Page History

...

Current Issues to Work out

1) Work with DefaultPartitioner class and with custom partitioners. Currently hardwired using a copy of the default Partitioner code for key != null logic.

2) Testing Framework. I have done all of my testing outside of the kafka streams unit testing framework, using my own framework that brings up a kafka cluster with user-specified partition counts. I was able to test the copartitioning and

3) Ensuring covereage for restoring from a failed state. I am unsure how logging state stores back data up to the cluster with regards to internal topic propagation, which has influenced the decision to include a "propagate" boolean in the PropagationWrapper, versus letting the high-water mark resolve it.