[
https://issues.apache.org/jira/browse/HIVE-1750?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Siying Dong updated HIVE-1750:
------------------------------
Attachment: HIVE-1750.3.patch
1. fix the bug of opOr -> opAnd
2. remove walkEvalExpr, as it is not really needed
3. Instead of removing operator after walking its expression tree, save the
operator and remove them one by one after walking the operation tree. It is to
handle the case like:
from src_tbl
insert overwrite table tbl1 where ds='1'
insert overwrite table tbl2 where ds='1';
Removing operators when walking the tree is dangerous in these cases.
4. fix comments and add some. Fix variable name.
5. rename classes for partition condition remover to be different from ppr
6. add some queries in the test: 1. cover opAnd case; 2. cover multiple insert
from the same partition
7. always put pruned partitions into the hash map. (we cannot remove to calling
calling partition pruners multiple places for PartitionConditionRemover is not
guaranteed to prune partitions, for cases like table is not partitioned)
8. PartitionPruner and PartitionConditonRemover to share some codes for
evaluating the expression with partition columns
9. some other code cleaning up
ran the test pcr.q but still running the whole test suites. Will "submit patch"
when other tests pass.
> Remove Partition Filtering Conditions when Possible
> ---------------------------------------------------
>
> Key: HIVE-1750
> URL: https://issues.apache.org/jira/browse/HIVE-1750
> Project: Hive
> Issue Type: Improvement
> Reporter: Siying Dong
> Assignee: Siying Dong
> Attachments: HIVE-1750.1.patch, HIVE-1750.2.patch, HIVE-1750.3.patch
>
>
> For some simple queries, partition filtering constraints take 8% of CPU time
> (now 16% since we filter twice) even if the result is always true. When
> possible, we should remove these constraints to save CPU times.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
