I have read three articles of lewissbaker, and I am totally lost. And now I have also read https://blog.eiler.eu/posts/20210512/, and I am even more lost. The author gives an example https://gist.github.com/MichaEiler/99c3ed529d4fd19c4289fd04672a1a7c, and it does compile. My question is about the example.

In MichaEiler’s gist, I find the following code very strange:

inline sync_wait_task make_sync_wait_task(task& t)
{
    co_await t;
}

I have several questions:

1. Why does the make_sync_wait_task function return sync_wait_task without a return?

   I guess it is because co_await will do some rewriting that will “return” some awaiter class. However, even if I am right here, sync_wait_task is NOT an awaiter class, because it has no await_ methods.

2. Furthermore, why is the return type of this function sync_wait_task? I mean not task or the awaiter in threadpool?

   In fact, sync_wait_task has nothing to do with task. The task class is defined without knowing about the existence of sync_wait_task. Then, why can co_awaiting a task “return” a sync_wait_task instance? What is the relationship here?

   I guess this is because sync_wait_task is the continuation of task, so some symmetry transfer thing happens here. Execution flows from task to sync_wait_task. If I am right, could you help me understand what exactly happens here, or relate it to the relevant sections in lewissbaker’s articles?