David Brochart
David Brochart
Both releases 6.1.13 and 6.2.0 have been pulled, we are working towards a 7.0.0 release, see #642.
Maybe worth opening a new issue? This one is for jupyter_client 6.1.13 which was yanked.
Could you elaborate in which context this is happening?
This PR should fix it: https://github.com/jupyter/jupyter_console/pull/244 But I don't have the rights to merge it.
To illustrate, this is a POC where all cells are executed with `bg: true`. The widget runs in parallel with some CPU-bound code in another cell: https://user-images.githubusercontent.com/4711805/172254781-54a3d8ec-ef89-4c31-9029-235bc64d73d6.mp4
> This sounds like it might be better in the metadata? i.e., it's independent of the message type/content, but is metadata telling the receiver how to process the message. Maybe,...
> Is the proposal here that each shell message with this bit set will have a new thread created to handle it? I don't know, that might be a lot...
I didn't work on putting control channel processing on a separate thread, maybe @JohanMabille did as part of the debugger? I remember @SylvainCorlay saying that the control channel should not...
I don't like this approach. It basically means that the user has no guarantee about what they are asking for. I prefer the "subshell" solution (see https://github.com/ipython/ipykernel/pull/955), where the user...
I don't see them as orthogonal problems. Subshells can serve both purposes. It is fine to let the user choose if they want a new subshell (i.e. a new thread)...