Problem while changing current figure size in Jupyter notebook

[Pibmy]

Bug report

Bug summary

In Jupyter Notebook, the problem occurs when dynamicaly changing the size of a ploted figure. The image is cropped and not totally displayed. (See code and example below)

Code for reproduction

1st cell > import declaration:

%matplotlib notebook

import pandas as pd
import matplotlib.pyplot as plt

2nd cell > declaring dataframe and plot:

df = pd.DataFrame([1,2,3,4,6])
df.plot()

3rd cell > changing current figsize:

fig = plt.gcf()
fig.set_size_inches(5,5)

As seen on the second screenshot the image is cropped and I had to manualy resize it for a good display.

Note

• the problem doesn't occur when ploting and changing size are in the same notebook cell
• the problem doesn't occur when the zoom lvl of my browser is between 50% 70%

Matplotlib version

• Operating system: 'Windows 10 Profesionnel' - v1709 - vsystem 16299.371
• Browser : 'Firefox 59.0.3 (64 bits)'
• Matplotlib version: '2.1.0'
• Matplotlib backend (print(matplotlib.get_backend())): nbAgg
• Python version: '3.6.3 |Anaconda, Inc.| (default, Oct 15 2017, 03:27:45) [MSC v.1900 64 bit (AMD64)]'
• Jupyter version (if applicable): '4.4.0'
• Other libraries: pandas version '0.20.3'

Installation by default with anaconda with version 'conda 4.3.30'

[avinashmnit30]

Use fig.set_size_inches(5,5, forward = False) .
Forward parameter is True by default.

[tacaswell]

So it works if you do not forward the size?

Can you try with mpl 2.2.2?

[avinashmnit30]

I checked. Similar problem (with the same solution) exists in mpl 2.2.2 as well.

[Pibmy]

Well changing forward option to false isn't really correcting the problem. It's more a turnaround. It seems to disable the autoresize of the figure window.

So if you want to scale down the figure it's ok, but if you want to scale it up you steel have a croped image. See example bellow.

%matplotlib notebook
import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame([1,2,3,4,6])
df.plot()

Scale down exemple:

fig = plt.gcf()
fig.set_size_inches(5,5, forward = False)

Scale up exemple:

fig = plt.gcf()
fig.set_size_inches(8,8, forward = False)

This behavior seems logical after all.

---

An other turn around solution could be to first display a big image then scale it down.

[ImportanceOfBeingErnest]

It would be a bit strange if using forward=False would solve this problem - after all you want to forward the figure size to take immediate effect.

In any case, I cannot reproduce this problem with matplotlib 2.2.2. The figure size changes as expected.

[Pibmy]

Agreed with you @ImportanceOfBeingErnest

Maybe the pb has been fixed in matplotlib 2.2.2.
I'll try it but I have some computation for the next hours so i have to wait a bit...

[Pibmy]

Sorry i was absent for a week.

The problem seems to be corected in matplotlib 2.2.2 i don't have this issue in version 2.2.2

Thanks every one for your participation :)

Config

• Operating system: 'Windows 10 Profesionnel' - v1803 - vsystem 17134.48
• Browser : 'Firefox 60.0.1 (64 bits)'
• Matplotlib version: '2.2.2'
• Matplotlib backend (print(matplotlib.get_backend())): nbAgg
• Python version: '3.6.5 |Anaconda, Inc.| (default, Mar 29 2018, 13:32:41) [MSC v.1900 64 bit (AMD64)]'
• Jupyter version (if applicable): '4.4.0'
• Other libraries: pandas version '0.23.0'