Parallel processing

We have seen that changing from a sequential stream to a parallel stream can lead to incorrect results if the code was not written and tested for processing a parallel stream. The following are a few more considerations related to the parallel stream.

Stateless and stateful operations

There are stateless operations, such as filter(), map(), and flatMap(), which do not keep data around (do not maintain state) while moving from processing from one stream element to the next. And there are stateful operations, such as distinct(), limit(), sorted(), reduce(), and collect(), which may pass the state from previously processed elements to the processing of the next element.

Stateless operations usually do not pose a problem when switching from a sequential stream to a parallel one. Each element is processed independently and the stream can be broken into any number of sub-streams for independent processing.

With stateful operations, the situation is different. To start with,using...