It'd be awesome if event_transforms (the refine UDF in Scala that seems to be used for transforming events into the tables in Hive that I've been working with via Spark) would also compute isp_data. This would help in early prototype work around T364164.