Can you use multiple backends at the same time?

HoloViews
,
1

Is it possible to use two backends within the same HoloViews application or do all the elements have to use the same backend?

I would like to use bokeh for some views but plotly (or some other backend) for others. Is there an easy way to do this?

2 Likes
2

Hoping to bump this issue and see if there is a way to do this. I’m working on a dashboard where most plots are generated with the bokeh backend but some other plots need to be generated with Plotly. I’m wondering if I can mix backends like this.

I’ve tried loading both extensions and passing the backend="..." option without much luck. Hoping to do something like

import holoviews as hv
hv.extension('bokeh', 'plotly')

# Plot with bokeh
hv.Points(df, kdims=['x', 'y'], vdims=['v']).opts(color=hv.dim('v'), backend='bokeh')

# Plot with plotly
hv.Points(df, kdims=['x', 'y'], vdims=['v']).opts(color=hv.dim('v'), backend='plotly')

But no matter what, both plots are the same backend and it seems the backend just controls where the opts are being captured for later static rendering which I think is outlined in the documentation here.

I’m curious if there is a way to have one plot with bokeh and one plot with plotly in the same dashboard

3

Maybe something like this

import holoviews as hv
import pandas as pd
import panel as pn
hv.extension('bokeh', 'matplotlib')

df = pd.DataFrame({'x': [1, 2, 3], 'y': [1, 2, 3], 'v': [1, 2, 3]})

# Plot with bokeh
pn.Row(
    hv.render(hv.Points(df, kdims=['x', 'y'], vdims=['v']).opts(color=hv.dim('v')), backend='bokeh'),
    hv.render(hv.Points(df, kdims=['x', 'y'], vdims=['v']).opts(color=hv.dim('v')), backend='matplotlib')
).show()

image
image1448×956 36.2 KB

1 Like
4

Thanks @ahuang11! That appears to do what I need!

5

Hi @banesullivan-kb

The HoloViews pane also supports a backend parameter. The code below will produce the same.

import holoviews as hv
import pandas as pd
import panel as pn

pn.extension()
hv.extension("bokeh", "matplotlib")

df = pd.DataFrame({"x": [1, 2, 3], "y": [1, 2, 3], "v": [1, 2, 3]})

# Plot with bokeh
pn.Row(
    pn.pane.HoloViews(
        hv.Points(df, kdims=["x", "y"], vdims=["v"]).opts(color=hv.dim("v")),
        backend="bokeh",
    ),
    pn.pane.HoloViews(
        hv.Points(df, kdims=["x", "y"], vdims=["v"]).opts(color=hv.dim("v")),
        backend="matplotlib",
    ),
).servable()
1 Like