when do you expect to see this working?

[johanneswilm]

Hey,
great that you started this project! Do I understand correctly that the setup you are running consists of:

NGINX, Tornado Server, TornadIO2, django-tornadio2, Django?

That's what I gather is what is needed from the TornadIO docs and your statement here: stephenmcd/django-socketio#19 (comment) .

I also understand that you already achieve this or that you are close to it, but that you cannot open source the needed code for now. May I ask how much code you would reckon would have to be written (and committed as open source) in order for this stack to be functional?

[tabouassaleh]

Regarding the stack: Correct as a goal. In my dev setup, I haven't put it all behind nginx yet - I'm communicating with Tornado directly.

Minimal code required to get started is a management command similar to the one in https://github.com/k1000/django-tornadio.

I will try to get some stuff in soon. At this point, I'm starting to feel there is more good than harm in releasing some code without proper cleanup/commenting/testing.

[johanneswilm]

Ok very good! I will be waiting for that then...

[tabouassaleh]

I haven't released any new code yet because it's still under heavy development. Turns out once your combine tornado with django you kill the async part of tornado and end up with a single-threaded synchronous server that block on every request until the client consume all the data. Needless to say, this is horrible even for a toy project if you're going to have external users.

I'm working with a revised architecture that uses Brukva and Celery to dispatch socket.io events to async workers, and once they're done, asynchronously send the data back to the client.