[GitHub] [spark] HeartSaVioR commented on pull request #28831: [SPARK-31993][SQL] Don't split code blocks in generated code for 'concat_ws' for mixed string/array types of columns

2020-06-15 Thread GitBox
HeartSaVioR commented on pull request #28831: URL: https://github.com/apache/spark/pull/28831#issuecomment-644120378 My bad I missed the existing test, as I target on older Spark version originally. The existing UT works with this patch, because we don't create local variables for dealing

[GitHub] [spark] HeartSaVioR commented on pull request #28831: [SPARK-31993][SQL] Don't split code blocks in generated code for 'concat_ws' for mixed string/array types of columns

2020-06-15 Thread GitBox
HeartSaVioR commented on pull request #28831: URL: https://github.com/apache/spark/pull/28831#issuecomment-643978420 There might be still a chance to compose these all parts into one and pass to splitExpressionsWithCurrentInputs, but it requires rewrite of code because there're two