Description
Bug report
Bug summary
The code from the Parasite Axes2 demo crashes when the x-axis is turned to a logarithmic scale.
This problem is new in the 2.2 branch.
Code for reproduction
Execute the code from Demo Parasite Axes2
https://matplotlib.org/gallery/axisartist/demo_parasite_axes2.html
This will work correctly.
Then set the x scale to log (before the show):
plt.xlim(.1)
plt.xscale('log')Now the program will produce many errors.
Actual outcome
Traceback (most recent call last):
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/backends/backend_qt5.py", line 519, in _draw_idle
self.draw()
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/backends/backend_agg.py", line 433, in draw
self.figure.draw(self.renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/artist.py", line 55, in draw_wrapper
return draw(artist, renderer, *args, **kwargs)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/figure.py", line 1475, in draw
renderer, self, artists, self.suppressComposite)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/image.py", line 141, in _draw_list_compositing_images
a.draw(renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/mpl_toolkits/axes_grid1/parasite_axes.py", line 286, in draw
self._get_base_axes_attr("draw")(self, renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/mpl_toolkits/axisartist/axislines.py", line 756, in draw
super(Axes, self).draw(renderer, inframe)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/artist.py", line 55, in draw_wrapper
return draw(artist, renderer, *args, **kwargs)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/axes/_base.py", line 2607, in draw
mimage._draw_list_compositing_images(renderer, self, artists)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/image.py", line 141, in _draw_list_compositing_images
a.draw(renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/artist.py", line 55, in draw_wrapper
return draw(artist, renderer, *args, **kwargs)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/mpl_toolkits/axisartist/axis_artist.py", line 1479, in draw
self._draw_ticks(renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/mpl_toolkits/axisartist/axis_artist.py", line 1211, in _draw_ticks
self.major_ticks.draw(renderer)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/mpl_toolkits/axisartist/axis_artist.py", line 330, in draw
locs = path_trans.transform_non_affine([loc])
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/transforms.py", line 2489, in transform_non_affine
return self._a.transform_non_affine(points)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/transforms.py", line 2263, in transform_non_affine
x_points = x.transform_non_affine(points)[:, 0:1]
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/transforms.py", line 2489, in transform_non_affine
return self._a.transform_non_affine(points)
File "/opt/anaconda2/envs/py2_conda52/lib/python2.7/site-packages/matplotlib/transforms.py", line 2265, in transform_non_affine
x_points = x.transform_non_affine(points[:, 0])
TypeError: list indices must be integers, not tuple
Expected outcome
There should be no errors.
Matplotlib version
- Operating system: linux and windows
- Matplotlib version: 2.2.2
- Matplotlib backend: Qt5Agg
- Python version: 2.7.15
Installed with conda (anaconda=5.2 python=2).
Same problem with python 3.6.5.
Problem is probably there in 3.0.0 but I have not tested that.
Note that it used to work without errors with anaconda 5.1 (matplotlib=2.1.2).
Possible fix
The problem is probably because of commit ffb5461
I can fix it by changing line 330 in mpl_toolkits/axisartist/axis_artist.py (methode draw of Ticks) from
locs = path_trans.transform_non_affine([loc])
to
locs = path_trans.transform_non_affine(np.array([loc]))
Another option would be to NOT use numpy indexing in transform_non_affine of BlendedGenericTransform of matplotlib/transforms.py