ot;,"table":"todo_list","change_lsn":"0025:0608:0002","commit_lsn":"0025:0608:0003","event_serial_no":2},"op":"u","ts_ms":1652104513257,"transaction":null}
And finally a DELETE
,
'value.format'='debezium-json',
'value.debezium-json.schema-include'='false',
'value.fields-include'='EXCEPT_KEY'
);
SELECT * FROM todo_list;
Now, when I perform a query like this in the MS-SQL database:
UPDATE todo_list SET name='test2' WHERE id=3;
Now I see that the Flink-SQL client updates the row with id=3 to have the new
value "test2" for the 'name' field, as I was expecting. However, when I
duplicate the 'done' field to have a different value, Flink-SQL seems to leave
the old row with values (3, False, 'test2') intact, and shows a new row with
values (3, True, 'test2').
I tried to append a `PRIMARY KEY (k_id) NOT ENFORCED` line between the first
parentheses in the CREATE TABLE statement, but this seems to make no
difference, except when running `DESCRIBE todo_list` in Flink-SQL.
I have no idea why the boolean field would cause different behavior than the
text field. Am I missing some piece of configuration, are my expectations
wrong?
Regards,
Joost Molenaar
