support async kernel shutdown #4425
Conversation
This is an "add-on" commit to PR jupyter#4412 in which kernel startup was addressed.
|
I believe what @kevin-bates means is: could this be also merged into 5.7.x branch and be part of the possible release that is going to handle the tornado version cap. |
|
EDIT: Issue was in jupyter_client. Since we're going to want to address this shortly, we should continue with this change since its independent of jc. |
kevin-bates
changed the title
kevin-bates
changed the title
| @@ -300,7 +301,8 @@ def shutdown_kernel(self, kernel_id, now=False): | |||
| type=self._kernels[kernel_id].kernel_name | |||
| ).dec() | |||
|
|
|||
| return super(MappingKernelManager, self).shutdown_kernel(kernel_id, now=now) | |||
| ret = yield gen.maybe_future(super(MappingKernelManager, self).shutdown_kernel(kernel_id, now=now)) | |||
| raise gen.Return(ret) | |||
There was a problem hiding this comment.
This already supported async shutdown_kernel without any change, so I don't think we should merge this. Since shutdown_kernel is not itself async, returning the super().shutdown_kernel() works if shutdown_kernel is async, since it will return a Future or coroutine object, which gets passed through to the caller of this method.
|
That's fine. If need be we can revisit since I had to back out the jupyter_client portion of this since there were restart side-effects that require further investigation. Closing. |
This is an "add-on" commit to PR #4412 in which kernel startup was addressed. As with that PR, this should be considered a "sibling PR" to the same jupyter_client PR - which I just updated with a shutdown-related commit.
I should have looked into shutdowns sooner. I thought it would be more involved than it was. Sorry about that.