Switch to event based listening

[LivInTheLookingGlass]

The current daemon is brittle and inefficient. What needs to happen is a simple event loop for me to tie into. Unfortunately this is where my Python skills are less than stellar, so I'll need to learn what frameworks are available.

[LivInTheLookingGlass]

As a sidenote, this would allow for supporting WebSocket and potentially UDP

[Kentoseth]

Will this library be useful: AsyncIO

It is part of the standard-library and will thus have no external dependencies.

Alternatively, there is also this: aioHTTP

Supports both Server WebSockets and Client WebSockets out-of-the-box.

aioHTTP is built on top of asyncIO, so the 2 could work together.

And it looks like AsyncIO itself also has UDP support: link

[LivInTheLookingGlass]

I'm definitely aware of AsyncIO. The problem is that I don't know how to use it. I'll definitely get there eventually, but classes have kept me from working on this too much.

[Kentoseth]

Can you break down exactly what is the first minimum-viable goal for a rewrite using asyncIO/aioHTTP?

If the goals/targets are clear, I can attempt a small re-write on an experimental branch.

[LivInTheLookingGlass]

The minimum viable goal would be

1. Supporting all the same transports it does now
2. Modeled at least vaguely like the Javascript one (although that might not be possible, so loose goal)
3. The total elimination of the BaseDaemon clases
4. Stretch goal: The ability to understand WebSockets

If 4 is achieved, then you would no longer need bridge nodes between browsers and python.

[LivInTheLookingGlass]

Also, if it's able to support multiple event loops, that's definitely a large plus. Because some of them are significantly faster than AsyncIO.