twalthr commented on issue #6534: [FLINK-10107] [sql-client] Relocate Flink Kafka connectors for SQL JARs URL: https://github.com/apache/flink/pull/6534#issuecomment-412001055 @zentol The goal of SQL jars was to provide ready-to-use libraries that could be added on a drag and drop basis. We do not need to support multiple Kafka versions. Even though I can image ETL cases where you might want to read from a global Kafka storage (maybe using an old version) to a testing storage (maybe using a newer version). This PR tries to solve the problem of having SQL jars that interfere with each other. The root cause of this is actually our current dependency structure for Kafka connectors. We have a Kafka-base but actually depend on the previous versions. After sleeping a night over this PR, I think we cannot merge this because we expose FlinkKafkaPartitioner which is also relocated now. A perfect solution would be: - Move version-specific Kafka code into version-specific packages (e.g. o.a.f.kafka10) - Move version-agnostic Kafka code into a common package (e.g. o.a.f.kafka)
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on 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
