Github user hhbyyh commented on the issue: https://github.com/apache/spark/pull/19439 @thunterdb @WeichenXu123 Let's keep only Array[Byte] for now. @WeichenXu123 for the origin column. Surely it maybe handy in some scenarios, but I'm most concerned about the object blending and uncertainty brought by this field (Some data may have it, while others don't), which can further bring out the problems I listed in my last comment. So my suggestion is to separate "origin" info out of the image column and serve it as an independent column.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org