hashAggregates: group by sum decimal issue #1554
Description
Aggregating negative decimal values produces incorrect results.
table:
| charge_item_name (string) | charge_type (string) | charge_amount (decimal(16,2) |
|---|---|---|
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -240.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
| WHT | tax | -200.00 |
| WHT | tax | -300.00 |
| WHT | tax | -300.00 |
Spark SQL query:
SELECT itemchargingtype, itemdescription, count(*) AS `count`, sum(itemamount) AS `sum`
FROM `transaction_items`
where itemdescription = 'WHT'
GROUP BY itemdescription, itemchargingtype
result:
| charge_type | charge_item_name | count | sum |
|---|---|---|---|
| tax | WHT | 14 | 1,202.88 |
After disabling
snappydata.sql.hashAggregateSize=-1
snappydata.sql.useOptimizedHashAggregateForSingleKey=false
it produces correct values.
result:
| charge_type | charge_item_name | count | sum |
|---|---|---|---|
| tax | WHT | 14 | -4040.00 |