hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sergio Peña (JIRA) <j...@apache.org>
Subject [jira] [Commented] (HIVE-13039) BETWEEN predicate is not functioning correctly with predicate pushdown on Parquet table
Date Thu, 11 Feb 2016 20:44:18 GMT

    [ https://issues.apache.org/jira/browse/HIVE-13039?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15143475#comment-15143475
] 

Sergio Peña commented on HIVE-13039:
------------------------------------

Thanks. Those tests are good.
Btw, {{testFilterFloatColumns}} is failing:

{noformat}
Expected :and(and(not(eq(a, null)), not(and(lt(a, 20.3), not(lteq(a, 10.2))))), not(or(or(eq(b,
1), eq(b, 2)), eq(b, 3))))
Actual   :and(and(not(eq(a, null)), not(and(lteq(a, 20.3), not(lt(a, 10.2))))), not(or(or(eq(b,
1), eq(b, 2)), eq(b, 3))))
{noformat}

It's related to your change. I did not notice that we're using a {{between}} call there as
well.

> BETWEEN predicate is not functioning correctly with predicate pushdown on Parquet table
> ---------------------------------------------------------------------------------------
>
>                 Key: HIVE-13039
>                 URL: https://issues.apache.org/jira/browse/HIVE-13039
>             Project: Hive
>          Issue Type: Bug
>          Components: Physical Optimizer
>    Affects Versions: 1.2.1, 2.0.0
>            Reporter: Yongzhi Chen
>            Assignee: Yongzhi Chen
>         Attachments: HIVE-13039.1.patch, HIVE-13039.2.patch
>
>
> BETWEEN becomes exclusive in parquet table when predicate pushdown is on (as it is by
default in newer Hive versions). To reproduce(in a cluster, not local setup):
> CREATE TABLE parquet_tbl(
>   key int,
>   ldate string)
>  PARTITIONED BY (
>  lyear string )
>  ROW FORMAT SERDE
>  'org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe'
>  STORED AS INPUTFORMAT
>  'org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat'
>  OUTPUTFORMAT
>  'org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat';
> insert overwrite table parquet_tbl partition (lyear='2016') select
>   1,
>   '2016-02-03' from src limit 1;
> set hive.optimize.ppd.storage = true;
> set hive.optimize.ppd = true;
> select * from parquet_tbl where ldate between '2016-02-03' and '2016-02-03';
> No row will be returned in a cluster.
> But if you turn off hive.optimize.ppd, one row will be returned.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message