Templates overwrite colorscale property of traces.

[SaschaJust]

Hi,

first of all, thank you for introducing template support to plotly.

As far as I understand, template settings should only "replace" default values but never overwrite explicit properties of a plot. However, when using colorscale in heatmaps, the colorscale is always overwritten by the template, no matter if it's specified by alias or as a list.

Consider the following example:

import plotly.graph_objs as go
import plotly.offline as py
from plotly import __version__ as V
py.init_notebook_mode(connected=True)

print(V)

trace = go.Heatmap(
    x = list(range(1,10)),
    y = list(range(1,10)),
    z = list(range(1,10)),
    colorscale = [[0, 'rgb(255,0,0)'], [1, 'rgb(0,255,0)']]
)

layout = go.Layout(
    title = 'Colorscale templating test',
    template = 'ggplot2'
)

fig = go.Figure([trace], layout)

py.iplot(fig)

In this case, the ggplot2 theme's colorscale will be used rather the one explicitly specified in the trace.

[jonmmease]

Hi @SaschaJust ,

Glad you're enjoying templates and thanks for the report. It looks like the issue is that the templates set autocolorscale to True which overrides the explicit colorscale. It will take a little digging to figure out whether the issue is with the plotly.py template or with the plotly.js template engine.

In the meantime, you can set autocolorscale to False in your heatmap definition

import plotly.graph_objs as go
import plotly.offline as py
from plotly import __version__ as V
py.init_notebook_mode(connected=True)

print(V)

trace = go.Heatmap(
    x = list(range(1,10)),
    y = list(range(1,10)),
    z = list(range(1,10)),
    colorscale = 'Viridis',
    autocolorscale = False
)

layout = go.Layout(
    title = 'Colorscale templating test',
    template = 'ggplot2'
)

fig = go.Figure([trace], layout)

py.iplot(fig)

[SaschaJust]

Hi @jonmmease,

thanks for getting back to me so quickly. That workaround does just fine.

[SaschaJust]

Just to confirm: tested with the 3.9.0 rc and works as expected.
Thanks!