Starting having MainLoopTests.InvokeLeakTest failure after push develop branch to my fork .

Starting test execution, please wait...
A total of 1 test files matched the specified pattern.
  Failed Terminal.Gui.Core.MainLoopTests.InvokeLeakTest [1 s]
  Error Message:
[xUnit.net 00:01:04.81]     Terminal.Gui.Core.MainLoopTests.InvokeLeakTest [FAIL]
   System.TimeoutException : Timeout: Increment lost. tbCounter (41) didn't change after waiting 500 ms. Failed to reach 10000 on pass 1
  Stack Trace:
     at Terminal.Gui.Core.MainLoopTests.RunTest(Random r, TextField tf, Int32 numPasses, Int32 numIncrements, Int32 pollMs) in /home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/MainLoopTests.cs:line 556
   at Terminal.Gui.Core.MainLoopTests.<>c__DisplayClass24_0.<InvokeLeakTest>b__0() in /home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/MainLoopTests.cs:line 577
   at System.Threading.ExecutionContext.RunFromThreadPoolDispatchLoop(Thread threadPoolThread, ExecutionContext executionContext, ContextCallback callback, Object state)
--- End of stack trace from previous location ---
   at System.Threading.Tasks.Task.ExecuteWithThreadLocal(Task& currentTaskSlot, Thread threadPoolThread)
--- End of stack trace from previous location ---
   at Terminal.Gui.Core.MainLoopTests.InvokeLeakTest() in /home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/MainLoopTests.cs:line 582
--- End of stack trace from previous location ---

Attachments:
  /home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/TestResults/5774b41c-7d1a-461a-a399-f2edfe5328aa/coverage.opencover.xml
Failed!  - Failed:     1, Passed:   873, Skipped:     0, Total:   874, Duration: 1 m 9 s - /home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/bin/Debug/net6.0/UnitTests.dll (net6.0)
     4>Done Building Project "/home/runner/work/Terminal.Gui/Terminal.Gui/UnitTests/UnitTests.csproj" (VSTest target(s)) -- FAILED.
     1>Done Building Project "/home/runner/work/Terminal.Gui/Terminal.Gui/Terminal.sln" (VSTest target(s)) -- FAILED.

Build FAILED.
    0 Warning(s)
    0 Error(s)

Time Elapsed 00:01:19.40
Error: Process completed with exit code 1.

[montekarlos]

My suggestion would be to change the test to

volatile static int tbCounter = 0;
static ManualResetEventSlim _wakeUp = new ManualResetEventSlim (false);

private static void Launch (Random r, TextField tf, int target)
{
	Task.Run (() => {
		Thread.Sleep (r.Next (2, 4));
		Application.MainLoop.Invoke (() => {
			tf.Text = $"index{r.Next ()}";
			Interlocked.Increment (ref tbCounter);
			if (target == tbCounter) {
				// On last increment wake up the check
				_wakeUp.Set ();
			}
		});
	});
}

private static void RunTest (Random r, TextField tf, int numPasses, int numIncrements, int pollMs)
{
	for (int j = 0; j < numPasses; j++) {

		_wakeUp.Reset ();
		for (var i = 0; i < numIncrements; i++) {
			Launch (r, tf, (j + 1) * numIncrements);
		}

		
		while (tbCounter != (j + 1) * numIncrements) // Wait for tbCounter to reach expected value
		{
			var tbNow = tbCounter;
			_wakeUp.Wait (pollMs);
			if (tbCounter == tbNow) {
				// No change after wait: Idle handlers added via Application.MainLoop.Invoke have gone missing
				Application.MainLoop.Invoke (() => Application.RequestStop ());
				throw new TimeoutException (
					$"Timeout: Increment lost. tbCounter ({tbCounter}) didn't " +
					$"change after waiting {pollMs} ms. Failed to reach {(j + 1) * numIncrements} on pass {j + 1}");
			}
		};
	}
	Application.MainLoop.Invoke (() => Application.RequestStop ());
}

[Fact]
[AutoInitShutdown]
public async Task InvokeLeakTest ()
{
	Random r = new ();
	TextField tf = new ();
	Application.Top.Add (tf);

	const int numPasses = 10;
	const int numIncrements = 10000;
	const int pollMs = 20000;

	var task = Task.Run (() => RunTest (r, tf, numPasses, numIncrements, pollMs));

	// blocks here until the RequestStop is processed at the end of the test
	Application.Run ();

	await task; // Propagate exception if any occurred

	Assert.Equal ((numIncrements * numPasses), tbCounter);
}

This gives a much longer timeout for slow build systems but is still ‘fast’ locally

It’s been my experience that it’s quite difficult to unit test non deterministic threading behaviour effectively

Note I haven’t tested if this reproduces the original bug prefix

[BDisp]

I think it’s already fixed.