Github user medcv commented on the issue: https://github.com/apache/flink/pull/6259 @dawidwys Thanks for your suggestions. New commit extend `SchemaCoder` with `writeSchema` method that helps to move the writing schema logic away from `AvroSerializationSchema` as you suggested. Totally agree with you that having dynamic `subject` variables make the implementations more generic but as this is `Confluent` specific implementations and this variable is only presented for `ConfluentRegistryAvroSerializationSchema`, I think a user of this method should be aware of how `Confluent` requires this variable when they setup their Kafka Producer and Schema Registry. I am open to suggestions to fix the issue ( by changing `FlinkKafkaProducer`) if you still thinking this is a blocker for this PR.
---