MatplotlibDeprecationWarning: Support for FigureCanvases without a required_interactive_framework attribute

Question

I'm trying to use matplotlib and numpy to plot a graph but keep running into this error:

MatplotlibDeprecationWarning: Support for FigureCanvases without a required_interactive_framework attribute was deprecated in Matplotlib 3.6 and will be removed two minor releases later.
plt.plot(cache_sizes, hit_rates[i])

Any idea how to solve it?

Here's the code:

#!/usr/bin/env python3
import os
import subprocess

import matplotlib.pyplot as plt
import numpy as np

# cache_sizes = np.arange(0, 120, 20)
cache_sizes = np.arange(1, 5)
policies = ["FIFO", "LRU", "OPT", "UNOPT", "RAND", "CLOCK"]
# these were acheived after running `run.sh`
hit_rates = [
    # FIFO
    [45.03, 83.08, 93.53, 97.42],
    # LRU
    [45.03, 88.04, 95.20, 98.30],
    # OPT
    [45.03, 88.46, 96.35, 98.73],
    # UNOPT
    # NOTE: was unable to finish running this one, as it took too long.
    [45.03, None, None, None],
    # RAND
    [45.03, 82.06, 93.16, 97.36],
    # CLOCK
    [45.03, 83.59, 94.09, 97.73],
]

for i in range(len(policies)):
    plt.plot(cache_sizes, hit_rates[i])

plt.legend(policies)
plt.margins(0)
plt.xticks(cache_sizes, cache_sizes)
plt.xlabel("Cache Size (Blocks)")
plt.ylabel("Hit Rate")
plt.savefig("workload.png", dpi=227)

Answer 1

Best I can tell, this is related to the backend you're using and how it generates the FigureCanvas object. Matplotlib has introduced a deprecation and some other softwares are still in the process of being updated accordingly. Sadly, for now, it's generating a warning that is not very user friendly and the matplotlib documentation provides no help.

I was seeing this error in PyCharm, which by default uses its own internal backend for matplotlib: 'module://backend_interagg'. Google provides a bunch of examples of people getting errors to do with FigureCanvas and backend_interagg. Fortunately, updating to the newest version of PyCharm solved it for me.

If your problem is not PyCharm related, then perhaps try checking which backend you're using, and perhaps use a different one.

import matplotlib
matplotlib.get_backend()

will show you which backend you're using

matplotlib.use('')

will give you a list of available backends to try and then use will also allow you to select one. e.g. matplotlib.use('GTK3Agg')