[
https://issues.apache.org/jira/browse/PHOENIX-944?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14177137#comment-14177137
]
Maryann Xue commented on PHOENIX-944:
-------------------------------------
The ClientGroupedAggregatingResultIterator is different from
GroupedAggregatingResultIterator not because of the post processing, but in
that it pipelines the input tuple through both server aggregators and client
aggregators, which means it combines the work of both
GroupedAggregatorRegionObserver and GroupedAggregatingResultIterator.
ClientScanPlan is not a good name, but simply handles
non-aggregate/non-distinct outer queries, like "select ... from (select ...
from XX limit 10) where ... order by ...". In this case, even it's not
two-level aggregation, it still cannot be flattened, so the outer query simply
iterate the inner result and do post steps.
Yes, they look similar, but think ClientAggegatePlan should not derive from
BaseQueryPlan for it actually includes another inner plan.
> Support derived tables in FROM clause that needs extra steps of client-side
> aggregation or other processing
> -----------------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-944
> URL: https://issues.apache.org/jira/browse/PHOENIX-944
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 3.0.0, 4.0.0, 5.0.0
> Reporter: Maryann Xue
> Assignee: Maryann Xue
> Fix For: 3.0.0, 4.0.0, 5.0.0
>
> Attachments: 944.patch
>
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> Groupby in both outer and inner queries cannot be flattened. We can apply an
> extra step of client-side aggregation to handle such cases.
> See DerivedTableIT.java for all those examples that are currently not
> supported.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
