Grouped Bars not rendering correctly

[philippjfr]

Noticed this in hvplot documentation and also on the latest dev build:

http://dev.holoviews.org/reference/elements/bokeh/Bars.html

[hoxbro]

This is a Bokeh problem with ObjectNDArray.

# This is mostly code from webpage.
import holoviews as hv
import numpy as np
import pandas as pd

hv.extension("bokeh")
samples = 100

pets = ["Dog"]
genders = ["Female", "Male"]

pets_sample = np.random.choice(pets, samples)
gender_sample = np.random.choice(genders, samples)
count = np.random.randint(1, 5, size=samples)
df = pd.DataFrame({"Pets": pets_sample, "Gender": gender_sample, "Count": count})
bars = hv.Bars(df, kdims=["Pets", "Gender"]).aggregate(function=np.sum)

Converting the code to Bokeh,

from bokeh.io import show
from bokeh.models import Row

bp_org = hv.render(bars)
bp_fix = hv.render(bars)
data = bp_fix.renderers[0].data_source.data
data["xoffsets"] = data["xoffsets"].tolist()

show(Row(bp_org, bp_fix))

[nicolaskruchten]

Is the same thing happening for box plots or is that a different issue? https://holoviews.org/gallery/demos/bokeh/boxplot_chart.html today looks like this, whereas I would expect the boxes to be non-overlapping:

[hoxbro]

This is the same issue. If you need to use this functionality right now, you can use the following hook:

def dtype_fix_hook(plot, element):
    renderers = plot.handles["plot"].renderers
    for renderer in renderers:
        data = renderer.data_source.data
        for k, v in data.items():
            if hasattr(v, "dtype") and v.dtype.kind == "U":
                data[k] = v.tolist()

boxwhisker.opts(hooks=[dtype_fix_hook])

[hoxbro]

I will close the issue as we have a workaround in Holoviews.

https://holoviews.org/reference/elements/bokeh/Bars.html

https://holoviews.org/gallery/demos/bokeh/boxplot_chart.html