PR: Fix conflict between AsyncDispatcher and run_sync function of jupyter_core

[hlouzada]

Description of Changes

• Wrote at least one-line docstrings (for any new functions)
• Added unit test(s) covering the changes (if testable)

Allow AsyncDispatcher to get the "main thread" running loop without depending on asyncio.set_event_loop and asyncio.get_running_loop so it won't trigger conflict with how jupyter_core.utils.run_sync handles creating, setting and executing event loops.

Also, refactor type hints to help with hinting decoreted async functions.

Affirmation

By submitting this Pull Request or typing my (user)name below,
I affirm the Developer Certificate of Origin
with respect to all commits and content included in this PR,
and understand I am releasing the same under Spyder's MIT (Expat) license.

I certify the above statement is true and correct:

@hlouzada

[pep8speaks]

Hello @hlouzada! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2024-06-09 23:28:07 UTC

[ccordoba12]

Thanks @hlouzada for your work on this! I only left some formatting suggestions for you.

[ccordoba12]

@dalthviz, could you check that you can start remote consoles on Windows with this PR? Thanks!

@hlouzada discovered that these fixes are necessary mainly there (although they'll be useful for Linux and Mac too).

[dalthviz]

Gave this a check on Windows and got some different behaviors. In an old dev env I have I'm got a RuntimeError:

Traceback (most recent call last):
  File "e:\acer\documentos\spyder\spyder otros\hlouzada\spyder\external-deps\qtconsole\qtconsole\manager.py", line 27, in poll
    super().poll()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_client\restarter.py", line 121, in poll
    if not self.kernel_manager.is_alive():
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_core\utils\__init__.py", line 173, in wrapped
    return loop.run_until_complete(inner)
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 625, in run_until_complete
    self._check_running()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 584, in _check_running
    raise RuntimeError('This event loop is already running')
RuntimeError: This event loop is already running
Traceback (most recent call last):
  File "e:\acer\documentos\spyder\spyder otros\hlouzada\spyder\external-deps\qtconsole\qtconsole\manager.py", line 27, in poll
    super().poll()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_client\restarter.py", line 121, in poll
    if not self.kernel_manager.is_alive():
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_core\utils\__init__.py", line 173, in wrapped
    return loop.run_until_complete(inner)
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 625, in run_until_complete
    self._check_running()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 584, in _check_running
    raise RuntimeError('This event loop is already running')
RuntimeError: This event loop is already running

However, in a newer env I got just the connection dialog showing that the connection errored (when using the container private ip):

Checking the container files did a test using localhost/127.0.0.1 and the mapped port and I got a connection to be stablished but then all you see is the console restarting continuously:

and for some reason from the cmd that I used to lauch Spyder seems like the script to install spyder on dev mode starts to run?:

Also, after closing Spyder, I still got some output:

so I had to interrupt things (Ctrl + C):

[hlouzada]

Gave this a check on Windows and got some different behaviors. In an old dev env I have I'm got a RuntimeError:

Traceback (most recent call last):
  File "e:\acer\documentos\spyder\spyder otros\hlouzada\spyder\external-deps\qtconsole\qtconsole\manager.py", line 27, in poll
    super().poll()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_client\restarter.py", line 121, in poll
    if not self.kernel_manager.is_alive():
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_core\utils\__init__.py", line 173, in wrapped
    return loop.run_until_complete(inner)
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 625, in run_until_complete
    self._check_running()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 584, in _check_running
    raise RuntimeError('This event loop is already running')
RuntimeError: This event loop is already running
Traceback (most recent call last):
  File "e:\acer\documentos\spyder\spyder otros\hlouzada\spyder\external-deps\qtconsole\qtconsole\manager.py", line 27, in poll
    super().poll()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_client\restarter.py", line 121, in poll
    if not self.kernel_manager.is_alive():
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\site-packages\jupyter_core\utils\__init__.py", line 173, in wrapped
    return loop.run_until_complete(inner)
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 625, in run_until_complete
    self._check_running()
  File "C:\Users\dalth\anaconda3\envs\spyder-dev\lib\asyncio\base_events.py", line 584, in _check_running
    raise RuntimeError('This event loop is already running')
RuntimeError: This event loop is already running

However, in a newer env I got just the connection dialog showing that the connection errored (when using the container private ip):

Checking the container files did a test using localhost/127.0.0.1 and the mapped port and I got a connection to be stablished but then all you see is the console restarting continuously:

and for some reason from the cmd that I used to lauch Spyder seems like the script to install spyder on dev mode starts to run?:

Also, after closing Spyder, I still got some output:

so I had to interrupt things (Ctrl + C):

Thanks for the info @dalthviz, I'll take a look and get back to you.

[dalthviz]

Just in case, did a check on a Linux VM and I got the same restarting kernel behavior when conneting:

Maybe I'm missing setting up something? 🤔

[ccordoba12]

Maybe I'm missing setting up something? 🤔

After some digging, @hlouzada and @dalthviz think this could be due to a networking problem in @dalthviz's system. However, we'll continue to investigate that and try to find the issue before 6.0 final is released.

[ccordoba12]

Thanks @hlouzada!