[ https://issues.apache.org/jira/browse/FLINK-17879?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17113854#comment-17113854 ]
Timo Walther commented on FLINK-17879: -------------------------------------- Yes, for `collect()` the `Row` type might be more suitable. And `RowData` for UDF process exchange? People more familiar with the Python API should decide this. I will assign this issue to you. Thanks. > Update Python row serializer > ---------------------------- > > Key: FLINK-17879 > URL: https://issues.apache.org/jira/browse/FLINK-17879 > Project: Flink > Issue Type: Improvement > Components: API / Python > Reporter: Timo Walther > Priority: Major > > Flink's `RowSerializer` has been updated and encodes now a `RowKind` in its > format (see FLINK-16998). However, Python code still depends on the old > format which is why these code paths are currently using > `legacyModeEnabled=true`. The Python side should be updated. Either to use > the new format, or `RowData` instead? -- This message was sent by Atlassian Jira (v8.3.4#803005)