"RuntimeError: Event loop is closed" on 3.8/win32

[chlowell]

Using aiohttp 3.6.2 and azure-core 1.1.1 with Python 3.8/win32, this program:

import asyncio
from azure.core.pipeline.transport import AioHttpTransport, HttpRequest

async def main():
    transport = AioHttpTransport()
    request = HttpRequest('GET', 'https://www.bing.com')
    await transport.send(request)

asyncio.run(main())

Logs this to stderr:

Exception ignored in: <function _ProactorBasePipeTransport.__del__ at 0x0348F3D0>
Traceback (most recent call last):
  File "C:\Python38-32\Lib\asyncio\proactor_events.py", line 116, in __del__
    self.close()
  File "C:\Python38-32\Lib\asyncio\proactor_events.py", line 108, in close
    self._loop.call_soon(self._call_connection_lost, None)
  File "C:\Python38-32\Lib\asyncio\base_events.py", line 711, in call_soon
    self._check_closed()
  File "C:\Python38-32\Lib\asyncio\base_events.py", line 504, in _check_closed
    raise RuntimeError('Event loop is closed')
RuntimeError: Event loop is closed

Note the exception is logged to stderr; it doesn't crash the program. This may be related to aio-libs/aiohttp#4324. I see the same behavior with aiohttp 4.0.0a1 (and an appropriately modified AioHttpTransport).

This may also be related to aiohttp's shutdown behavior (see aio-libs/aiohttp#1925). The exception isn't logged when asyncio.run above is replaced with with:

loop = asyncio.get_event_loop()
loop.run_until_complete(main())

[chlowell]

I'm closing this issue because the SDK can't prevent this error being raised. As described in the first aiohttp issue I linked above, the error is due to different behavior of ProactorEventLoop as compared to SelectorEventLoop, will not be raised in a future version of aiohttp, and can be prevented by application-level workarounds (though those may not be worth using because the error typically causes only a noisy shutdown).