[
https://issues.apache.org/jira/browse/HIVE-17775?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Shelukhin updated HIVE-17775:
------------------------------------
Description:
If a PTF partitions-by a partition column of the table (or a bucketing column,
although that could be more difficult), or orders-by a partition column or a
column on which the table is already sorted being bucketed+sorted, it should
take advantage of these properties of the table.
That can turn some PTF operators (that I'm presently observing on a real query
from a user) into a single scan, or at least a much simpler DAG; as it is, the
query is very slow.
was:
When the user is running a PTF that partitions by a partition column of the
table (or a bucketing column, although that could be more difficult), or orders
by a partition column, or a column on which the table is already sorted being
bucketed+sorted, the PTF should take advantage of these.
That can turn some PTF operators (that I'm presently observing on a real query
from a user) into a single scan, or at least a much simpler DAG; as it is, the
query is very slow.
> PTF should take advantage of partitioning, bucketing and sorting
> ----------------------------------------------------------------
>
> Key: HIVE-17775
> URL: https://issues.apache.org/jira/browse/HIVE-17775
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
>
> If a PTF partitions-by a partition column of the table (or a bucketing
> column, although that could be more difficult), or orders-by a partition
> column or a column on which the table is already sorted being
> bucketed+sorted, it should take advantage of these properties of the table.
> That can turn some PTF operators (that I'm presently observing on a real
> query from a user) into a single scan, or at least a much simpler DAG; as it
> is, the query is very slow.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
