*loop* parameter was removed from asyncio.Queue() in Py3.10

[prjemian]

As part of unit testing a new feature in apstools, encountered this error in the Py3.10 unit testing workflow:

>               RE(my_plan(msg, default, agree, bypass))

/home/runner/work/apstools/apstools/apstools/plans/tests/test_input_plan.py:38: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:893: in __call__
    plan_return = self._resume_task(init_func=_build_task)
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:1032: in _resume_task
    raise exc
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:1663: in _run
    raise err
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:1498: in _run
    msg = self._plan_stack[-1].throw(stashed_exception or resp)
/home/runner/work/apstools/apstools/apstools/plans/tests/test_input_plan.py:13: in my_plan
    yield from request_input(
/home/runner/work/apstools/apstools/apstools/plans/input_plan.py:45: in request_input
    r = yield from bps.input_plan(full_text)
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/plan_stubs.py:554: in input_plan
    return (yield Msg('input', prompt=prompt))
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:1583: in _run
    new_response = await coro(msg)
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/run_engine.py:2431: in _input
    async_input = AsyncInput(self.loop)
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/site-packages/bluesky/utils/__init__.py:1052: in __init__
    self.q = asyncio.Queue(loop=self.loop)
/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/asyncio/queues.py:34: in __init__
    super().__init__(loop=loop)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <[AttributeError("'Queue' object has no attribute '_maxsize'") raised in repr()] Queue object at 0x7fce7f4ae9b0>

    def __init__(self, *, loop=_marker):
        if loop is not _marker:
>           raise TypeError(
                f'As of 3.10, the *loop* parameter was removed from '
                f'{type(self).__name__}() since it is no longer necessary'
            )
E           TypeError: As of 3.10, the *loop* parameter was removed from Queue() since it is no longer necessary

/home/runner/micromamba-root/envs/anaconda-test-env-py-3.10/lib/python3.10/asyncio/mixins.py:17: TypeError

The code in question is: https://github.com/bluesky/bluesky/blob/72ea2473359e3020a27f4338bad5446534dc7d02/bluesky/utils/init.py#L1052

Since it breaks unit tests at this time, it is labeled High Priority here.

[tacaswell]

I could have sworn I got all of these.

We apparently do not set that code.....

[hyperrealist]

My understanding is that in python 3.10 asyncio synchronization primitives like Lock and Queue can be instantiated anywhere without tying them to a specific loop at instantiation. The high-level asyncio API then reads the currently running loop using the _get_loop() method of the primitive when needed, i.e., where we would await self.q.get(), for example. I think the main advantage is that loop capturing is now automatically handled in a thread-safe way, also leading to cleaner code if one were to stick to python 3.10+.

See this discussion for a more detailed reasoning behind this change in python.

I propose we simply let python handle the loop parameter on capable versions.

[tacaswell]

Closed by #1566