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 'scheduler.enter()' 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 structured concurrency apis 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.

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. The implementation of those features needs to be specific to the event-loop you are using. You've already witnessed one, the sleep.

asyncgui is not usefull then.

There is at least one situation that 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 apis 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

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-0.5.5.tar.gz (12.2 kB view details)

Uploaded Source

Built Distribution

asyncgui-0.5.5-py3-none-any.whl (11.5 kB view details)

Uploaded Python 3

File details

Details for the file asyncgui-0.5.5.tar.gz.

File metadata

  • Download URL: asyncgui-0.5.5.tar.gz
  • Upload date:
  • Size: 12.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.1.11 CPython/3.8.12 Linux/5.4.0-122-generic

File hashes

Hashes for asyncgui-0.5.5.tar.gz
Algorithm Hash digest
SHA256 f5b0749a11b1766cb6ca5ced54a3a41807cbee17e3ee4b3866b985953243f0d1
MD5 106b3c5e12e8903a929a41f21c09a832
BLAKE2b-256 e82a2049a49aee3a1ef19385e5ca8da7420f07034014a1b1f9912a17ab41d6db

See more details on using hashes here.

File details

Details for the file asyncgui-0.5.5-py3-none-any.whl.

File metadata

  • Download URL: asyncgui-0.5.5-py3-none-any.whl
  • Upload date:
  • Size: 11.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.1.11 CPython/3.8.12 Linux/5.4.0-122-generic

File hashes

Hashes for asyncgui-0.5.5-py3-none-any.whl
Algorithm Hash digest
SHA256 f90548952c5cef3b13dfc2d95a7e44c613935453ef51241b9205d41694b0b6fc
MD5 19b53a7cb34bf22d54664091c300ea0d
BLAKE2b-256 7bc324dccd553edcd18c16d7639ce8f22118cdc4fb719cb5644cc7c95b04fed1

See more details on using hashes here.

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