mxtymoshyk commented on issue #26911: URL: https://github.com/apache/beam/issues/26911#issuecomment-6439325831
I checked this on current master. At first it looked fixed, since the query from the issue passes when run through `SqlTransform`. I wanted a regression test before closing it, so I also ran it through the JDBC driver, which is what the SQL shell uses. There it still fails with the same `Row expected 4 fields ... initialized with 5 fields` error, so a fix is needed after all. The two paths plan the query with different Calcite planners. The one behind `SqlTransform` flattens the nested `author` row into separate columns. The one behind the JDBC driver keeps `author` as a single ROW column. `UnnestFn` always flattened, so it only matched `SqlTransform`. I'm opening a PR that makes `UnnestFn` follow the output schema, with tests for both paths. -- 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]
