pyplot fill_between warning since upgrade of numpy to 1.10.10

[mhooreman]

Hello,

Since I've upgraded my python 3 environment, I get a warning, as demonstrated below:

%matplotlib inline
import matplotlib.pyplot as plt
plt.fill_between([0, 1], [0, 1])

This gives the warning:

/usr/local/lib/python3.4/dist-packages/matplotlib/collections.py:590: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
    if self._edgecolors == str('face')

The upgraded libraries are listed below, I'm pretty sure that this comes from the numpy upgrade

• Pillow: 2.9.0 -> 3.0.0
• XlsxWriter: 0.7.3 -> 0.7.5
• numpy: 1.9.3 -> 1.10.0
• path.py: 8.1.1 -> 8.1.2
• pexpect: 3.3 -> 4.0.1
• psutil: 3.2.1 -> 3.2.2
• requests: 2.7.0 -> 2.8.0

Best regards.

[jenshnielsen]

Which version of matplotlib? I think we have already fixed this on master and it will be in matplotlib 1.5.0

At the moment you can ignore the warning. It will not actually break anything before Numpy actually makes the change.

[tlnagy]

I also ran into this problem with numpy v1.10.0 and matplotlib 1.4.3

[jenshnielsen]

Please test if this is still an issue with 1.5.0rc2 or master. I think we have already fixed this between 1.4.3 and the current master

[jenshnielsen]

Closing as this is fixed in master https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/collections.py#L624

[mhooreman]

Thanks a lot.

Do you have any idea of the release date of the 1.5.0?

Best regards.

[tacaswell]

"soon", I think we are down to documentation related issues.

https://github.com/matplotlib/matplotlib/milestones/next%20point%20release%20(1.5.0)

[pm0kjp]

I've upgraded to matplotlib 1.5.1, but still get the error when doing a scatterplot (regular plot is no problem...) Any ideas? Should I downgrade numpy?

[jenshnielsen]

@pm0kjp The original issue should be fixed but you are perhaps seeing something similar. Can you please provide a minimal example that demonstrates the issue that you are seeing. Also this is not an error with the latest release of numpy but a warning that something will change in the future thus there is no reason to downgrade numpy.