THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
...
JIRA:
Jira | ||||||
---|---|---|---|---|---|---|
|
Jira | ||||||||
---|---|---|---|---|---|---|---|---|
|
Motivation
For users who cannot read their source topic as a changelog stream from the beginning we need to provide a way for event streams to be translated into changelog stream. As pointed out by guozhang Wang this should not be confused with KStream.reduce(), these functions should:
- completely change the semantics of the input stream from event stream to changelog stream any null-values will still be serialized, and if the resulted bytes are also null they will be interpreted as "deletes" to the materialized KTable (i.e. tombstones in the changelog stream).
- Materialization of the KTable will follow the usual process:
- if Optimization is turned off, the KTable will always be materialized (but the store will not be queryable)
- if Optimization is turned on and if .toTable() is used, the Ktable may or may not be materialized. The store still cannot be queried.
- if .toTable(final Materialized<K, VRV, KeyValueStore<Bytes, byte[]>> materialized) is used irrespective of the optimization strategy used the KTable will be materialized and will be queryable.
...