Using Barrier to Implement Go's Wait Group

Question

I have implemented a WaitGroup class to simulate WaitGroup in Go lang. Then I've found that I can use Barrier in a different manner, to achieve the same thing. My WaitGroup has some additional functionality, but in most scenarios Barrier just would do.

Question: Do you spot any design flaws in this style of using Barrier?

Code:

I have an instance of Barrier:

static Barrier WaitGroupBarrier = new Barrier(0);

Then I define my tasks (or threads) this way in multiple places - so I have not the actual number of tasks/threads:

for (int i = 0; i < 1000; i++) // in multiple places
{
    var t = new Task(() =>
    {
        try
        {
            WaitGroupBarrier.AddParticipant();

            // body
        }
        finally
        {
            WaitGroupBarrier.RemoveParticipant();
        }
    });

    t.Start();
}

And I wait for them to complete, this way:

if (WaitGroupBarrier.ParticipantsRemaining > 0)
{
    Console.WriteLine("waiting...");
    WaitGroupBarrier.SignalAndWait();
}

Answer 1

if (WaitGroupBarrier.ParticipantsRemaining > 0) {
    WaitGroupBarrier.SignalAndWait();
}

The above is a non-atomic operation. Thus if the if branch gets evaluated and there is one remaining, but finishes before SignalAndWait, then this will throw an exception.