SEZ9 commented on issue #12752: URL: https://github.com/apache/seatunnel/issues/12752#issuecomment-6307133931
@zhengxiang378928908-code thanks for following up with the linked PR and the `LENGTH$` analysis. Reading `SYS.SYSCOLUMNS.LENGTH$` directly for `NVARCHAR` / `NVARCHAR2`, instead of deriving length from `ALL_TAB_COLUMNS.DATA_LENGTH`, is the right direction for the doubled-length symptom reported here, and the measurements across the `LENGTH_IN_CHAR` × charset combinations are exactly the evidence we asked for. A few things I'd like to see confirmed on the PR side before this issue can be closed: - The catalog tests should include the `NVARCHAR2` `CHAR_LENGTH = 0` case explicitly, since that was the reason an unconditional `DATA_LENGTH / 2` fallback was ruled out. - Please make sure the read-only user E2E demonstrates that the `LENGTH$` lookup works without extra privileges, as that is the scenario most likely to regress for users whose catalog account is not the schema owner. - Keep CHAR/VARCHAR/NCHAR handling unchanged and covered by the existing tests, as you noted. @hutiefang76 thanks for offering and for stepping back to avoid a parallel patch; reviews on the linked PR are of course welcome. Let's keep implementation discussion on the PR. This issue stays open until that PR is reviewed and merged. <!-- streview-comment:1917 --> -- 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]
