[
https://issues.apache.org/jira/browse/HIVE-11865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jesus Camacho Rodriguez updated HIVE-11865:
-------------------------------------------
Attachment: HIVE-11865.03.patch
New patch contains the following parts:
- Disabling Hive PPD. It was just necessary to keep a small part of the code
that is responsible for pushing Filter predicates to TableScan operators
(SimplePredicatePushDown).
- Disabling Hive inference for _isnotnull_ predicates on equi-join inputs. This
was done in SemanticAnalyzer, and it is not necessary anymore when we run
purely through Calcite.
- It introduces a new rule in Calcite that pushes Filter through Sort operator.
This was present in Hive, but it was missing on the Calcite side.
- It includes logic related to pushing Filter down when return path was on.
This should have been added when HIVE-11110 went it, but it was difficult to
detect as Hive PPD was doing the work for us.
I already went through the changes in the q files: they are either changes in
the order of Filter predicate factors, or removal of redundant _isnotnull_
factors. I will post the patch to RB for review.
[~jpullokkaran], [~ashutoshc], could you take a look? Thanks
> Disable Hive PPD optimizer when CBO has optimized the plan
> ----------------------------------------------------------
>
> Key: HIVE-11865
> URL: https://issues.apache.org/jira/browse/HIVE-11865
> Project: Hive
> Issue Type: Bug
> Components: CBO, Logical Optimizer
> Affects Versions: 2.0.0
> Reporter: Jesus Camacho Rodriguez
> Assignee: Jesus Camacho Rodriguez
> Attachments: HIVE-11865.01.patch, HIVE-11865.02.patch,
> HIVE-11865.02.patch, HIVE-11865.03.patch, HIVE-11865.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
