update() produces 'invalid type promotion' error when used with datetime columns and overwrite=False #9065
Description
xref #3025
To reproduce:
import pandas as pd
df = pd.DataFrame({'A': [1, 2, 3],
'B': ['20140101', nan, '20140101'],
'C': [nan, nan, nan]})
df2 = pd.DataFrame({'A': [1, 2],
'B': ['20140104', '20140105'],
})
df.B = pd.to_datetime(df.B, coerce=True)
df2.B = pd.to_datetime(df2.B, coerce=True)
df.set_index('A', inplace=True)
df2.set_index('A', inplace=True)
#df.update(df2[['B']]) # Works fine
df.update(df2[['B']], overwrite=False) # Fails with 'invalid type promotion' error.