show actual values in error message

[asolntsev]

... instead of just throwing an empty runtime exception.

Problem

Before this change, test might fail with an unclear message:

java.lang.RuntimeException
	at net.datafaker.Issue759.issue759Test(Issue759.java:93)

or the output of multiple failing tests might be mixed:

 0 0 0 0 0
 0 0 0 0 0
 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0
 0 0 0
 0 0 0
 0 0 0
 0

 0 0 0 0 0

java.lang.RuntimeException
	at net.datafaker.Issue759.issue759Test(Issue759.java:93)

Solution

Now the error message is clear:

java.lang.AssertionError: Not all of s[0, 0, 0, 0, 0] are not equal to 20000

	at net.datafaker.Issue759Test.issue759Test(Issue759Test.java:85)

P.S.

To reproduce the issue, just put breakpoint, say, to any field of WorkerThread, run Issue759 test in debug mode and resume the execution after stopping on the breakpoint.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.10%. Comparing base (b37c566) to head (e973bce).
Report is 97 commits behind head on main.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #1196      +/-   ##
============================================
- Coverage     92.35%   92.10%   -0.25%     
- Complexity     2821     2891      +70     
============================================
  Files           292      300       +8     
  Lines          5609     5726     +117     
  Branches        599      618      +19     
============================================
+ Hits           5180     5274      +94     
- Misses          275      293      +18     
- Partials        154      159       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[snuyanzin]

Thanks for your contribution @asolntsev

it seems there was a long standing issue with this test, which apparently started to reproduce with your PR [1] on Windows only...
I submitted a fix for that (a bit more description is below)
If you are ok with it, we can merge it, otherwise please express your concerns

About the issue:
It looks the problem was with the code

        int[] lastIters = getIterations(threads);
        while (true) {
            Thread.sleep(100);
            int[] iters = getIterations(threads);
            if (Arrays.equals(lastIters, iters)) {
                // Either all threads are done, or something is probably stuck
                System.err.println();
                if (allElementsEqual(iters, iterationsPerThread)) {
                    break;
                } else {
                    printIterations(iters);
                    throw new RuntimeException();
                }
            }

On Windows it could happen that we reach the line if (allElementsEqual(iters, iterationsPerThread)) { before the threads are even started... as a result there was none iterations executed...
The idea of fix is to measure amount of time of being unchanged and react on the fact that this amount of time is exceeded some limit.

[1] https://github.com/datafaker-net/datafaker/actions/runs/9138381122/job/25130594393#step:4:792

[asolntsev]

@snuyanzin Thank you for sharing the information.
Now I looked at the test once more and realized it can be simplified using Java built-in class CountDownLatch.
Look my last commit. :)

[snuyanzin]

nit: i guess we could reuse the second var as well

Suggested change

	CountDownLatch countDownLatch = new CountDownLatch(numThreads * 20000);
	CountDownLatch countDownLatch = new CountDownLatch(numThreads * iterationsPerThread);

[asolntsev]

Oops, thank you. Fixed. @snuyanzin

[snuyanzin]

@asolntsev Indeed it is better approach
Thanks a lot!
It look great

I left one tiny comment