Followup to https://github.com/pydata/pandas/pull/8307
In [1]: df = pd.DataFrame(np.random.rand(4, 4))
In [2]: df2 = df.copy()
In [3]: for i in range(len(df)):
...: df.iloc[i, i] = np.nan
...:
In [4]: fig, axes = plt.subplots(figsize=(16, 8), nrows=2, ncols=4)
In [5]: df.plot(kind='pie', subplots=True, ax=axes.ravel()[:4])
In [6]: df2.plot(kind='pie', subplots=True, ax=axes.ravel()[4:])
Right now, when subplots=True, the NaN/zero labels aren't being put in the legend (or they aren't showing up).
In [8]: df[0].plot(kind='pie', figsize=(4,4), legend=True)
Not a huge priority, we'll see if anyone notices and wants to implement a fix. I think it would involve refactoring all our legend and ticklabels code to be subplot aware.
Comment From: TomAugspurger
Not clear what needs to be done here.