use geom_segment when scale x is datetime

[gaorongchao]

if the scale x is not datetime,it work

from plotnine import * 
from plotnine.data import *
import datetime

p = ggplot(economics,aes(x='pop',y='pce'))
(p + geom_line()
  + geom_segment(aes(x=240000,xend=240000,y=0,yend=12500)))

The flow doesnot work

from mizani.breaks import date_breaks
from mizani.formatters import date_format

p = ggplot(economics,aes(x='date',y='pce'))
(p + geom_line()
   + geom_segment(aes(x=datetime.datetime.strptime('1970-01-01','%Y-%m-%d'),xend=datetime.datetime.strptime('1970-01-01','%Y-%m-%d'),y=0,yend=12500)) 
   + scale_x_datetime(breaks=date_breaks('10 years'),labels=date_format('%Y')))

[has2k1]

Yes, that is a bug. For a work around, enclose x and xend in pd.Series

i.e

geom_segment(aes(x=pd.Series([datetime.datetime.strptime('1970-01-01','%Y-%m-%d')]), xend=pd.Series([datetime.datetime.strptime('1970-01-01','%Y-%m-%d')]), y=0, yend=12500))

Sorry to ask, but what are you doing to run into all these edge cases?

[gaorongchao]

I don't think this is an edage cases,when I use this in my work, something like financial，Time Series is the most common thing. And always need to drawn the average rate,income in the pic. So if other people know and want use plotnine I think they will encounter these cases. now I think plotnine is not as famous as seaborn or matplotlib, so I encounter these cases first. I hope plotnine will be better than ggplot in R, come on !