Preferred Way to Create Failing Unit Test

[ghost]

Hi there, I've found that JINT will sometimes fail with a JavaScriptException

"<parameter name> is not defined"

I have created a unit test that replicates this failure. The core of the code looks like this:

delegate int RandoDelegate(string text, SomeClass obj);

class CallJavascriptCallback
{
    static RandoDelegate _func;

    static SomeClass _obj;

    public static void Register(RandoDelegate func)
    {
        _func = func;
    }

    public static void Run()
    {
        if (_func != null)
        {
            _obj = new SomeClass();
            var result = _func("asdfklhj", _obj);
        }
    }
}

_engine.SetValue("numInStringOut2",
    new Func<int, string>(number =>
    {
        cSharpMethodCalled = true;
        return "C# can see that you passed: " + number;
    })
);
_engine.SetValue("CallJavascriptCallback", typeof(CallJavascriptCallback));
_engine.SetValue("setTimeout", new Action<Action, int>((action, msec) =>
{
    System.Threading.Tasks.Task.Run(() =>
    {
        System.Threading.Tasks.Task.Delay(msec);
        action();
    });
}));

_engine.Execute(@"
	{
		let counter = 0;
		var callback = function(){
			++counter;
			var result = numInStringOut2(42);
		}

		CallJavascriptCallback.Register(function(a, b){
			return 52;
		});

		setTimeout(callback, 5);
	}
");

CallJavascriptCallback.Run();

The above generates the error "a is not defined".

I'm having trouble tracking down the offending code (culprit) for the failure.

I was thinking of just creating a branch in my fork and creating a PR with that. The PR will obviously fail the CI/CD (built in test pipeline) - but I'd like to see if anyone could give me some pointers on where I could start digging to try to track down the problem.

I created a branch here:
https://github.com/source-transformer/jint/tree/issues/tpc/parameterIsNotDefined

If you want to test against mainline - you can just copy the only new file I've added which is:
Jint.Tests/Runtime/AsyncAndDisposeTests.cs

into whatever branch you're using locally.

[ghost]

It looks like this might be a symptom of using the JINT engine across multiple threads.

When I changed the above setTimeout implementation to be on the main thread - the issue doesn't occur.

Just wanted to call that out since I think we can restrict our usage of each Engine to a single thread.