RFR: ClassicProblem_02_ProducerConsumerProblem - third try

[Michael Mirwaldt]

On Thu, 18 Nov 2021 16:28:15 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> This sample implements some solutions from the wikipedia article about the producer-consumer-problem on https://en.wikipedia.org/wiki/Producer%E2%80%93consumer_problem .
>> It contains these solutions:
>> - _OneProducerOneConsumer_ with semaphores
>> - _FlawedTwoProducersOneConsumer_ with semaphores in which the producers sometimes use the same index for their items and overwrite each others item in the buffer
>> - _FixedTwoProducersOneConsumer_ with semaphores which uses an extra lock for the index so that the producers never use the same index for their items
>> - _Lock_ doesn't use any sempahores but a reentrant lock with two conditions. It supports many producers and many consumers
>> - _AtomicIntegers_ neither uses any semaphores but two atomic integers and supports one producer and one consumer.
>> 
>> I have decided not to implement the last solution with finite-range single writer registers in the wikipedia article. It looked far too complicated to me.
>> 
>> This is my third try to avoid squash commits.
>
> jcstress-samples/src/main/java/org/openjdk/jcstress/samples/concurrency/classic/ClassicProblem_02_ProducerConsumerProblem.java line 49:
> 
>> 47:  */
>> 48: public abstract class ClassicProblem_02_ProducerConsumerProblem {
>> 49:     private final static int BUFFER_SIZE = 2;
> 
> This should be inside `SemaphoresBase`?

No, I don't think so because the two examples _Lock_ and _AtomicIntegers_ both use _BUFFER_SIZE_ but they are not sub classes of _SemaphoresBase_. They don't use any semaphores.

-------------

PR: https://git.openjdk.java.net/jcstress/pull/104