zhengjiewen created FLINK-23761:
-----------------------------------
Summary: FlinkSQL1.12 BatchMode Group By with INT type can not
produce the correct result
Key: FLINK-23761
URL: https://issues.apache.org/jira/browse/FLINK-23761
Project: Flink
Issue Type: Bug
Components: Table SQL / Planner
Affects Versions: 1.12.2
Reporter: zhengjiewen
Attachments: image-2021-08-13-16-08-31-970.png
When Run a Flink Sql which have Group By oprator, and filed type is INT
type，the result is incorrect.
I try to CAST the field to BIGINT type OR STRING type, the result is correct.
the SQL is folllow: 【receipt_year ，receipt_month {color:#172b4d}is INT
type{color} 】
{code:java}
" SELECT " +
" ware_id, " +
" supplier_id," +
" supplier_name," +
" receipt_year as years, " +
" receipt_month as months, " +
" SUM(COALESCE(receipt_count,0) * COALESCE(price,0.0)) /
SUM(COALESCE(receipt_count,0)) as supply_avg_price" +
" FROM " +
"
`kudu`.`default_database`.`impala::cube_kudu.dwd_storage_purchase_receipt` " +
" WHERE " +
" receipt_date BETWEEN TO_TIMESTAMP('"+ beginTime +"')
AND TO_TIMESTAMP('"+ endTime +"') " +
" GROUP BY " +
" ware_id, supplier_id, supplier_name, receipt_year ,
receipt_month "
{code}
{color:#172b4d}Here is the execution plan:{color}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
