Dave DeCaprio created SPARK-31982: ------------------------------------- Summary: Spark sequence doesn't handle date increments that cross DST Key: SPARK-31982 URL: https://issues.apache.org/jira/browse/SPARK-31982 Project: Spark Issue Type: Bug Components: SQL Affects Versions: 2.4.4 Reporter: Dave DeCaprio
Run the following query with the Spark default time zone set to "America/Chicago" {code:java} sequence(cast("2011-03-01" as date), cast("2011-05-01" as date), interval 1 month) {code} The result is: {noformat} [2011-03-01, 2011-03-31, 2011-04-30]{noformat} I've traced the code what happens is that during this month there is daylight savings and so the time ends up being 11pm on 3/31 instead of midnight on 4/1. If you run this with timezone set to GMT or run the sequence functions over months that don't have daylight savings moving forward, it works correctly. I think this is because the toLong and from long functions are not using a timezone but timestampAddInterval is. I think the solution is probably to make everything time zone aware, but maybe you can just always use GMT for timestmapAddInterval. -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org