Class RowTimeRowsUnboundedPrecedingFunction<K>

java.lang.Object
org.apache.flink.api.common.functions.AbstractRichFunction
org.apache.flink.streaming.api.functions.KeyedProcessFunction<K,IN,OUT>
org.apache.flink.table.runtime.functions.KeyedProcessFunctionWithCleanupState<K,org.apache.flink.table.data.RowData,org.apache.flink.table.data.RowData>
org.apache.flink.table.runtime.operators.over.AbstractRowTimeUnboundedPrecedingOver<K>
org.apache.flink.table.runtime.operators.over.RowTimeRowsUnboundedPrecedingFunction<K>
All Implemented Interfaces:
Serializable, org.apache.flink.api.common.functions.Function, org.apache.flink.api.common.functions.RichFunction, CleanupState

public class RowTimeRowsUnboundedPrecedingFunction<K> extends AbstractRowTimeUnboundedPrecedingOver<K>
A ProcessFunction to support unbounded ROWS window. The ROWS clause defines on a physical level how many rows are included in a window frame.

E.g.: SELECT rowtime, b, c, min(c) OVER (PARTITION BY b ORDER BY rowtime ROWS BETWEEN UNBOUNDED preceding AND CURRENT ROW), max(c) OVER (PARTITION BY b ORDER BY rowtime ROWS BETWEEN UNBOUNDED preceding AND CURRENT ROW) FROM T.

See Also:
  • Constructor Details

    • RowTimeRowsUnboundedPrecedingFunction

      public RowTimeRowsUnboundedPrecedingFunction(long minRetentionTime, long maxRetentionTime, GeneratedAggsHandleFunction genAggsHandler, org.apache.flink.table.types.logical.LogicalType[] accTypes, org.apache.flink.table.types.logical.LogicalType[] inputFieldTypes, int rowTimeIdx)
  • Method Details