[HIVE-20174] Vectorization: Fix NULL / Wrong Results issues in GROUP BY Aggregation Functions

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Component/s: Hive
• Labels: hive-3.2.0-candidate
• Assignee: Matt McCline
• Reporter: Matt McCline
• Created: 13/Jul/18 23:03
• Updated: 01/Dec/22 09:14
• Resolved: 22/Jul/18 08:18

Description

Write new UT tests that use random data and intentional isRepeating batches to checks for NULL and Wrong Results for vectorized aggregation functions.

 

BUGs found:

1) AVG/VARIANCE (family) in PARTIAL1 mode was returning NULL instead of count = 0, sum = 0 (All data types).  For AVG DECIMAL, only return NULL if there was an overflow.

2) AVG/MIN/MAX was not detecting repeated NULL correctly for the TIMESTAMP, INTERVAL_DAY_TIME, and String Family.  Eliminated redundant code.

3) Fix incorrect calculation  for VARIANCE (family) in PARTIAL2 and FINAL modes (HIVE-18758).

4) Fix row-mode AVG DECIMAL to enforce output type precision and scale in COMPLETE and FINAL modes.

 

Issue Links

• links to: RB #67920