Github user jainaks commented on the issue: https://github.com/apache/spark/pull/21320 Thanks @mallman for making this huge contribution. 3 years is really a long time to keep patience for concluding things. I am attaching the sample parquet file for your reference with which you can reproduce the Window function, wrong column selection issue. [sample.parquet.txt](https://github.com/apache/spark/files/2230873/sample.parquet.txt) Please remove .txt from the filename. Following are the simple steps you can follow to reproduce this issue via spark shell. ``` import org.apache.spark.sql.SparkSession val ss = SparkSession.builder().config("spark.sql.nestedSchemaPruning.enabled", "true").getOrCreate() val inputdf = ss.read.parquet("sample.parquet") inputdf.createOrReplaceTempView(âtemptableâ) ss.sql("select page.url, first(page.url) over (partition by id order by timestamp rows between current row and unbounded following) from temptable").collect.foreach(println) ``` Result: `[https://adobeid-na1.services.adobe.com/renga-idprovider/pages/login,Account:IMS:onLoad_SignInForm]` Please let me know if you need any help from my side. PS: Sorry for responding late to it. Few high priority items kept me busy.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org