Holoviews: Error bars not working for categorical axis with bokeh

[bstadlbauer]

ALL software version info

holoviews = 1.13.3
python = 3.6.10
bokeh = 2.2.1

on MacOs 10.14.6

Description of expected behavior and the observed behavior

Expected: Error bars are shown for categorical values

Error bars are not shown, issue should already be resolved:
https://github.com/holoviz/holoviews/issues/1470

Complete, minimal, self-contained example code that reproduces the issue

%%opts ErrorBars (line_width=2 line_color='darkblue')
%%opts Points    [invert_axes=False] (color='darkred' size=6)
import holoviews as hv
import numpy as np

x=['a','b','c','d','e']
points = [(x[i], 5*np.random.rand()) for i in range(5)]
errors = [points[i]+(np.random.rand()/2,) for i in range(5)]

h1 = hv.Points(points)
h2 = hv.ErrorBars(errors)
h2*h1

Screenshots or screencasts of the bug in action

[jlstevens]

Based on the issue you linked to, this definitely sounds like a regression (since it looks like it was fixed in 2017!)

[kubekmonika]

Hello, I have just stambled upon this error. Is there any workaround I can use before it is fixed?

[lgonzalezsa]

Hola,
Seems to be related to categorical data in X axis; was facing the issue today and at the time I changed the X axis to numerical, the error bars start working. Need to show the categorical values in the x axis; any suggestion for workaround more than welcome.

Look the new definition of pointsvariable.

%%opts ErrorBars (line_width=2 line_color='darkblue')
%%opts Points    [invert_axes=False] (color='darkred' size=6)

x=['a','b','c','d','e']
points = [(i,5*np.random.rand()) for i in range(5)]
#[(x[i], 5*np.random.rand()) for i in range(5)]
errors = [points[i]+(np.random.rand()/2,) for i in range(5)]

h1 = hv.Points(points)
h2 = hv.ErrorBars(errors)
h2*h1