dongjoon-hyun commented on PR #59751: URL: https://github.com/apache/spark/pull/59751#issuecomment-6388064499
Thank you for running the suite and for the suggestion, @Sid4030. The existing test already fails when chaining is disabled entirely. Its second part checks the same-depth case, `transform(values, row -> transform(row, x -> g(g(x))))`, where both `g` calls are lifted to depth 2, and asserts that one lifted UDF still chains the other: https://github.com/apache/spark/blob/c439a32ffab8ea0076b638ce4a02b8b0b570e6ce/sql/core/src/test/scala/org/apache/spark/sql/execution/python/ExtractPythonUDFFromLambdaSuite.scala#L287-L292 I checked it by making `shouldExtractUDFExpressionTree` never chain. Each `g` is then evaluated by its own `ArrowEvalPython`, so no lifted UDF has a `PythonUDF` child, and the test fails at L292. Your case reaches the same check. Once `f` is extracted, `h(g(...))` is a chain of two depth-2 UDFs over the result column of `f`, just like `g(g(...))` above. So I'd like to keep the test as it is. -- 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]
