Hello, So I have come up with a possible solution to the described ticket. My idea here was to avoid any breaking changes and introduce a new field which will determine the behavior of the processor, called 'SCHEMA_WRITING_STRATEGY'. The default option, 'USE_FIRST_SCHEMA' will stick to the old behaviour and abide by the first transformed records schema, whereas the option 'PARTITION_BY_SCHEMA' will split the output into multiple flowfiles, depending on the number of unique schemas generated.
The implementation is by no means done in my opinion and would possibly like to include a few more tests, however I wanted to get your feedback on the possible implementation if possible. Thank you! https://github.com/apache/nifi/pull/10545 Kind regards, Jordan On Mon, 17 Nov 2025 at 13:21, Jordan Sammut <[email protected]> wrote: > Hello, > > Just bumping this thread, would really want to get some feedback on it. > > Kind regards, > Jordan > > On Thu, 13 Nov 2025 at 09:37, Jordan Sammut <[email protected]> wrote: > >> Hi all, >> >> Yesterday I raised an issue regarding the JoltTransformRecord only >> inheriting the schema from the first transformed record. Was wondering if I >> can get some feedback on this as it is a very problematic issue for us. I >> would also be more than happy to contribute in any way if possible, but >> would like to get some tips on how this can be tackled. >> >> This is the ticket: https://issues.apache.org/jira/browse/NIFI-15209 >> >> Thank you! >> >> Kind regards, >> Jordan >> >
