Java Reference
In-Depth Information
Appendix C. Performing multiple operations in
parallel on a stream
One of the biggest limitations of a Java 8 stream is that you can operate on it only once and get
only one result while processing it. Indeed, if you try to traverse a stream for a second time, the
only thing you can achieve is an exception like this:
java.lang.IllegalStateException: stream has already been operated upon or closed
Despite this, there are situations where you'd like to get several results when processing a single
stream. For instance, you may want to parse a log file in a stream, as we did in
section 5.7.3
, but
gather multiple statistics in a single step. Or, keeping with the menu data model used to explain
Stream's features in
chapters 4
-
6
, you may want to retrieve different information while
traversing the stream of dishes.
In other words, you'd like to push a stream through more than one lambda on a single pass, and
to do this you need a type of fork method and to apply different functions to each forked stream.
Even better, it would be great if you could perform those operations in parallel, using different
threads to calculate the different required results.
Unfortunately, these features aren't currently available on the stream implementation provided
in Java 8, but in this appendix we'll show you a way to use a Spliterator and in particular its
late-binding capacity, together with BlockingQueues and Futures, to implement this useful
feature and make it available with a convenient API.
[
1
]
1
The implementation presented in the rest of this appendix is based on the solution posted by
Paul Sandoz in the email he sent to the lambda-dev mailing list:
C.1. Forking a stream
The first thing necessary to execute multiple operations in parallel on a stream is to create a
StreamForker that wraps the original stream, on which you can define the different operations
you want to perform. Take a look at the following listing.
Search WWH ::
Custom Search