Using imshow with Quantity, LogNorm, and colorbar crashes

[dstansby]

Trying to use imshow() with a Quantity and LogNorm, and then trying to draw a colorbar doesn't work. I suspect this needs to be fixed in matplotlib, but thought I would open it here too just in case.

A simple example is:

import astropy.units as u
import matplotlib.colors as mcolor
import matplotlib.pyplot as plt
import numpy as np

data = np.random.rand(2, 2) * u.m

fig, ax = plt.subplots()
im = ax.imshow(data, norm=mcolor.LogNorm())
fig.colorbar(im)

plt.show()

which fails with

Traceback (most recent call last):
  File "/Users/dstansby/github/matplotlib/examples/units/units_imshow.py", line 9, in <module>
    im = ax.imshow(data, norm=mcolor.LogNorm())
  File "/Users/dstansby/github/matplotlib/lib/matplotlib/__init__.py", line 1348, in inner
    return func(ax, *map(sanitize_sequence, args), **kwargs)
  File "/Users/dstansby/github/matplotlib/lib/matplotlib/axes/_axes.py", line 5591, in imshow
    im._scale_norm(norm, vmin, vmax)
  File "/Users/dstansby/github/matplotlib/lib/matplotlib/cm.py", line 291, in _scale_norm
    self.autoscale_None()
  File "/Users/dstansby/github/matplotlib/lib/matplotlib/cm.py", line 476, in autoscale_None
    self.norm.autoscale_None(self._A)
  File "/Users/dstansby/github/matplotlib/lib/matplotlib/colors.py", line 1522, in autoscale_None
    super().autoscale_None(np.ma.masked_less_equal(A, 0, copy=False))
  File "/Users/dstansby/miniconda3/envs/dev/lib/python3.9/site-packages/numpy/ma/core.py", line 2043, in masked_less_equal
    return masked_where(less_equal(x, value), x, copy=copy)
  File "/Users/dstansby/miniconda3/envs/dev/lib/python3.9/site-packages/numpy/ma/core.py", line 1932, in masked_where
    cond = mask_or(cond, a._mask)
  File "/Users/dstansby/miniconda3/envs/dev/lib/python3.9/site-packages/numpy/ma/core.py", line 1742, in mask_or
    return make_mask(m1, copy=copy, shrink=shrink, dtype=dtype)
  File "/Users/dstansby/miniconda3/envs/dev/lib/python3.9/site-packages/numpy/ma/core.py", line 1635, in make_mask
    result = _shrink_mask(result)
  File "/Users/dstansby/miniconda3/envs/dev/lib/python3.9/site-packages/numpy/ma/core.py", line 1544, in _shrink_mask
    if m.dtype.names is None and not m.any():
  File "/Users/dstansby/github/astropy/astropy/units/quantity.py", line 1645, in any
    raise TypeError("cannot evaluate truth value of quantities. "
TypeError: cannot evaluate truth value of quantities. Evaluate array with q.value.any(...)

This is present on both (Matplotlib master, astropy master) and (Matplotlib 3.3.4, astropy 4.2) with the same error message.

[pllim]

From matplotlib/matplotlib#19476 , seems like it doesn't work without using Quantity either. Hmm.

[dstansby]

Can you please provide version info? Thanks.

I've updated the original post. Yes, I don't think this is supported in Matplotlib at the moment, but I'm taking a look at the upstream issue.

[neutrinoceros]

I've opened matplotlib/matplotlib#27682 to try and fix this issue.
I expect I'll need to open a PR here too to add a test. To be discussed upstream.