Seeing inconsistent test results in TEST(RedisSubscriber, SubscribeMultiplePublished)

[trondrift]

I've been developing a proof of concept windows app using cpp_redis for pubsub and noticed my subscriber doesn't always catch consecutive publishes to its channel. I can re-create this behavior using the provided unit tests. If I vary the number of publish calls in the SubscribeMultiplePublished test, it seems to fail sometimes, but not others.

This is the test code I've been using:

 TEST(RedisSubscriber, SubscribeMultiplePublished) {
  cpp_redis::redis_subscriber sub;
  cpp_redis::redis_client client;
  std::condition_variable cv;

  sub.connect();
  client.connect();

  std::atomic<int> number_times_called = ATOMIC_VAR_INIT(0);
  sub.subscribe("/chan",
    [&](const std::string& channel, const std::string& message) {
      ++number_times_called;
      EXPECT_TRUE(channel == "/chan");
      if (number_times_called == 1)
        EXPECT_TRUE(message == "first");
      else if (number_times_called == 2)
        EXPECT_TRUE(message == "second");
      else if (number_times_called == 3)
	EXPECT_TRUE(message == "third");
      else if (number_times_called == 4)
	EXPECT_TRUE(message == "fourth");
      else if (number_times_called == 5)
	EXPECT_TRUE(message == "fifth");

      cv.notify_all();
    },
    [&](int64_t) {
      client.publish("/chan", "first");
      client.publish("/chan", "second");
      client.publish("/chan", "third");
      client.publish("/chan", "fourth");
      client.publish("/chan", "fifth");
      client.commit();
    });

  sub.commit();

  std::mutex mutex;
  std::unique_lock<std::mutex> lock(mutex);
  cv.wait_for(lock, std::chrono::seconds(60), [&]() -> bool { return number_times_called == 5; });

  EXPECT_TRUE(number_times_called == 5);
}

TEST RESULTS

My setup uses 64-bit Windows 7, Visual Studio 2015 (The project however is built 32-bit). I see this both building cpp_redis statically as well as dynamically.

[Cylix]

Hi,

I was not able to reproduce this bug on my macbook.
However, there is a high change that this issue is related to tacopie#17.

Can you just confirm me which version of cpp_redis are you using?
My guess is that you are using the latest version and that this bug is not happening in earlier versions?

If it is the case, then this issue is indeed related to tacopie#17 and a fix will be submitted by tomorrow night.

Thanks for reporting this issue!

Best

[Cylix]

I published a fix in v3.5.3.
I closed this issue but feel free to re-open it if the bug is still there or if you have any other issues to report!

Best

[trondrift]

Simon, I haven't seen the issue occur since the tacopie fix. Thanks a lot!

…

On Sun, Jul 2, 2017 at 3:35 PM, Simon Ninon ***@***.***> wrote: I published a fix in v3.5.3. I closed this issue but feel free to re-open it if the bug is still there or if you have any other issues to report! Best — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#85 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AcbAM3oTvBx-CmPxSrr6AmnWM0mTELDoks5sJ_EPgaJpZM4OK2yY> .

[Cylix]

Awesome, thank you for your feedback!