[HIVE-5660] Hive incorrecly handles data overflow/underflow

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.10.0, 0.11.0, 0.12.0
• Component/s: Types
• Assignee: Xuefu Zhang
• Reporter: Xuefu Zhang
• Created: 26/Oct/13 02:48
• Updated: 26/Oct/13 03:29

Description

Unlike other databases, Hive has only one server mode w.r.t error handling, in which NULL value is placed in case of errors such as divide-by-zero and data overflow/underflow. However, it appears that hive isn't consistent in this prospect. In case of data overflow or underflow, Hive returns a value that wraps around, which can produce erroneous result without user noticing it. For instance:

hive> select 2147483640 + 2147483645 from tmp2 limit 1;
...
OK
-11
Time taken: 6.549 seconds, Fetched: 1 row(s)

To be consistent in terms of error handling, Hive should produce NULL value instead.

Issue Links

• is related to: HIVE-5656 Hive produces unclear, confusing SemanticException when dealing with mod or pmod by zero
• is related to: HIVE-5438 Introduce different server modes for error handling in Hive server2
• is related to: HIVE-5655 Hive incorrecly handles divide-by-zero case