe we can create a test and evaluate the overhead if you think this
is the best option. WDYT?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
