zhuxiangyi commented on PR #10203:
URL: https://github.com/apache/paimon/pull/10203#issuecomment-5997585030
Thanks @JingsongLi for the review!
This didn't come from a production issue. I ran into it while addressing
your signed-zero P1 on #10144: the top-level column turned out to lose the row
too, and I then confirmed it on master with plain SQL, e.g. on Spark:
```sql
CREATE TABLE t (id INT, d DOUBLE) USING paimon;
INSERT INTO t VALUES (1, CAST('-0.0' AS DOUBLE));
SELECT id FROM t WHERE d = 0.0; -- returns nothing
```
You're right that NaN and ±0 are handled inconsistently in many places in
Paimon, so fixing just this one spot has limited value, and the first version
of this PR did introduce the Flink issue you found. If you don't think this
edge case is worth the added complexity, I can close this PR and record the
problem in an issue for reference. If you think it's worth fixing, I'm happy to
keep adjusting it based on your feedback.
--
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]
