Hi, Grant, The append api seems indeed a bit weird. The compression type is a producer level config. Instead of passing it in for each append, we probably should just pass it in once during the creation RecordAccumulator. Could you file a jira to track this?
Thanks, Jun On Mon, Mar 23, 2015 at 7:16 PM, Grant Henke <ghe...@cloudera.com> wrote: > I am reading over the new producer code in an effort to understand the > implementation more thoroughly and had some questions/feedback. > > Currently org.apache.kafka.clients.producer.internals.RecordAccumulator > append method accepts the compressionType on a per record basis. It looks > like the code would only work on a per batch basis because the > CompressionType is only used when creating a new RecordBatch. My > understanding is this should only support setting per batch at most. I may > have misread this though. Is there a time where setting per record would > make sense? > > public RecordAppendResult append(TopicPartition tp, byte[] key, byte[] > value, CompressionType compression, Callback callback) throws > InterruptedException; > > Why does org.apache.kafka.common.serialization.Serializer Interface require > a topic? Is there a use case where serialization would change based on > topic? > > public byte[] serialize(String topic, T data); > > Thank you, > Grant > > -- > Grant Henke > Solutions Consultant | Cloudera > ghe...@cloudera.com | 920-980-8979 > twitter.com/ghenke <http://twitter.com/gchenke> | > linkedin.com/in/granthenke >
