Rationale behind iteration context and the stream method about aiostream HOT 2 CLOSED

Thanks for creating this issue, maybe it's something that should be better documented.

Consider the following asynchronous generator:

async def count_to_ten():
    try:
        for i in range(10):
            yield i
            await asyncio.sleep(1)
    finally:
        print('cleaning up')

Now, say you're only interested in the first three values:

async def main():
    xs = stream.enumerate(agen())
    async for i, x in xs:
        if i == 3:
            break
        print(x)
    print('done!')

And you run the main coroutine using the standard boiler-plate code:

if __name__ == '__main__':
    loop = asyncio.get_event_loop()
    loop.run_until_complete(main())
    loop.close()

Let's check the output:

[...]: UserWarning: AsyncIteratorContext is iterated outside of its context
  "AsyncIteratorContext is iterated outside of its context")
0
1
2
done!

And that's it, no clean up! According to PEP 525, that's because you're actually supposed to add the following line before closing the loop:

loop.run_until_complete(loop.shutdown_asyncgens())

Notice however that 'cleaning up' will appear after 'done!' which is usually not what you want. By using the stream method, the clean up gets to run before leaving the corresponding context:

async def main():
    xs = stream.enumerate(agen())
    async with xs.stream() as streamer:
        async for i, x in streamer:
            if i == 3:
                break
            print(x)
    print('done!')

Note that if you use await instead of the asynchronous iteration, the streaming context is not necessary:

async def main():
    xs = stream.iterate(agen())
    print(await xs[:3])
    print('done!')

Now if you can think of a more intuitive interface, I'd be happy to look into it!

from aiostream.

Looks like I stepped into a minefield here.

dabeaz/curio#176

https://vorpus.org/blog/some-thoughts-on-asynchronous-api-design-in-a-post-asyncawait-world/

This is actually broken in python IMHO.. If DB can't come up with any good solution, I am not even going to try. Closing.

from aiostream.