Streaming patterns — dedup, top-N, split/merge, MATCH_RECOGNIZE, OVER

5 h
requires FLK-04

From your world. the recurring SQL shapes you already write (QUALIFY row_number dedup, window functions), now as streaming queries whose output is a changelog

Practice quiz with instant feedback. For a scored run that counts toward your local progress, use uv run de quiz FLK-09.

1A deduplication query uses ROW_NUMBER() OVER (PARTITION BY k ORDER BY ts) WHERE rn = 1 with an ascending ORDER BY. What kind of changelog does it produce? AUpdating. It retracts the previous first row when a duplicate arrives BAppend-only. After the query emits the first row per key, it drops later duplicates and never retracts the first row CIt produces only deletes DIt depends on the sink

6How does an OVER aggregation differ from a GROUP BY aggregation in Flink SQL? AOVER collapses rows into one row per group. GROUP BY keeps every row BOVER produces exactly one output row per input row, which carries the running aggregate. It does not collapse rows. GROUP BY collapses rows per key CThey are the same DOVER cannot be used in streaming mode