forked from prestodb/presto
-
Notifications
You must be signed in to change notification settings - Fork 21
Open
Description
Teradata Presto release appears to be deserializing the timestamp column incorrectly in ORC files with time zone information. (ORC written from Hive 1.2.1)
Note the query returns correctly in standard Presto and in Hive. The table was built via Hive, transforming the first column, a string, into a timestamp column; the local time zone of Hive/Presto is America/Los_Angeles.
Teradata Presto 0.167-t.0.2:
time | tm | _col2
---------------------+-------------------------+---------------------------------------------
01-30-2016-00:11:02 | 2016-01-30 08:11:02.000 | 2016-01-30 07:11:02.000 America/Los_Angeles
01-30-2016-00:39:28 | 2016-01-30 08:39:28.000 | 2016-01-30 07:39:28.000 America/Los_Angeles
Mini cluster (0.17x)
time | tm | _col2
---------------------+-------------------------+---------------------------------------------
01-30-2016-00:11:02 | 2016-01-30 00:11:02.000 | 2016-01-30 00:11:02.000 America/Los_Angeles
01-30-2016-00:39:28 | 2016-01-30 00:39:28.000 | 2016-01-30 00:39:28.000 America/Los_Angeles
presto:mesoads> select time,tm, tm at time zone 'America/Los_Angeles' from orc_table
Metadata
Metadata
Assignees
Labels
No labels