Page History

...

In KStreamImpl, implementation is fairly simple:

1. We call op, passing our current KStream as its argument. This produces our output KStream.
2. We determine if repartitioning is required on the op stream, and if it is, we automatically include a repartition node, equivalent to adding .repartition()  to the end of the op stream.We call op, passing our current KStream as its argument. This produces our output KStream.
3. We wire up the graphNode  from the output KStream  as a parent of the current KStream. This takes care of the recursion.
4. Finally, we return the output  KStream. This enables users to operate on the records that are being recursively produced, as above.

...