Polar and polar subplot (make_subplots) not displaying correctly in FigureWidget Jupyter Ipython notebook

[openafox]

plotly==4.1.1 jupyter==1.0.0 ipython==7.8.0

Plotting a polar in a subplot works fine (#758 resolved in 4.1.1?) but wrapping a polar figure in a FigureWidget and using display causes theta values to become very strange, scattering the data. This happens for subplots and solo polar plots. This would appear to be a Bug. Any thoughts on a workaround?

import plotly
import plotly.graph_objs as go
import numpy as np
from plotly.subplots import make_subplots

# Data
# the first subplot: a histogram (If using)
trace1 = go.Histogram(
    x=np.random.uniform(1,6,size=62),
)

# the second subplot: a polar graph
trace2 = go.Scatterpolar(
    r = np.random.uniform(1,6,size=62),
    theta = np.random.uniform(30,5,size=62),
    mode='markers',
)

# # # # # # # # # # # # # # # # # # 
# single polar example
# # # # # # # # # # # # # # # # # # 
fig = go.Figure(
    trace2, 
    layout=go.Layout(
        title=go.layout.Title(text="Polar chart - normal fig.show()")
    )
)

fig.show()

# Now wrap in a figure widget and use same traces
# not working correctly

fig.update_layout(go.Layout(
    title=go.layout.Title(text="Polar chart - FigureWidget")
    )
)
fig2 = go.FigureWidget(
    fig,
)
display(fig2)  # allows traces to be added but displays wrong

print("FigureWidget.show()")
fig2.show()  # This will not be updateable with data - just shows base fig

# # # # # # # # # # # # # # # # # # 
# Subplots example
# # # # # # # # # # # # # # # # # # 
# normal use of make_subplots

# make 2 plots sublplot
fig = make_subplots(rows=1, cols=2, specs=[[{}, {'type': 'polar'}]],)
fig.update_layout(go.Layout(
    title=go.layout.Title(text="Polar chart - normal fig.show()")
    ))

fig.append_trace(trace1, 1, 1)

fig.append_trace(trace2, 1, 2)

fig.show()

# Now wrap in a figure widget and use same traces
# not working correctly                
fig2 = go.FigureWidget(make_subplots(rows=1, cols=2, specs=[[{'type': 'bar'}, {'type': 'polar'}]]))

fig2.update_layout(go.Layout(
    title=go.layout.Title(text="Polar chart - FigureWidget")
    ))

display(fig2)  # allows traces to be added but the polar one displays wrong
print("FigureWidget.show()")
fig2.show()  # This will not be updateable with data - just shows base fig

# the first subplot: a histogram
fig2.append_trace(trace1, 1, 1)

# the second subplot: a polar graph
fig2.append_trace(trace2, 1, 2)

print("FigureWidget.show()")
fig2.show() # This fig show will work and display correct data

[openafox]

Temporary Workaround: A bit more digging has shown that the theta values are being plotted as Radians even though Degrees are default. Setting thetaunit = "degrees" does not work but thetaunit = "radians" and converting np.degreees() values beforehand does work.

[zouhairm]

confirming this is still an issue in plotly==5.1.0

[zouhairm]

Some further investigation seems to indicate that this is not just an issue of radians/degrees, but rather of array types:

Code below produces different traces depending on whether theta is given as a list or numpy array!

f = go.FigureWidget()

f.add_scatterpolar(r=[10., 15., 15.], theta=[  0., -45.,  45.], thetaunit='degrees', mode='markers')
f.add_scatterpolar(r=[10., 15., 15.], theta=np.array([  0., -45.,  45.]), thetaunit='degrees', mode='markers')

f