Github user mgaido91 commented on the issue: https://github.com/apache/spark/pull/22364 @maropu anyway I checked and that is the only other places where this pattern happens. So I am ok including it here. The point is that there the situation is a bit different, ie. it is not an `(AttributeSet -- AttributeSet).nonEmpty` case, but it is a `(AttributeSet -- Seq[Attribute]).nonEmpty`. So I am not sure the performance gain/difference will be significant in this case. I'll try and do some benchmarks and I'll do the change if I see a significant difference. Thanks.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org