[HIVE-24435] Vectorized unix_timestamp is inconsistent with non-vectorized counterpart

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Labels: pull-request-available
• Assignee: Zoltan Haindrich
• Reporter: Zoltan Haindrich
• Created: 26/Nov/20 11:33
• Updated: 17/Nov/22 08:47
• Resolved: 03/Dec/20 11:07

Description

create table t (d string);
insert into t values('2020-11-16 22:18:40 UTC');

select
  '>' || d || '<' , unix_timestamp(d), from_unixtime(unix_timestamp(d)), to_date(from_unixtime(unix_timestamp(d)))
from t
;

set hive.fetch.task.conversion=none;

select
  '>' || d || '<' , unix_timestamp(d), from_unixtime(unix_timestamp(d)), to_date(from_unixtime(unix_timestamp(d)))
from t
;

results:

-- std udf:
>2020-11-16 22:18:40 UTC<       1605593920      2020-11-16 22:18:40     2020-11-16
-- vectorized udf
>2020-11-16 22:18:40 UTC<       NULL    NULL    NULL

Issue Links

• links to: GitHub Pull Request #1713