Is is possible to run reactor's loop iteration by iteration manually? about async HOT 11 CLOSED

I want to use this project from ruby script for weechat. How that works is that I can register hooks to get my callback called every X ms (in my case probably 10, to be determined), "do my work" (in this case I would check if any data arrived and process them) and return from the callback. At that point weechat takes back the control and I'm waiting for another callback call.

However once I leave the ruby code, the GVL is locked and any background processing is therefore not possible since no thread can acquire GVL... This could possibly be solved somehow by modifying the weechat's ruby plugin (I don't know), but still, I would like to keep this singlethreaded :)

from async.

Okay, there is an example how to do it with current interface.

from async.

I've implemented a similar design to this in the past.

We could certainly look at doing it. I think it's a reasonable approach.

Certain invariants are retained when using the current design, i.e. guarantees about task state.

from async.

If you want to try this another way, you could probably do the following (untested):

reactor = Async::Reactor.new

reactor.run do
    # ...
    reactor.stop
end

In the past, I've done something similar but there are sometimes useful variants like: run_once, run_until(time), etc.

from async.

#run_once would be great :) Thanks for the suggestion, this looks like I could avoid monkey-patching reactor all together, so will try it and see what it does. Thanks :)

from async.

The #stop solution might be the simplest and easiest solution to use rather than #run_once. Because it gives the flexibility to the user as to the stopping conditions (timeout, number of iterations). However, I'm not adverse to #run_once as long as it doesn't make the code more fragile.

One thing I'd like to understand is what # Do some of my own work is. In almost every case, you'd simply be better off scheduling a task or doing your own work on a separate thread. Is there some reason why this is useful/necessary?

from async.

Ah yep.

I think in the first case, can you try using #stop. Can you report back whether that works or not? If not, let's see if we can find another solution (e.g. #run_once).

from async.

How is this working out for you?

from async.

I'm sorry it took me so long to get back to this, I've picked example from async-io and tried to do it (https://github.com/socketry/async-io/blob/master/spec/async/io/echo_spec.rb) , basically I've copied the code from the readme, however I cannot figure out where should I put the #stop to in order for it to work :/

Could you please nudge me in the right direction? :)

from async.

That indeed seems to work! Thank you very much :)

from async.

You are most welcome, let me know how you get on. The only thing I can think of is it might still make sense to have a "Run one iteration" function rather than the timer... but for now this is the simplest thing I can think of without digging into the code.

from async.