Plot forgets the order of categorical values

[swenlaur]

If you draw a plot with a discrete x-scale and the first geom does not use all x-values and the second geom introduces new values that should be in front of old ones they are added to the end. Example code snippet

VarNames = CategoricalDtype(['x_{}'.format(i) for i in range(10)], ordered = True)
df1 = DataFrame({'x': ['x_1', 'x_2']}).assign(x = lambda df: df['x'].astype(VarNames))
df2 = DataFrame({'x': ['x_0', 'x_2']}).assign(x = lambda df: df['x'].astype(VarNames))

p = ggplot(df1) + geom_point(aes(x =  'x', y = 'x'), size=6) 
p = p + geom_point(aes(x =  'x', y = 'x'), data=df2, color = 'red', size = 3)
p 

[TyberiusPrime]

obviously, this is a bug.
Quick workaround for now, specify the limits on the x-scale:

p = p + scale_x_discrete(limits=VarNames.categories)