Changing scale type on existing figure

I’m building a multipurpose dashboard that should be able to graph a wide variety of data with different settings. One of the requirements is being able to graph against milliseconds while another is to graph against datetimes. I currently have code similar to the following:

x_sc = DateScale()
y_sc = LinearScale()

ax_x = Axis(label='time', scale=x_sc, grid_lines='solid', tick_format='%m/%d/%y', tick_rotate=90)
ax_y = Axis(label='y_val', scale=y_sc, orientation='vertical', tick_format='0.2f')
pz = PanZoom(scales={'x': [x_sc], 'y': [y_sc]})
data = Lines(x=x_data, y=y_data, scales={'x': x_sc, 'y': y_sc})

zoom_interacts = widgets.ToggleButtons(options=OrderedDict([
                                        (' ', None),
                                        ('xy ', pz)]),
                                        icons = ["stop", "arrows"]
                                        )



fig = Figure(axes=[ax_x, ax_y], marks=[data])
link((zoom_interacts, 'value'), (fig, 'interaction'))
widgets.HBox([fig, zoom_interacts])

swap = widgets.Select(
    options=['dates', 'ms'],
    description='Swap',
    disabled=False
)
def swapAxes(b):
    global x_data
    global ax_x
    global data
    global fig
    if(b == 'dates'):
        ax_x.scale = DateScale()
        x_data = date_data
    else:
        ax_x.scale = LinearScale()
        x_data = ms_data
 
    
interactive(swapAxes, b=swap)

where y_data, date_data, and ms_data are all numpy arrays. When the figure first renders it works as long as I’m using ms_data with a linear scale or date_data with a date scale. But when I try to change the scale using the select the values on the x_axis simply disappear. Is there a way to change this scale back and forth between a date scale and a linear scale?

[shorstman]

That worked great! Thank you so much for the help!

[DougRzz]