[HIVE-19354] from_utc_timestamp returns incorrect results for datetime values with timezone

• Type: Bug
• Status: Patch Available
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.1.0
• Component/s: Hive
• Assignee: Unassigned
• Reporter: Bruce Robbins
• Created: 29/Apr/18 03:03
• Updated: 09/Nov/18 23:02

Description

On the master branch, from_utc_timestamp returns incorrect results for datetime strings that contain a timezone:

hive> select from_utc_timestamp('2000-10-10 00:00:00+00:00', 'America/Los_Angeles');
OK
2000-10-09 10:00:00
Time taken: 0.294 seconds, Fetched: 1 row(s)
hive> select from_utc_timestamp('2000-10-10 00:00:00', 'America/Los_Angeles');
OK
2000-10-09 17:00:00
Time taken: 0.121 seconds, Fetched: 1 row(s)
hive> 

Both inputs are 2000-10-10 00:00:00 in UTC time, but I got two different results.

In version 2.3.3, from_utc_timestamp doesn't accept timezones in its input strings, so it does not have this bug:

hive> select from_utc_timestamp('2000-10-10 00:00:00+00:00', 'America/Los_Angeles');
OK
NULL
Time taken: 5.152 seconds, Fetched: 1 row(s)
hive> select from_utc_timestamp('2000-10-10 00:00:00', 'America/Los_Angeles');
OK
2000-10-09 17:00:00
Time taken: 0.069 seconds, Fetched: 1 row(s)
hive> 

Since the function is expecting a UTC datetime value, it probably should continue to reject input that contains a timezone component.