I’ve logged a JIRA case[1] and sketched out some code[2] for a streaming TCK. The idea is that we define the TCK once in Calcite, then each engine (Samza, Flink, Storm, etc., plus Calcite’s reference implementation) implements the sole abstract method in the TCK so that all of the tests in the TCK get run against that particular engine.
If you’re interested in helping design streaming SQL, join the conversation at that JIRA case. Julian [1] https://issues.apache.org/jira/browse/CALCITE-1114 <https://issues.apache.org/jira/browse/CALCITE-1114> [2] https://github.com/julianhyde/calcite/tree/1114-stream-tck <https://github.com/julianhyde/calcite/tree/1114-stream-tck>
