[Bug]: Animation seems to keep the last frame of an old loop

[sdasda7777]

Bug summary

I noticed this behaviour few days ago. I'm not very familiar with matplotlib so this might be user error, but it's really strange and unexpected.

Code for reproduction

import matplotlib.pyplot as plt
import matplotlib.animation as animation
from matplotlib.collections import LineCollection
import numpy as np

fig, ax = plt.subplots()

dots = []

def init():
    plt.cla()
    ax.set_xlim(-1, 3)
    ax.set_ylim(-1, 2)
    
    global dots
    
    # Initialize vertices
    dot1, = ax.plot([0], [0], 'o', color='red')
    dot2, = ax.plot([0], [1], 'o', color='green')
    dots = [dot1, dot2]
    
    return sum([dots], [])

def animate(frame_index):
    for k, dot in enumerate(dots):
        y = (frame_index % 15) / 15
        dot.set_data([[k], [y]])
    
    return sum([dots], [])

ani = animation.FuncAnimation(fig, animate, frames=15, init_func=init, blit=True)

plt.show()

Actual outcome

When I move the camera using the move tool during the animation, last frame of the old animation seems to linger in some sort of buffer, until the camera is moved again.

Here, the top two dots are from the old animation loop, and the two bottom ones are from the new animation loop.

Expected outcome

I would not expect old frames to be kept anywhere, period.

Additional information

No response

Operating system

No response

Matplotlib Version

3.8.4

Matplotlib Backend

No response

Python version

Python 3.10.4

Jupyter version

No response

Installation

None

[QuLogic]

I cannot reproduce what you are saying; there are never two copies of the markers.

The only thing odd is that as the animation loops, it calls init again, which calls sets the limits, so everything snaps back at that point, though if you have the mouse down, then moving it a little bit will jump back to wherever you were.

[sdasda7777]

Thanks for trying, may I ask about your library version? I was able to replicate it on both 3.8.4 and 3.9.0. I also asked about this on Gitter before posting here, and one person was supposedly able to replicate it to some extent (also hinted that it could be a race condition, which does seem increasingly likely).

For me the axis does not seem to move on animation reset: