Request kernel shutdown about nteract HOT 10 CLOSED

ah, yes

from nteract.

Safia, do this!

from nteract.

Is this the issue about kernels being alive after quitting the app?

from nteract.

Yeah our cleanup isn't fully happening right now, leaving the connection files around.

from nteract.

The process seems to keep running as well. Check pgrep -f "python3 -m ipy" after starting and quitting nteract a few times.

from nteract.

Pressing Cmd+q is wired directly to app.quit() without creating a KILL_KERNEL action or calling cleanupKernel().

We can add a send(focusedWindow, 'menu:kill-kernel') to the quit menu item, but then you need to delay calling app.quit() to give it a chance to do stuff. What is a good way of ipc()ing and waiting for that message to have been completed. Should we send a message back via ipc and then app.quit()?

from nteract.

The awkward part here is the difference between OS X and all the others. cmd-q should tell each running window to close their kernel (which means ipc) or we'd have to track the kernels on the main thread to shut them down there (what if the BrowserWindow is no longer responding).

from nteract.

Ok, I've noticed that beforeunload and unload events are both fired when you ctrl-c kill an application, as well as the rest of the quit action. What we're definitely running into though is that we've been relying on an asynchronous flow for killing the kernel. When these two are triggered, we need to handle them immediately.

from nteract.

Basically we can't (and shouldn't) rely on dispatch for killing a kernel for forced lifecycle events. Otherwise our dispatched kill kernel won't happen because the reduction won't happen until the next tick.

from nteract.

This thread has been automatically locked because it has not had recent activity. Please open a new issue for related bugs and link to relevant comments in this thread.

from nteract.