FYI: This discussion is initiated from this PR: https://github.com/apache/spark/pull/22282#discussion_r214632570 (Supporting Kafka Headers functionality in Spark Structured Streaming)
Thanks, Dongjin On Wed, Sep 5, 2018 at 7:41 AM Jungtaek Lim <kabh...@gmail.com> wrote: > Hi Kafka devs, > > I just noticed that the only implementation for > org.apache.kafka.common.header.Header is RecordHeader which is exposed to > the 'internal' package, and some of end users just picked the class because > they don't want to create their own. IMHO it can give false signal for end > users on taking risk to access internal thing. > > Would we be better to provide it without having to touch internal package > from end users? While I think moving RecordHeader would be tough one > (allowing duplicates works though) we could do it with various approach > like builder my folk suggested if we really want to hide the implementation > and let end users only refer Header interface. > > If that's only missing spot I'm happy to take it forward. (I might request > contributor role then.) If that's by intention please also let me know that > we can discuss about it. > > Thanks! > Jungtaek Lim (HeartSaVioR) > -- *Dongjin Lee* *A hitchhiker in the mathematical world.* *github: <http://goog_969573159/>github.com/dongjinleekr <http://github.com/dongjinleekr>linkedin: kr.linkedin.com/in/dongjinleekr <http://kr.linkedin.com/in/dongjinleekr>slideshare: www.slideshare.net/dongjinleekr <http://www.slideshare.net/dongjinleekr>*
