SEZ9 commented on issue #12590: URL: https://github.com/apache/seatunnel/issues/12590#issuecomment-5945992518
Thanks for the detailed report and clear reproduction. Preserving an unknown (null) length for BIT/ENUM so the sink falls back to its unbounded string type (LONGTEXT on MySQL) sounds like the right direction; please open the PR with the converter fix plus the unit tests (null/zero/positive length, ENUM array fallback) and the JdbcMysqlIT case you mentioned. In the PR, please document the behavior change: existing destination tables created with VARCHAR(1)/VARCHAR(255) will not be altered automatically and need a manual schema migration. Also confirm that a positive length reported by DuckDB metadata (if any) is still honored rather than always widening to unbounded. Could you confirm whether other sinks with auto-DDL (e.g. PostgreSQL, Oracle, Doris/StarRocks) behave correctly when the source string length is null, so the fix does not regress them? A quick check or an extra IT for at least one more sink would help. <!-- streview-comment:1467 --> -- 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]
