Github user maropu commented on the issue:
https://github.com/apache/spark/pull/22198
Thanks for the sum-up. I like simpler one, too. Le me just describe more to
make me more understood; IIUC we have the two case: (1) fully-qualified case
`/*+ MAPJOIN(testDb.t) */` and (2) non-qualified case `/*+ MAPJOIN(t) */`.
(1) no ambiguity, as @dongjoon-hyun said, we just exactly map `testDb.t` to
`testDb.t`
(2) the case @dongjoon-hyun worry about ,right?
Since I think most users meet this case (2) (they don't add database names
there in most case I probably think...), IMHO it is important to support the
syntax for usability. Based on the thought, my proposal is that we handle `/*+
MAPJOIN(t) */` as `/*+ MAPJOIN(*.t) */`to pick up all the matched tables in
views and tables. Then, we print warning message for users like this:
https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/Column.scala#L265
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
