fatmanverse commented on PR #12802: URL: https://github.com/apache/seatunnel/pull/12802#issuecomment-6261819821
@DanielLeens a heads-up, since this changed after your approval: 1. **Production code change in ffc8b1f.** While adding a sink-to-reader round trip test for both engines, I found that the EasyExcel engine still cut DOUBLE values (`10.285714285` -> `10.28571428`): EasyExcel passes the text of a number cell formatted with its display format, the same issue the POI path had. Numeric columns now read the stored number of the cell (`ReadCellData#getNumberValue`) with EasyExcel as well. The numeric types moved to `ExcelCellUtils#NUMERIC_SQL_TYPES`, shared by both engines. `incompatible-changes.md` (en/zh) now says both engines are affected. 2. **Merge of `dev` in 3284e8f** (conflict resolution only, nothing else). #12768 rewrote the same POI read loop for header-name mapping. I kept its loop and only pass the extra `fieldType` argument, and kept both the XML (#12801) and the Excel entries in `incompatible-changes.md`. 3. **Tests:** `ExcelReadStrategyTest#testSinkOutputRoundTripsDoubleAndDecimal` (POI and EasyExcel) and `#testHeaderMappedReadKeepsNumericPrecision` (`excel_use_header_line` with both engines). Verified on JDK 17: the `connector-file-base` unit tests all pass (`ExcelReadStrategyTest` 20 tests including the #12768 cases). On the deployed Zeta cluster, the 3000-row MySQL -> Excel -> MySQL round trip now has 0 DOUBLE and 0 DECIMAL mismatches with both POI and EasyExcel. Could you take another look when you have time? Thanks! -- 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]
