[
https://issues.apache.org/jira/browse/HIVE-27293?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Riju Trivedi updated HIVE-27293:
--------------------------------
Description:
Attached repro.q file and data file used to reproduce the issue.
{code:java}
Insert overwrite table etarget
select mt.*, floor(rand() * 1) as bdata_no from (select nvl(np.client_id,'
'),nvl(np.id_enddate,cast(0 as decimal(10,0))),nvl(np.client_gender,'
'),nvl(np.birthday,cast(0 as decimal(10,0))),nvl(np.nationality,'
'),nvl(np.address_zipcode,' '),nvl(np.income,cast(0 as
decimal(15,2))),nvl(np.address,' '),nvl(np.part_date,cast(0 as int)) from
(select * from esource where part_date = 20230414) np) mt;
{code}
Outcome:
{code:java}
select client_id,birthday,income from etarget;
15678 0 0.00
67891 19313 -1.00
12345 0 0.00{code}
Expected Result :
{code:java}
select client_id,birthday,income from etarget;
12345 19613 -1.00
67891 19313 -1.00
15678 0 0.00{code}
Disabling hive.vectorized.use.vectorized.input.format produces correct output.
was:
Attached repro.q file and data file used to reproduce the issue.
{code:java}
Insert overwrite table etarget
select mt.*, floor(rand() * 1) as bdata_no from (select nvl(np.client_id,'
'),nvl(np.id_enddate,cast(0 as decimal(10,0))),nvl(np.client_gender,'
'),nvl(np.birthday,cast(0 as decimal(10,0))),nvl(np.nationality,'
'),nvl(np.address_zipcode,' '),nvl(np.income,cast(0 as
decimal(15,2))),nvl(np.address,' '),nvl(np.part_date,cast(0 as int)) from
(select * from esource where part_date = 20230414) np) mt;
{code}
Outcome:
{code:java}
select client_id,birthday,income from etarget;
889004570706 0 0.00
889004570838 19880313 -1.00
889005389931 0 0.00 {code}
Expected Result :
{code:java}
select client_id,birthday,income from etarget;
889004570706 0 0.00
889004570838 19880313 -1.00
889005389931 19880613 -1.00 {code}
Disabling hive.vectorized.use.vectorized.input.format produces correct output.
> Vectorization: Incorrect results with nvl for ORC table
> -------------------------------------------------------
>
> Key: HIVE-27293
> URL: https://issues.apache.org/jira/browse/HIVE-27293
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 4.0.0-alpha-2
> Reporter: Riju Trivedi
> Priority: Major
> Attachments: vectorization_nvl.q
>
>
> Attached repro.q file and data file used to reproduce the issue.
> {code:java}
> Insert overwrite table etarget
> select mt.*, floor(rand() * 1) as bdata_no from (select nvl(np.client_id,'
> '),nvl(np.id_enddate,cast(0 as decimal(10,0))),nvl(np.client_gender,'
> '),nvl(np.birthday,cast(0 as decimal(10,0))),nvl(np.nationality,'
> '),nvl(np.address_zipcode,' '),nvl(np.income,cast(0 as
> decimal(15,2))),nvl(np.address,' '),nvl(np.part_date,cast(0 as int)) from
> (select * from esource where part_date = 20230414) np) mt;
> {code}
> Outcome:
> {code:java}
> select client_id,birthday,income from etarget;
> 15678 0 0.00
> 67891 19313 -1.00
> 12345 0 0.00{code}
> Expected Result :
> {code:java}
> select client_id,birthday,income from etarget;
> 12345 19613 -1.00
> 67891 19313 -1.00
> 15678 0 0.00{code}
> Disabling hive.vectorized.use.vectorized.input.format produces correct output.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
