Reset axes button in dcc.Graph modebar doesn't apply the range specified in yaxis_range

If the figure is included from the start in a dcc.Graph, clicking Reset axes 🏠 in the modebar triggers this:

{'xaxis.autorange': True, 'xaxis.showspikes': False, 'yaxis.range': 0, 0.1, 'yaxis.showspikes': False}

However, if the dcc.Graph figure property is populated in a callback, Reset Axes 🏠 triggers this (the same effect as clicking Autoscale):

{'xaxis.autorange': True, 'xaxis.showspikes': False, 'yaxis.autorange': True, 'yaxis.showspikes': False} 

even if the figure has a specific yaxis_range=[0,0.1] and we have set yaxis_autorange=False

I believe this is because the modebar is created when the dcc.Graph component is initialized and it’s not updated if the figure changes.

Additional information:

• Clicking Reset Axes 🏠 triggers this event: https://github.com/plotly/plotly.js/blob/master/src/components/modebar/buttons.js#L215
• I’ve (unsuccessfully) tried forcing a modebar update by:
  • Making some change to the modebar when I create fig2 in the callback
  • Modifying the config property of the Graph at the same time I return the figure

Workaround

Create an empty figure with the desired/eventual layout, for example:

empty_fig = go.Figure(layout=go.Layout(yaxis_range=[0, 0.1])) 

and assign this as the initial value for the figure that you are going to later update with a callback:

dcc.Graph(id='fig2', figure=empty_fig)

Code to replicate the issue

Env
Python 3.10
dash==2.18.1
plotly==5.24.1

from dash import Dash, html, dcc, callback, Input, Output
import pandas as pd
import plotly.graph_objects as go
import random
import plotly
import dash

print(f"Plotly version is {plotly.__version__} and dash version is {dash.__version__}")

def generate_figs():

    N = 50

    well_df = pd.DataFrame(
        {
            "EventTime": [i for i in range(N)],
            "cmax": [random.randint(0, 100) for i in range(N)],
        }
    )

    well_names = [f"Well{i}" for i in range(N)]

    threshold = 1000
    max_y = 0.07

    fig = go.Figure()

    for well_name in well_names:

        fig.add_trace(
            go.Scatter(
                x=well_df["EventTime"],
                y=well_df["cmax"] / threshold,
                name=well_name,
                mode="lines",
            )
        )

    fig.update_layout(
        title="Combined anomaly detection",
        xaxis=dict(title=dict(text="Time")),
        yaxis=dict(range=[0, 0.1], showticklabels=True, autorange=False),
        plot_bgcolor="#ebfaeb",
        legend=dict(y=1.08, x=0.97),
    )

    # create a copy of the figure
    fig2 = go.Figure(fig)

    # your figure
    fig.add_hrect(y0=1.0, y1=max_y, line_width=0, fillcolor="red", opacity=0.2)
    # fig2
    fig2.update_layout(plot_bgcolor="rgba(255,0,0,0.2)")  # red with alpha 0.2
    fig2.add_hrect(y0=0, y1=max_y, line_width=0, fillcolor="#ebfaeb", layer="below")
    return fig, fig2

fig, fig2 = generate_figs()

# WORKAROUND
empty_fig = go.Figure(layout=go.Layout(yaxis_range=[0, 0.1]))

@callback(Output("fig2", "figure"), Input("btn", "n_clicks"), prevent_initial_call=True)
def show_fig(n):
    return fig2

app = Dash(__name__)

app.layout = html.Div(
    [
        html.Button("Click me", id="btn"),
        dcc.Graph(id="fig2"),  # add figure=empty_fig for workaround
        dcc.Markdown(id="out2"),
        dcc.Markdown(id="out2b"),
        dcc.Graph(id="fig2b", figure=fig2),
    ]
)

# callbacks to check the effects on the figure
@callback(
    Output("out2", "children"), Input("fig2", "relayoutData"), prevent_initial_call=True
)
def showchanges2(rl):
    return f"**Relayout:**{rl}"

@callback(
    Output("out2b", "children"),
    Input("fig2b", "relayoutData"),
    prevent_initial_call=True,
)
def showchanges2b(rl):
    return f"**Relayout:**{rl}"

if __name__ == "__main__":
    app.run_server(debug=True)

Reproduce el caso de un dcc.Graph rellenado mediante un callback usando el ejemplo de Dash proporcionado e inspecciona src/components/modebar/buttons.js en los handlers de Reset Axes enlazados. Compara los relayoutData emitidos para una figura configurada inicialmente con los de una figura rellenada mediante un callback. Se considera terminado cuando Reset Axes conserva el yaxis_range especificado en lugar de emitir yaxis.autorange: true.