RFR: 8314515: java/util/concurrent/SynchronousQueue/Fairness.java failed with "Error: fair=false i=8 j=0"

[Martin Buchholz]

On Fri, 18 Aug 2023 08:15:58 GMT, Viktor Klang <duke at openjdk.org> wrote:

>> The usual tiny improvements, with no guarantee that the intermittent test failure is actually fixed.
>
> test/jdk/java/util/concurrent/SynchronousQueue/Fairness.java line 41:
> 
>> 39:         throws Throwable
>> 40:     {
>> 41:         int threadCount = ThreadLocalRandom.current().nextInt(2, 8);
> 
> @Martin-Buchholz I'm a bit weary about making the number of threads random—might make it tricky to get to reproduce if the test fails?

Suppose the test fails only with a specific number of threads.  I would want to know that!  At least once in the past a j.u.c. bug has been found due to randomization of concurrency level in a test.  I can't understand the current industry obsession with deterministic tests, except that too many  organizations don't commit to hunting down flakes.

Here the error detail has been changed to report the threadcount, in case of failure.

> test/jdk/java/util/concurrent/SynchronousQueue/Fairness.java line 55:
> 
>> 53:             Thread t = new Thread(put);
>> 54:             t.start();
>> 55:             ts.add(t);
> 
> @Martin-Buchholz Any specific reason to add the thread *after* it has been started?

I've never considered that the order would matter - so, no reason.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/15337#discussion_r1299024335
PR Review Comment: https://git.openjdk.org/jdk/pull/15337#discussion_r1299027961