Hover not working if set before applying linked selections

[MarcSkovMadsen]

HoloViews: 1.14.0, bokeh: 2.2.3, 0.10.2

I'm trying to create an app for awesome-panel.org showcasing linked brushing but have difficulties getting hover to work.

Here is my example

import holoviews as hv
import panel as pn
from holoviews import opts
from plotly.data import iris

hv.extension("bokeh")
pn.config.sizing_mode="stretch_width"

IRIS_DATASET = iris()

def _get_linked_plots():
    dataset = hv.Dataset(IRIS_DATASET)

    scatter = hv.Scatter(dataset, kdims=["sepal_length"], vdims=["sepal_width"]).opts(tools=["hover"], size=10)
    hist = hv.operation.histogram(dataset, dimension="petal_width", normed=False).opts(
        tools=["hover"]
    )

    # pylint: disable=no-value-for-parameter
    selection_linker = hv.selection.link_selections.instance()
    scatter = selection_linker(scatter)
    hist = selection_linker(hist)

    scatter=pn.pane.HoloViews(scatter, sizing_mode="stretch_both")
    hist=pn.pane.HoloViews(hist, sizing_mode="stretch_both")

    return pn.Column(scatter, hist, height=600)

_get_linked_plots().servable()

[MarcSkovMadsen]

Additional Context: Works if you hover opts after linking plots.

I can see that it works if I set the tools opts after then plots have been linked. See below. I spent a long time figuring this out. It's confusing and frustrating as some opts like color and active_tools work if you set them before setting the linked selections. And it's not documented anywhere (I believe). So your guessing here.

import holoviews as hv
import panel as pn
from holoviews import opts
from plotly.data import iris

hv.extension("bokeh")
pn.config.sizing_mode="stretch_width"

IRIS_DATASET = iris()

def _get_linked_plots():
    dataset = hv.Dataset(IRIS_DATASET)

    scatter = hv.Scatter(dataset, kdims=["sepal_length"], vdims=["sepal_width"]).opts(size=10)
    hist = hv.operation.histogram(dataset, dimension="petal_width", normed=False)

    # pylint: disable=no-value-for-parameter
    selection_linker = hv.selection.link_selections.instance()
    scatter = selection_linker(scatter).opts(opts.Scatter(tools=["hover"]))
    hist = selection_linker(hist).opts(opts.Histogram(tools=["hover"]))

    scatter=pn.pane.HoloViews(scatter, sizing_mode="stretch_both")
    hist=pn.pane.HoloViews(hist, sizing_mode="stretch_both")

    return pn.Column(scatter, hist, height=600)

_get_linked_plots().servable()

[jbednar]

I agree that's confusing; I would have expected the options for the new object created by selection_linker to inherit the options from the original object.