Holoviews.operation.datashader.spread() support for streams?

I’d like to implement the spread(datashade()) example from the Large Data example in the Holoviews User Guide; it’s the third plot here:

However my source data comes from an hv.DynamicMap which returns gv.Points from an hv.streams.Buffer like this:

template_df   = pd.DataFrame({'x': [], 'y': []}, columns=['x', 'y'])
dfstream      = hv.streams.Buffer(template_df, index=False, length=99999999, following=False)
points        = hv.DynamicMap(gv.Points, streams=[dfstream])

I’ve imported like this:

from holoviews.operation.datashader import datashade, spread

And while datashade(points) works as expected, using spread(datashade(points)) as per the example in the user guide yields a ZeroDivisionError: float division by zero error. I think this is because the stream is empty upon initialization and only receives data after user interaction, but as I said, datashade is able to handle this but spread is not.

I have Holoviews version 1.14.0

Can spread be made to work on streaming data that is empty upon initialization?

Could you file an issue on the HoloViews repo?

Will do. Thanks.

[Update:] …actually looks like it was a problem with my code. When I wrote a minimal example for the issue on GitHub it worked as I expected.