holoviews.streams.update is not embedded into HTML on save

In the first cell I do this:

%%opts Scatter3D [width=800 height=800 color_index=2] (cmap='Viridis')
import holoviews as hv
import numpy as np
from holoviews.plotting import plotly
hv.notebook_extension(plotly=True)
from holoviews.streams import PositionX
xs,ys = np.mgrid[-50:51, -50:51] * 0.07
N = 100
random_x = np.random.randn(N)
random_y = np.random.randn(N)
random_z = np.random.randn(N)
stream = PositionX()
hv.DynamicMap(lambda x: hv.Scatter3D((random_x, random_y, random_z*x)), kdims=[], streams=[stream])

And at the next cell I put this:

stream.update(x=100)

Then I’m saving the resulted plot as HTML - but getting the plot before the update and not after the update.

Is it a bug in Plotly - or it’s by design and there is no way to fix it? Thanks!

[jlstevens]

I suspect this is because streams are currently integrated with the matplotlib and bokeh backends but not the plotly backend. I’ll assign this issue to the 1.7 milestone as we need to investigate how plotly will work with streams before then.

[philippjfr]

I did the cast but the plot disappeared…

You need to cast it after you’ve displayed it once.

I meant expanding the cache size to be greater than 1 so you can capture a sequence of updates via streams.

The default cache size is 500 anyway, the example above use streams so by default the cache is fixed at length 1. A user would have to explicitly declare key dimensions for the streams for the cache to work properly.