SEZ9 commented on PR #11493: URL: https://github.com/apache/seatunnel/pull/11493#issuecomment-6568614682
Thanks for the follow-up on `925fb9c1fe0da5b6cd1b0a65b7fbcd036f22a16b`. Checking the earlier points against the current diff: - **F1 (STRING -> JSON change on upgrade)** – The source-level `handle_json_as_string` option providing a metadata-preserving STRING projection covers the opt-out. Could you confirm it is documented on each affected source connector page (Snowflake, Doris, StarRocks), not only in the incompatible-changes page, so users hitting this on upgrade can find it where they look first? - **F2 (JSON logical type is a plain String)** – Strict parsing with `FAIL_ON_TRAILING_TOKENS` in `RowToJsonConverters` gives the validity contract at the format boundary. It would help transform authors if the schema docs state explicitly that validation happens there rather than at construction. - **F3 (StarRocks JSON + default CSV Stream Load)** – The init-time rejection of JSON columns with the CSV format addresses this. Please make sure the StarRocks sink doc points users to the JSON load format for native JSON columns. - **F4 (Snowflake user-defined query skips PARSE_JSON)** – This looks like the remaining open item, and the Snowflake INSERT grammar omission has been noted as still present. Could you clarify how a user-supplied query targeting JSON-typed fields is handled now: is PARSE_JSON applied, is there a clear early error, or is this documented as a requirement on the user? Happy to take another look once F4 is clarified. <!-- streview-comment:2223 --> -- 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]
