becketqin commented on issue #8535: [FLINK-11693] Add KafkaSerializationSchema that uses ProducerRecord URL: https://github.com/apache/flink/pull/8535#issuecomment-505757170 @aljoscha It seems we have to do it in this PR to avoid potential backwards incompatible interface change or functionality regression. I am worrying about the following: 1. we do one of the following: a. We introduce a `KafkaSerializationSchema.serialize(T element, Long timestamp, KafkaContext context)`; b. We simply keep the patch as is, i.e. partitioner is no longer supported. 2. Flink 1.9 releases and users start to provide implementation of `KafkaSerializationSchema`; 3. We change the API to use `ContextAware`, so `KafkaSerializationSchema.serialize()` no longer needs a KafkaContext argument. Instead the KafkaContext argument will be set using `setContext(KafkaContext)` for the `KafkaSerializationSchema`. If we did 1a, for backwards compatible purpose, in the next release, we will have to keep the old method signature and add a new one. If we did 1b, Flink 1.9 users will not be able to use partitioner, which is a functionality regression. But if step 1 does not go into release 1.9, we can do 1b with an additional patch without any problem.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
