SEZ9 commented on PR #11493:
URL: https://github.com/apache/seatunnel/pull/11493#issuecomment-6207402839

   Status looks accurate: the only change on `58d1cf0d19d` is the typed 
`Mockito.<TableSchema>any()` in the two `JdbcSinkWriterTest` stubs, so all four 
findings from the previous round remain open. Noting what I'd like to see in 
the next push so re-review is quick:
   
   - **F1 (compatibility):** Either a config-level opt-out that keeps Snowflake 
`VARIANT` / Doris `JSON`/`JSONB` / StarRocks `JSON` mapped to `STRING`, or, if 
you decide the new mapping is unconditional, an explicit entry in 
`docs/en/introduction/concepts/incompatible-changes.md` describing the upgrade 
impact and the migration path. Please say which route you are taking.
   - **F2 (logic):** A clear statement in 
`docs/en/introduction/concepts/schema-feature.md` of what the JSON logical type 
guarantees. If validity is the sink's responsibility, say so; if it is a 
property of the type, the validation needs to live somewhere other than only 
the StarRocks load boundary.
   - **F3 (robustness):** Fail-fast at StarRocks sink initialization when the 
target table has native JSON columns and the Stream Load format is the default 
CSV, rather than surfacing the rejection at load time.
   - **F4 (robustness):** For the Snowflake sink with a user-defined query, 
either apply `PARSE_JSON` wrapping for JSON fields or fail at sink 
initialization with an actionable message, instead of letting the server reject 
mid-run.
   
   On CI: merging current `dev` to pick up the `Dead links` and `unit-test (8, 
ubuntu-latest)` fixes before the next push is the right call, since fail-fast 
on run 38351233448 meant the E2E lanes never executed on this head. Please 
include the E2E outcome (Doris, StarRocks, Snowflake and the format tests) in 
the ping when the code is up, and I will do the full re-review then.
   
   <!-- streview-comment:1792 -->


-- 
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