lisancao commented on PR #59778: URL: https://github.com/apache/spark/pull/59778#issuecomment-6648795773
Thanks @holdenk! Rebased onto master, and removed the migration note in 6fddee13206. Before I rework the check: the review bot suggests deciding the fallback on the executors, when the UDF runs, instead of on the driver while planning ([in its review](https://github.com/apache/spark/pull/59778#pullrequestreview-5720487728)). If we keep it on the driver, it's more aligned with classic and out of the shared execution path. Which would you prefer? If it stays on the driver, the check can go into a small module under `pyspark/sql/worker` and run through `PythonPlannerRunner`, like UDTF analyze and Python data sources, so it starts the same way as the workers and can fork from the daemon. It'll return a Boolean and throw errors for anything unexpected, as you suggested. UDTFs are purposely excluded from this PR for scope and will be in a follow-up (SPARK-60828). -- 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]
