Repository navigation
[MNT]: handle new deprecation from numpy (casting single-element arrays to scalar) #25744
Closed as not planned
Description
Summary
For context, see numpy/numpy#10615
One way matplotlib currently hits the new deprecation warning (there might be others) can be reproduced as
# t.py
import matplotlib.pyplot as plt
import numpy as np
fig, ax = plt.subplots()
ax.set_ylim(np.array([-1]), 1)$ python -Werror t.py
Traceback (most recent call last):
File "/Users/robcleme/dev/matplotlib/t.py", line 6, in <module>
ax.set_ylim(np.array([-1]), 1)
File "/Users/robcleme/dev/matplotlib/lib/matplotlib/axes/_base.py", line 3873, in set_ylim
return self.yaxis._set_lim(bottom, top, emit=emit, auto=auto)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/robcleme/dev/matplotlib/lib/matplotlib/axis.py", line 1214, in _set_lim
v0, v1 = self.get_major_locator().nonsingular(v0, v1)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/robcleme/dev/matplotlib/lib/matplotlib/ticker.py", line 1644, in nonsingular
return mtransforms.nonsingular(v0, v1, expander=.05)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/robcleme/dev/matplotlib/lib/matplotlib/transforms.py", line 2844, in nonsingular
vmin, vmax = map(float, [vmin, vmax])
^^^^^^^^^^
DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
Proposed fix
The following patch resolves the one issue for which I included a reproducer
diff --git a/lib/matplotlib/transforms.py b/lib/matplotlib/transforms.py
index 03c8a9f97a..aa345fb295 100644
--- a/lib/matplotlib/transforms.py
+++ b/lib/matplotlib/transforms.py
@@ -2841,7 +2841,7 @@ def nonsingular(vmin, vmax, expander=0.001, tiny=1e-15, increasing=True):
# Expand vmin, vmax to float: if they were integer types, they can wrap
# around in abs (abs(np.int8(-128)) == -128) and vmax - vmin can overflow.
- vmin, vmax = map(float, [vmin, vmax])
+ vmin, vmax = map(lambda x: float(np.asarray(x).item()), [vmin, vmax])
maxabsvalue = max(abs(vmin), abs(vmax))
if maxabsvalue < (1e6 / tiny) * np.finfo(float).tiny:I will open a PR with this patch, but careful inspection is needed to make sure the rest of the code base is safe.
Activity
jklymak commentedon Apr 21, 2023
Are we testing on numpy 1.25?
tacaswell commentedon Apr 21, 2023
our nightly cron job should be catching that and I am hitting it in an ad-hoc way on my local machine.
tacaswell commentedon Apr 21, 2023
See #25745 (comment)
I am going to close this as no-action as the user passing us
(1,)shaped arrays for scalars is a bug in the caller code that happened to work due to a (now deprecated) quirk of the numpy implementation.