[IMPALA-989] Average on decimals is not correct.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 1.4
• Fix Version/s: Impala 1.4
• Labels: decimal
• Assignee: Nong Li
• Reporter: Ishaan Joshi
• Created: 09/May/14 01:10
• Updated: 20/Dec/15 00:05
• Resolved: 11/May/14 19:34

Description

Computing Average on a set of decimal results is not correct.

mysql:

mysql> select sum(float_col_4) from table_1;
+-----------------------+
| sum(float_col_4)      |
+-----------------------+
| 319391280635.61476055 |
+-----------------------+
1 row in set (0.00 sec)

mysql> select avg(float_col_4) from table_1;
+-----------------------+
| avg(float_col_4)      |
+-----------------------+
| 31939128.063561476055 |
+-----------------------+
1 row in set (0.00 sec)

impala:

[localhost:21000] > select sum(float_col_4) from table_1;
Query: select sum(float_col_4) from table_1
+-----------------------+
| sum(float_col_4)      |
+-----------------------+
| 319391280635.61476055 |
+-----------------------+
Returned 1 row(s) in 0.55s
[localhost:21000] > select avg(float_col_4) from table_1;
Query: select avg(float_col_4) from table_1
+--------------------------------------+
| avg(float_col_4)                     |
+--------------------------------------+
| -47414.42700673951055721309858621187 |
+--------------------------------------+
Returned 1 row(s) in 0.56s

Division returns the right scale and precision (there is a unit test to confirm this), so I'm not sure what's going on.