nzw921rx commented on PR #13161:
URL: https://github.com/apache/seatunnel/pull/13161#issuecomment-6916561640

   ### P2: Cover the actual JDBC timestamp binding in the regression test
   
   The new timestamp tests exercise `nextAddTimeMicros()` and `formatAddTime()` 
directly, without going through `write()` and `processCdcRow()`.
   
   I verified this gap in an isolated copy: reverting the actual binding to the 
old `setTimestamp()` call still leaves all 13 `DatabendSinkWriterTest` tests 
passing. The E2E test does not force the events to share the same millisecond 
either.
   
   Please extend the existing writer test to exercise the real `write()` path 
with a mocked `PreparedStatement` and controlled time. Capture parameter 4 and 
assert that:
   
   - It is bound through `setString()`.
   - Microsecond precision is preserved.
   - Consecutive events within the same millisecond receive strictly increasing 
timestamps.
   
   The regression test should fail if the binding is reverted to 
`setTimestamp()`.
   
   ### P2: Qualify the ordering guarantee in the documentation
   
   The EN/ZH incompatible-change entries currently state that the latest event 
from one writer in the same merge always wins.
   
   However, `formatAddTime()` renders timestamps in the JVM’s local time zone. 
The new DST test explicitly demonstrates `01:59:59` followed by `01:00:00`. 
Since the merge orders by the stored `add_time DESC`, the later event can still 
lose during that transition.
   
   Please qualify this statement in both language versions and document the 
local-time/DST limitation. This is an existing limitation; clarifying the 
documentation is sufficient for this PR.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to