Ilya Egorov

I used inheritance for the reason that it looks the most sensible here: existing Condition classes are _extended_ with additional notification methods. The helper ones are prefixed with `__`, which...

Since 1d91757, the fix is really a side effect: there is no way to set an event loop in the `wait()` method other than the one bound in `async_lock`, since...

Unless we can reverse the order of changes and apply the PR with the fix first. It can be seen as removing unnecessary `parent._get_loop()` in nowait methods (which became unnecessary...

I agree @A5rocks, it does look weird, but the point is there: if a Trio run loop is running in a thread, then by default every `await` in that thread...

Now, since the [last commit](https://github.com/x42005e1f/aiologic/commit/e3920dd68a9c55600972b7ac26f214dd86efe111), the above behavior is clarified in my README.

The hub is created automatically for the thread when using `eventlet`. The problem is that there is no way for the user to reverse this side effect, unless they access...

I also looked at the `_AsyncioMutex` implementation and I cannot agree with the implemented approach. Polling will definitely load the CPU. This can be avoided by using something similar to...

I will also clarify again, as I feel a certain bias. My suggestions are not to improve the usability of `eventlet` in new code, but to make it painless to...

I think it is redundant to place warnings about the `eventlet` deprecation everywhere it is supported. My README just lists the supported libraries and does not focus on `eventlet` in...

A little bit about the lock implementation approach. In your comment to the `acquire()` method you said the following: "The chance of hitting this path is usually quite low (it...