KIP-259 — Improve Streams DSL Timestamp Propagation Semantics

Discussion Kafka 2.3 Streams

Formalizes timestamp propagation rules for Streams DSL operators: aggregations use the maximum timestamp of records in the window, joins use the maximum of the two input record timestamps, and stateless operators pass through the input timestamp unchanged. Without explicit rules, timestamp propagation through aggregations and joins is non-deterministic, making windowed computation semantics unpredictable.

Details