Describe the bug
These SQL functions produce a valid RAL but return wrong results i.e. the BSQL engine technically supports these functions but in reality, the results are not correct.
Steps/Code to reproduce bug
import blazingsql as bsql
import cudf
df = cudf.DataFrame({'datetime': [
np.datetime64('2020-10-15T10:58:02')
, np.datetime64('2020-10-01T10:02:23')
, np.datetime64('2020-09-30T14:36:26')
, np.datetime64('2020-10-10T08:34:36')
, np.datetime64('2020-10-09T08:34:40')
, np.datetime64('2020-10-03T03:31:21')
, np.datetime64('2020-10-05T03:21:28')
]})
bc = bsql.BlazingContext()
bc.create_table('dates_table', df)
bc.sql('''
SELECT datetime,
TIMESTAMP'2020-11-10 12:00:01' AS datetime2,
TIMESTAMPDIFF(DAY, datetime, TIMESTAMP'2020-11-10 12:00:01') AS diff,
TIMESTAMPADD(HOUR, 2, datetime) AS add
FROM dates_table
''')
The above produces the following table
datetime
datetime2
diff
add
2020-10-15 10:58:02
2020-11-10 12:00:01
1396630809
2021-01-06 18:58:02
2020-10-01 10:02:23
2020-11-10 12:00:01
1396630809
2020-12-23 18:02:23
2020-09-30 14:36:26
2020-11-10 12:00:01
1396630809
2020-12-22 22:36:26
2020-10-10 08:34:36
2020-11-10 12:00:01
1396630809
2021-01-01 16:34:36
2020-10-09 08:34:40
2020-11-10 12:00:01
1396630809
2020-12-31 16:34:40
2020-10-03 03:31:21
2020-11-10 12:00:01
1396630809
2020-12-25 11:31:21
2020-10-05 03:21:28
2020-11-10 12:00:01
1396630809
2020-12-27 11:21:28
Expected behavior
Proper values are returned
datetime
datetime2
diff
add
2020-10-15 10:58:02
2020-11-10 12:00:01
26
2020-10-15 12:58:02
2020-10-01 10:02:23
2020-11-10 12:00:01
40
2020-10-01 12:02:23
2020-09-30 14:36:26
2020-11-10 12:00:01
40
2020-09-30 16:36:26
2020-10-10 08:34:36
2020-11-10 12:00:01
31
2020-10-10 10:34:36
2020-10-09 08:34:40
2020-11-10 12:00:01
32
2020-10-09 10:34:40
2020-10-03 03:31:21
2020-11-10 12:00:01
38
2020-10-03 05:31:21
2020-10-05 03:21:28
2020-11-10 12:00:01
36
2020-10-05 05:21:28
Environment overview (please complete the following information)
Describe the bug These SQL functions produce a valid RAL but return wrong results i.e. the BSQL engine technically supports these functions but in reality, the results are not correct.
Steps/Code to reproduce bug
The above produces the following table
Expected behavior Proper values are returned
Environment overview (please complete the following information)
'ba4a97a3ce3f1c733a135bf88518f2f67b12b519'