Steven Silvester

No, it did work previously, this test was not skipped on Windows: https://github.com/ipython/ipykernel/blob/eddd3e666a82ebec287168b0da7cfa03639a3772/tests/test_async.py#L32. Additionally, the Jupyter Client test needs to pass before we merge, since it is treating the kernel...

Great, we can skip that test on Windows, and concentrate on getting client to pass.

Aha! The issue is that we need to update [`SignalTestKernel`](https://github.com/jupyter/jupyter_client/blob/4ffbfcf5809ad8913724dd1359089ac3d42031e1/tests/signalkernel.py#L14C7-L14C23) in `jupyter_client`, which is directly subclassing `BaseKernel`. I'd like to be able to update that class to prove it it...

More progress, I see now that we have to handle `self.kernel.shell_interrupt.get` now in `SignalTestKernel.do_execute`.

I'm afraid I don't know how to proceed. It seems as if the changes from #1210 are incompatible with the changes in this PR. I am getting deadlocked on reading...

What's really confusing to me is why `nest_asyncio` would show up in the stack trace: ```python async with asyncio.Lock(): File "/Users/silvester/workspace/.venvs/ipykernel/lib/python3.10/site-packages/pluggy/_manager.py", line 80, in _hookexec return self._inner_hookexec(hook_name, methods, kwargs, firstresult)...

@ccordoba12 @tacaswell @haperilio do you have an ideas what might be happening here?

It seems to be an intermittent failure. I saw it in #1089, but then it passed on the merge commit. I just kicked it.

cc @maartenbreddels @martinRenou

Hi @gramster, these types are provided by the jedi completer in [ipython](https://github.com/ipython/ipython/blob/3f0bf05f072a91b2a3042d23ce250e5e906183fd/IPython/core/completer.py#L1898)