Difference in behavior when mocking async method using .Result vs without

[cyungmann]

Using Moq 4.17.2, the first two mock setups in the code below work as expected and return a Task<string> that resolves to null, while the third mock setup simply returns null. Is this expected?

using Microsoft.VisualStudio.TestTools.UnitTesting;
using Moq;
using System.Threading.Tasks;

namespace MoqTest2
{
    public interface IFoo
    {
        Task<string?> Bar();
    }

    [TestClass]
    public class UnitTest1
    {
        [TestMethod]
        public async Task TestMethod1()
        {
            var mock = new Mock<IFoo>();

            mock.Setup(x => x.Bar()).Returns(() => Task.FromResult((string?)null));
            var result = await mock.Object.Bar().ConfigureAwait(false);

            mock.Setup(x => x.Bar()).ReturnsAsync((string?)null);
            result = await mock.Object.Bar().ConfigureAwait(false);

            mock.Setup(x => x.Bar().Result).Returns((string?)null);
            result = await mock.Object.Bar().ConfigureAwait(false);
        }
    }
}

[Ana-byte-bit]

It seems that problem is that Result allowed to have attribute of type TResult instead of Task<TResult>.

[stakx]

This is a bug. As it turns out, the logic that's responsible for wrapping the return value in a completed task only triggers for non-null return values, which is incorrect since null is a perfectly valid value (for reference types and nullable value types).