[
https://issues.apache.org/jira/browse/TAJO-1517?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14394177#comment-14394177
]
Keuntae Park commented on TAJO-1517:
------------------------------------
For INTERSECT case, I think it can be implemented as semi join + distinct.
However, for INTERSECT ALL, the problem becomes more difficult
because it should count the number of intersecting rows even with the same row
value.
For example,
when query1 results in {2, 2, 2} while query2 results in {2, 2},
query1 INTERSECT query2 is {2} while query1 INTERSECT ALL query2 is {2, 2}.
So, I'm wondering about implementing new INTERSECT ALL Exec,
which is very similar to semi-join
except INTERSECT ALL Exec also accounts for the number of intersecting rows of
the same value.
> Implement INTERSECT
> -------------------
>
> Key: TAJO-1517
> URL: https://issues.apache.org/jira/browse/TAJO-1517
> Project: Tajo
> Issue Type: Sub-task
> Reporter: Keuntae Park
>
> {code}
> query1 INTERSECT [ALL] query2
> {code}
> INTERSECT returns all rows that are both in the result of query1 and in the
> result of query2.
> Duplicate rows are eliminated unless INTERSECT ALL is used.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
