...

For Flink DataStream jobs:

• When creating a source, user provides a `WatermarkStrategy` to `StreamExecutionEnvironment#fromSource`.
• If the the source supports event time natively (e.g., KafkaSource) or the user provides a custom `TimestampAssigner` in the `WatermarkStrategy` to extract the timestamp from the record, Flink will add the timestamp to the record. Otherwise, the timestamp on the record will be `Long.MIN_VALUE`.
• If the user uses `NoWatermarkGenerator` in the `WatermarkStrategy`, the job will not generate watermarks. Otherwise, the job will periodically emit watermarks, and the watermark value depends on event time. The output frequency of watermarks is determined by `pipeline.auto-watermark-interval`, with a default value of 200ms.

Public Interfaces

1) Modify `org.apache.flink.api.common.eventtime.Watermark` and `org.apache.flink.streaming.api.watermark.Watermark `by adding a new field `useProcessingTime`, with a default value of false.

...