Closed
Description
Bug report
Bug summary
plt.errorbar borks with `TypeError' when the first element of the error array is masked. This does not seem to happen for other masked elements.
Code for reproduction
x, y, e = np.random.randn(3, 100)
e = np.ma.array(e)
e[0] = np.ma.masked
plt.errorbar(x, y, e)Actual outcome
Here's the traceback
TypeError Traceback (most recent call last)
<ipython-input-127-b58bf89a7e18> in <module>()
3 e[0] = np.ma.masked
4
----> 5 plt.errorbar(x, y, e)
/usr/local/lib/python3.5/dist-packages/matplotlib-2.1.0rc1+104.g479242506-py3.5-linux-x86_64.egg/matplotlib/pyplot.py in errorbar(x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, hold, data, **kwargs)
2850 xlolims=xlolims, xuplims=xuplims,
2851 errorevery=errorevery, capthick=capthick, data=data,
-> 2852 **kwargs)
2853 finally:
2854 ax._hold = washold
/usr/local/lib/python3.5/dist-packages/matplotlib-2.1.0rc1+104.g479242506-py3.5-linux-x86_64.egg/matplotlib/__init__.py in inner(ax, *args, **kwargs)
1708 warnings.warn(msg % (label_namer, func.__name__),
1709 RuntimeWarning, stacklevel=2)
-> 1710 return func(ax, *args, **kwargs)
1711 pre_doc = inner.__doc__
1712 if pre_doc is None:
/usr/local/lib/python3.5/dist-packages/matplotlib-2.1.0rc1+104.g479242506-py3.5-linux-x86_64.egg/matplotlib/axes/_axes.py in errorbar(self, x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, **kwargs)
3082
3083 if yerr is not None:
-> 3084 lower, upper = extract_err(yerr, y)
3085 # select points without upper/lower limits in y and
3086 # draw normal errorbars for these points
/usr/local/lib/python3.5/dist-packages/matplotlib-2.1.0rc1+104.g479242506-py3.5-linux-x86_64.egg/matplotlib/axes/_axes.py in extract_err(err, data)
3022 fe = safe_first_element(err)
3023 if (len(err) != len(data)
-> 3024 or isinstance(fe, Sized) and len(fe) > 1):
3025 raise ValueError("err must be [ scalar | N, Nx1 "
3026 "or 2xN array-like ]")
TypeError: len() of unsized object
Expected outcome
An errorbar plot with the first plot point errorbar being invisible, similar to what is produced when any other errorbar point is masked. The problem seems to stem from the fact that cbook.safe_first_element returns a np.ma.masked value, which for some reason is considered a Sized object: isinstance(np.ma.masked, Sized) returns True. A possible fix would be to prevent cbook.safe_first_element from returning masked values.
Matplotlib version
- Operating System: Ubuntu 17.04
- Matplotlib Version: 2.1.0rc1+104.g479242506
- Python Version: 3.5
Metadata
Metadata
Assignees
Labels
No labels