Closed
Description
I am testing matplotlib 2.2rc from a point of view of "what if pandas no longer registers its unit converters", and one of the issues raised when we temporarily did that for the 0.21.0 release was the following (pandas-dev/pandas#18283):
Code for reproduction
import pandas as pd
import matplotlib.pyplot as plt
pd.plotting.deregister_matplotlib_converters()
xticks = pd.date_range(start="10/10/2017", end="11/11/2017",
freq='D')
plt.xticks(xticks, xticks.strftime("%a %m-%d"))Actual outcome
The above raises TypeError: Cannot compare type 'Timestamp' with type 'float', because Timestamps are not recognized to use the DateConverter in the units framework.
Would it be possible to consider adding pandas Timestamps to the unit registry? It are subclasses of datetime.datetime, so code-wise it should not be hard to add support.
Matplotlib version
- Operating system: Linux
- Matplotlib version: '2.2.0rc1'
- Python version: 3.5