andygrove opened a new issue, #2021: URL: https://github.com/apache/datafusion-comet/issues/2021
### Describe the bug As part of exploring writing unit tests for serde code in https://github.com/apache/datafusion-comet/issues/2020, I discovered that we currently have incorrect behavior for `try_add` (and probably for many other `try_` functions). The issue can be reproduced with the following test: ```scala test("try_add") { val data = Seq((Integer.MAX_VALUE, 1)) withParquetTable(data, "tbl") { checkSparkAnswerAndOperator("SELECT try_add(_1, _2) FROM tbl") } } ``` This fails with: ``` == Results == !== Correct Answer - 1 == == Spark Answer - 1 == struct<try_add(_1, _2):int> struct<try_add(_1, _2):int> ![null] [-2147483648] ``` The issue is that we do not respect the `EvalMode` for this (and other) expressions. When serializing the expression we populate a `fail_on_error` flag based on `add.evalMode == EvalMode.ANSI`. However, we should really be serializing the `evalMode` and implementing all three distinct behaviors. In the short term, we should probably just fall back to Spark for `Try`, since the behavior is incorrect. ### Steps to reproduce _No response_ ### Expected behavior _No response_ ### Additional context _No response_ -- 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: github-unsubscr...@datafusion.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org For additional commands, e-mail: github-h...@datafusion.apache.org
