[
https://issues.apache.org/jira/browse/PHOENIX-4966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16648295#comment-16648295
]
Josh Mahonin commented on PHOENIX-4966:
---------------------------------------
[~tdsilva] I'm a bit worried about relying on a toString() implementation to
verify the query plan, but if it's the only way to get at it, it's better than
nothing. Do you know of any differences in the representation between different
Spark versions?
> Implement unhandledFilters in PhoenixRelation so that spark only evaluates
> filters when required
> ------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-4966
> URL: https://issues.apache.org/jira/browse/PHOENIX-4966
> Project: Phoenix
> Issue Type: Improvement
> Affects Versions: 4.14.0
> Reporter: Thomas D'Silva
> Assignee: Thomas D'Silva
> Priority: Major
> Labels: spark
> Fix For: 4.15.0, 5.1.0
>
> Attachments: PHOENIX-4966.patch
>
>
> PhoenixRelation does not implement unhandledFilters so the data returned from
> a scan is run through filters twice.
> {code}
> Project [ID#7L]
> +- Filter (((isnotnull(COL1#8) && isnotnull(ID#7L)) && (COL1#8 = test_row_1))
> && (ID#7L = 1))
> +- Scan PhoenixRelation(TABLE1,localhost:50068:/hbase,false)
> [ID#7L,COL1#8] PushedFilters: [IsNotNull(COL1), IsNotNull(ID),
> EqualTo(COL1,test_row_1), EqualTo(ID,1)], ReadSchema: struct<ID:bigint>
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
