Github user dilipbiswal commented on the issue: https://github.com/apache/spark/pull/22198 @maropu @gatorsmile I do have a question on the semantics. ```SQL use hint; explain extended SELECT /*+ BROADCASTJOIN(hint.s2) */ * FROM s1, s2 where s1.c1 = s2.c1; ``` In this case, aren't we supposed to apply the hint ? even though s2 is not explicitly qualified with the database ? Here is the optimized plan i see .. ``` *(5) SortMergeJoin [c1#30], [c1#32], Inner :- *(2) Sort [c1#30 ASC NULLS FIRST], false, 0 : +- Exchange hashpartitioning(c1#30, 200) : +- *(1) Filter isnotnull(c1#30) : +- HiveTableScan [c1#30, c2#31], HiveTableRelation `hint`.`s1`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [c1#30, c2#31] +- *(4) Sort [c1#32 ASC NULLS FIRST], false, 0 +- Exchange hashpartitioning(c1#32, 200) +- *(3) Filter isnotnull(c1#32) +- HiveTableScan [c1#32, c2#33], HiveTableRelation `hint`.`s2`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [c1#32, c2#33] ```
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org