c21 commented on pull request #29342: URL: https://github.com/apache/spark/pull/29342#issuecomment-671497515
@cloud-fan, @viirya - I just thought a bit more about it, and I think we need some tracking per row, but not per key. When doing full outer join, not only the join keys from both side should be matched, but the `condition` needs to be passed, e.g. in current implementation - https://github.com/apache/spark/pull/29342/files#diff-db4ffe4f0196a9d7cf1f04c350ee3381R139 . So from `ShuffledHashJoinExec` after checking join condition, we need have a way to `markRowLookedUp` to call into `HashedRelation`. I am not sure if a bitset for values of `BytesToBytesMap` (not for key, but for values) is feasible. Do you guys have any thoughts? Thanks. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
