Loss of Precision in Arithmetic Operations on BIGINT Data Type.
To Reproduce
Assume that we execute the following statement under a database named testdb.
DROP TABLE IF EXISTS t1;
CREATE TABLE t1(time TIMESTAMP, c1 BIGINT);
INSERT INTO t1(time, c1) VALUES (1641024000000, 9000000000000000000);
# query 1
SELECT c1 FROM t1 WHERE 9000000000000000001 <= c1;
# query 2
SELECT c1 FROM t1 WHERE 9000000000000000001 <= c1 + CAST(0 as BIGINT);
Expected Behavior
Query 1 returned result set: empty set
Query 2 returned result set: empty set
Actual behaviour
Query 1 returned result set: empty set
Query 2 returned result set: 9000000000000000000
Environment
OS：Ubuntu Server 22.04 LTS 64bit
TDengine Version：3.3.4.3
Additional Context
Hello, TDengine team. I found that arithmetic operations on the BIGINT type fail to meet the precision requirement of 2^63-1 supported by the BIGINT type, which might be a logical error.
LingweiKuang
commented
3 days ago
Bug Description
Loss of Precision in Arithmetic Operations on BIGINT Data Type.
To Reproduce
Assume that we execute the following statement under a database named testdb.
Expected Behavior
Query 1 returned result set: empty set
Query 2 returned result set: empty set
Actual behaviour
Query 1 returned result set: empty set
Query 2 returned result set: 9000000000000000000
Environment
Additional Context
Hello, TDengine team. I found that arithmetic operations on the BIGINT type fail to meet the precision requirement of 2^63-1 supported by the BIGINT type, which might be a logical error.