SEZ9 commented on issue #12571: URL: https://github.com/apache/seatunnel/issues/12571#issuecomment-5944267097
Thanks @SEPURI-SAI-KRISHNA — the matrix measured through `ZetaSQLEngine` on `dev` at `53733a112c9e935c5d97f7f24be249caa43b4fd5` is enough to proceed, and it matches what I see at `d7e9931bea0976547e9730675e7019bbef88ad83`. The scope stands as agreed: integral targets only, planner acceptance unchanged, `CAST` fails and `TRY_CAST` returns null on out-of-range numeric input at the shared cast boundary, floating point left for a separate decision. Two small notes on the tables: - Count: six cells change, not four. `-2147483649`, `2147483648` and `3000000000` each flip for both `CAST` and `TRY_CAST` (today `2147483647`, `-2147483648` and `-1294967296`). After the change all six become `CAST` error / `TRY_CAST` null, matching the STRING column. The exact-boundary rows and `0` stay as they are. - Since the range check is going in at the shared boundary for `TINYINT` and `SMALLINT` as well, please test that code directly with out-of-range numeric inputs for those targets, not only via the STRING matrix. The planner does not route numeric narrowing to them today, so without a direct test the new check for those two targets would have no coverage. Everything else from the earlier instructions stands; link the PR here once it is open. <!-- streview-comment:1454 --> -- 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]
