Metpy StationPlot does not like dates

[njmessina]

Howdy...
I was trying to make a single station time series of sky condition using StationPlot. I get an error ('invalid type promotion') when I send a Pandas date index for the x-axis values. I emailed with Ryan and advised me to use date2num, which works flawlessly. He mentioned I should send some code for a possible bug fix.
Here's some sample code:

import numpy as np
import dateutil
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.ticker import FuncFormatter
import matplotlib.dates as mdates
import matplotlib.ticker as ticker
import matplotlib.dates as dates

from metpy.calc import reduce_point_density
from metpy.calc import wind_components
from metpy.cbook import get_test_data
from metpy.plots import add_metpy_logo, current_weather, sky_cover, StationPlot, wx_code_map
from metpy.units import units


rng = pd.date_range('12/1/2017', periods=5, freq='D')
sc= [1,2,3,4,5]
ts= pd.Series(sc, index= rng)

fig, ax = plt.subplots()
y=  np.ones(len(ts.index)) 

# This does NOT work:

x=  ts.index    # metpy does not like dates
stationplot = StationPlot(ax, x , y, fontsize=12)
stationplot.plot_symbol('C', ts , sky_cover)

# This works! :
#x= dates.date2num(ts.index)   # works!  ... Thank you Ryan!
#stationplot = StationPlot(ax, x , y, fontsize=12)
#stationplot.plot_symbol('C', ts , sky_cover)

ax.xaxis.set_major_locator(mdates.DayLocator())   
ax.xaxis.set_major_formatter(mdates.DateFormatter('%-d'))

plt.show()

 

Thank you!!

Joe

[zbruick]

@dopplershift, how do you imagine this being handled? Is there a way to force any array to work here? Not sure if any others would give errors if we just handled the pandas case here. At any rate, the issue is within plot_text for this case, but the type conversion could likely be handled in the init function.