Dash scattermap browser display freezes with js errror

📊 Plotly Python
1

I have an app that overlays lines onto of a graph with a go.scattermap fig of satellite map. It has sliders controlling line spacing and position.

This all works fine for the first few adjustments, then the browser freezes with a js error. The figure in Dash is updating in response to additional inputs (verified by fig.show() in the debugger), but the browser display is frozen.

MNG screengrab
MNG screengrab1920×961 129 KB

The js console in the browser shows a Map error

Uncaught (in promise) Error: Map error.
    at t.Map.r (plotly.min.js:8:718596)
    at t.Map.fire (plotly.min.js:8:3714964)
    at de.fire (plotly.min.js:8:3715109)
    at de.removeLayer (plotly.min.js:8:4203279)
    at t.Map.removeLayer (plotly.min.js:8:4490265)
    at l.dispose (plotly.min.js:8:1359475)
    at _.updateData (plotly.min.js:8:718055)
    at plotly.min.js:8:716504

Edit: Solved, but not understood

Referencing a global color_lookup table when setting the color parameter of a trace triggers the js error after 3-10 callbacks. The first two calls always work.

I do not understand why, but empirically this change avoids the problem

There is a global color lookup dictionary:

lk_line_color = build_line_color_dict()

The update_graph callback rebuilds the figure containg traces like this one:

setting the color with color=lk_line_color[‘pac’] eventually fails after 3-10 calllbacks.

CASE 1 Works all the time: (color via discrete color)

fig.add_trace(go.Scattermap(lat=xyp_pac[:, 0],
lon=xyp_pac[:, 1],
mode=‘lines+markers’,
name=f’Pacific line {line_no}',
marker=dict(color=‘blue’, size=2),
line=dict(color=‘blue’) ) ) )

CASE 2 Fails after 3-10 calllbacks: (color via lookup_table)

fig.add_trace(go.Scattermap(lat=xyp_pac[:, 0],
lon=xyp_pac[:, 1],
mode=‘lines+markers’,
name=f’Pacific line {line_no}',
marker=dict(color=lk_line_color[‘pac’], size=2),
line=dict(color=lk_line_color[‘pac’]),
visible=‘legendonly’ ) )

Perhaps keeping the lookup table in the dcc.Store would avoid the error.

Previous problem description:
In reading I see in previous versions this was related to version mismatches. My current versions are:

dash==2.18.2
dash-bootstrap-components==1.6.0
dash-core-components==2.0.0
dash-html-components==2.0.0
dash-leaflet==1.0.15
dash-table==5.0.0

I am unclear whether I have conflicting callbacks or a javascript problem.

The callbacks are shown below. The settings_dict is stored in a dcc.Store with id=settings_store.

I hope this is enough. Constructing a full MWE will be daunting.

comment: updates fig and updated [connector_dict in settings_store  ('data')

@callback(
    [Output('graph', 'figure', allow_duplicate=True),
     Output('settings_store', 'data')],

    [Input('line_no', 'value'),
     Input('width_km', 'value'),
     Input('clip_range', 'value'),
     Input('shift', 'value'),
     Input('settings_store', 'data')],
    prevent_initial_call=True
)
def update_graph_line(line_no, width_km, clip_range, shift, settings_dict):
    if settings_dict is None:
        # fill settings_dict with defaults:   # width_km = 10, clip = [0, 100], shift = 0
        settings_dict = {str(n): [10, [0, 100], 0] for n in range(1, 23)}

    if ctx.triggered_id == 'line_no':
        width_km, clip_range, shift = settings_dict[str(line_no)]
    else:
        settings_dict[str(line_no)] = width_km, clip_range, shift

    fig, result_line = plot_line(line_no,
                                 line_dict,
                                 spacing=width_km,
                                 min_pct=clip_range[0],
                                 max_pct=clip_range[1],
                                 shift=shift
                                 )

    return fig, settings_dict


comment: on line change: restore slider settings or set defaults if settings_dict is None

@callback(
    [
        Output('graph', 'figure', allow_duplicate=True),
        Output('width_km', 'value', allow_duplicate=True),
        Output('clip_range', 'value', allow_duplicate=True),
        Output('shift', 'value', allow_duplicate=True),
        Output('settings_store', 'data', allow_duplicate=True)],

    [Input('line_no', 'value'),
     Input('settings_store', 'data')],
    prevent_initial_call=True
)
def specify_settings(line_no, settings_dict):
    if ctx.triggered_id == 'line_no':               # only update if line_no change
        if settings_dict is None:
            # fill settings_dict with defaults:     # width_km = 10, clip = [0, 100], shift = 0
            settings_dict = {str(n): [10, [0, 100], 0] for n in range(1, 23)}

        width_km, clip_range, shift = settings_dict[str(line_no)]

        fig, result_line = plot_line(line_no,
                                     line_dict,                 # global line coordinate data dictionary
                                     spacing=width_km,
                                     min_pct=clip_range[0],
                                     max_pct=clip_range[1],
                                     shift=shift
                                     )

        return fig, width_km, clip_range, shift, settings_dict

    return no_update

Thanks for any insight.