wForget commented on PR #56439: URL: https://github.com/apache/spark/pull/56439#issuecomment-6164490681
Would it be possible to backport this fix to `branch-4.2`? During performance testing with Spark 4.2.0, I found that [TypeOps.apply](https://github.com/apache/spark/blob/7f0631c94762ade281546690c6813f5c2a6bdad9/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/types/ops/TypeOps.scala#L231), including its callees, accounted for about **21.6% of wall-clock samples in one stage**. Almost all of that came from repeated `SQLConf.get` and `typesFrameworkEnabled` lookups while reading array elements in `array_union` and `array_intersect`. This fix appears to directly address that overhead. <img width="1892" height="702" alt="image" src="https://github.com/user-attachments/assets/b0a6fd52-39a4-4431-9e25-44f5d60fa3aa"; /> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
