andygrove commented on issue #2611: URL: https://github.com/apache/datafusion-comet/issues/2611#issuecomment-3423263022
As an example of the benefit of fuzz testing, I just found a correctness issue while experimenting with the proposed solution. Issue: https://github.com/apache/datafusion-comet/issues/2612 We had an existing test for `array_reverse` but it was only testing with inputs that Comet supports: ``` val fieldNames = table.schema.fields .filter(field => CometArrayReverse.isTypeSupported(field.dataType)) .map(_.name) for (fieldName <- fieldNames) { sql(s"SELECT $fieldName as a FROM t1") .createOrReplaceTempView("t2") checkSparkAnswer(sql("SELECT reverse(a) FROM t2")) } ``` Those cases work fine. The correctness issue happens when we fall back to Spark for the projection due to unsupported types. -- 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]
