Skip to main content

async/await without event-loop

Project description

AsyncGui

A thin layer that helps to build an async/await-based api using a callback-based api.

How to use

Despite its name, asyncgui has nothing to do with gui. You can wrap any kind of callback-based api in it. The simplest example of it would be sched, whose the whole feature is a timer. All you need is just few lines of code:

import types
import sched
import asyncgui

s = sched.scheduler()

# wrapping the timer api takes only three lines
@types.coroutine
def sleep(duration):
    yield lambda step_coro: s.enter(duration, 10, step_coro)


async def main():
    print('A')
    await sleep(1)  # Now you can sleep in an async-manner
    print('B')
    await sleep(1)
    print('C')

asyncgui.start(main())
s.run()

And you already have a structured concurrency api as well:

async def print_numbers():
    for i in range(10):
        await sleep(.1)
        print(i)


async def print_letters():
    for c in "ABCDE":
        await sleep(.1)
        print(c)


async def main():
    from asyncgui.structured_concurrency import or_
    # Let print_letters() and print_numbers() race.
    # As soon as one of them finishes, the other one gets cancelled.
    tasks = await or_(print_letters(), print_numbers())
    if tasks[0].done:
        print("print_letters() won")
    else:
        print("print_numbers() won")
    print('main end')
A
0
B
1
C
2
D
3
E
print_letters() won
main end

Why not asyncio ?

The above example may not attract you because you can just replace sched with asyncio or Trio, and can use thier sleep function (asyncio.sleep and trio.sleep). But in a read-world situation, that might not be an option: Kivy required massive changes in order to adapt to asyncio and Trio, asyncio-tkinter's codebase is quite big as well.

The reason they needed lots of work is that they had to merge two event-loops into one. One is from the gui libraries. The other one is from async libraries. You cannot just simply run multiple event-loops simultaneously in one thread... maybe.

On the other hand, asyncgui doesn't require a lot of work as shown above because it doesn't have an event-loop. asyncgui and a library who has an event-loop can live in the same thread seemlessly because of it.

So, is asyncgui superior to asyncio ?

No, it is not. For asyncgui, many features that exist in asyncio are either impossible or hard to implement because of the lack of event-loop. Thus those features need to be specific to each event-loop you are using. You've already witnessed one, the sleep.

asyncgui is not usefull then.

There is at least one situation where asyncgui shines. When you are creating a gui app, you probably want the app to quickly react to the gui events, like pressing a button. This is problematic for asyncio because it cannot immediately start/resume a task. It can schedule a task to eventually start/resume but not immediate, which causes to spill gui events. As a result, you need to use callback-based api for that, and thus you cannot fully receive the benefits of async/await.

If you use asyncgui, that never happens because:

  • asyncgui.start() immediately starts a task.
  • asyncgui.Event.set() immediately resumes the tasks waiting for it to happen.

In summary, if your program needs to react to something immediately, asyncgui is for you. Otherwise, it's probably not worth it.

Installation

If you use this module, it's recommended to pin the minor version, because if it changed, it means some important breaking changes occurred.

poetry add asyncgui@~0.5
pip install "asyncgui>=0.5,<0.6"

Tested on

  • CPython 3.7
  • CPython 3.8
  • CPython 3.9
  • CPython 3.10

Async-libraries who relies on this

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

asyncgui-0.5.3.tar.gz (17.3 kB view hashes)

Uploaded Source

Built Distribution

asyncgui-0.5.3-py3-none-any.whl (16.7 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page