Designing your own asynchronous functions, which pattern is correct?

Question

C++ / C#

public Task<int> Foo()
    {
        var tcs = new TaskCompletionSource<int>();
        Task.Run(() => {
            var result = SomeAction();
            tcs.SetResult(result);
        });
        return tcs.Task;
    }

I just can’t understand what benefit the TaskCompletionSource class plays here, why is it considered bad form to write return Task.Run(...)?

Answer 1

I always use this:
//for a method that returns nothing

public async Task Blabla()
{
    await Task.Run(()=>{

     //тут я что-то выполняю

     });
}

//for a method that returns a string type (for example)

public async Task<string> Blabla()
{
    return await Task.Run(()=>{

     return "я возвращаюсь!!!";

     });
}

and that's it.
simply. works. readable.
upd.
Petr - Mercy, reminded)
upd2
Slavka : here is the answer to where TaskCompletionSource is used in practice

Answer 2

In your example, the TaskCompletionSource is definitely redundant.
Enough:

public Task<int> Foo()
{
        return Task.Run(() => {
            return SomeAction();
        });
    }