[bug] Quits if "Preview Manim Cell" before it finishes

[VladimirFokow]

Is this happening on your side as well?

Example code I'm trying:

from manimlib import *

class Test(Scene):
    def construct(self):
        ##
        circle = Circle()
        self.play(ShowCreation(circle), run_time=3)

Steps:

manimgl

# Press "Preview Manim Cell"

# While it's still executing: "Preview Manim Cell" again -> it quits

[Splines]

This is wanted behavior as described here.

The currently running command is canceled and a new one is issued. I like this behavior as I don't want to be forced to wait for an animation to finish entirely before previewing it again (e.g. with adjusted parameters).

[VladimirFokow]

hmm.. After reading that text - I thought it would re-run the cell.. not quit.

For me it just quits (the window closes)

[Splines]

For me it works just fine (running in WSL Ubuntu):

2024-10-22.17-37-13.mp4

[VladimirFokow]

ok, then maybe OS differences (I'm on MacOS) - will have to look into that..

[bhoov]

I can replicate @VladimirFokow's experience. Pressing the "preview" button before the animation is finished rendering exits the iPython terminal.

I am also on MacOS

[VladimirFokow]

looks like on macos, if you press ctrl+c while checkpoint_paste() is still executing, it quits the whole ipython

[bhoov]

yes, this happens to me as well. I get the following error when pressing ctrl+c during checkpoint_paste() execution, which is the same error that we get when pressing preview cell before the previous preview is finished

.../lib/python3.10/multiprocessing/resource_tracker.py:224: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
  warnings.warn('resource_tracker: There appear to be %d '

I like the functionality of interrupting the current render and starting a new one when preview cell is clicked -- this is especially helpful for long running animations.

[Splines]

Just to be sure, could you try with Python 3.12 as well, that's the version I'm using.

[bhoov]

I can confirm this still happens with python=3.12

.../lib/python3.12/multiprocessing/resource_tracker.py:254: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
  warnings.warn('resource_tracker: There appear to be %d '

[Splines]

Mmh ok, that's unfortunate. So maybe that's something for a new Manim issue with your specific system specs? Feel free to link or use the video from here to show how it's supposed to work.

[bhoov]

Just started debugging this tonight, haven't figured out exactly how to fix.

It seems like the issue of Ctrl-C killing the ipython instance is because of the way the ipython shell is launched. Specifically, if the shell is initialized through manim (see this line), ctrl+c will kill the terminal.

However, normal ipython does not behave this way. Instead, ctrl+c sends a "keyboard interrupt" to the running ipython, which corresponds to the nice behavior we see in @Splines' video

What is different about the ipython shell created in these two manners such that it fails on MacOS but not on Windows? I'm baffled 🤷