slinkydeveloper commented on pull request #17544: URL: https://github.com/apache/flink/pull/17544#issuecomment-953184692
The last commit I made https://github.com/apache/flink/pull/17544/commits/a20fe2427c03ceb71b8d4b5f30bdc282aba530fe fixes the behaviour of `TestCsvDeserializationSchema`. In particular, before this PR, `TestCsvDeserializationSchema` was implicitly assuming that the input data type was a 1:1 mapping to the parsed csv row. This is not the case, and the DeserializationSchema should take care of performing the proper mapping between the input records (in whatever format) and the output `RowData`, which may or may not contain all the fields of the input record and may or may not contain those ordered in the same way the input is. This PR addresses the issue. Note that the problem exists for this particular csv format, because `flink-csv` is still using the old interface (hence, no change) and because json deserialization schema properly map the input object fields to the output row data, without performing any bad assumption: https://github.com/apache/flink/blob/master/flink-formats/fli nk-json/src/main/java/org/apache/flink/formats/json/JsonToRowDataConverters.java#L334 -- 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]
