Dash() add new parameter hide_all_callbacks

[CNFeffery]

For the scenario corresponding to #3493, add a new parameter hide_all_callbacks to Dash(), which is used to change the default application value of the hidden parameter in app.callback(), improving the convenience of callback debugging for complex applications. Example:

import dash
from dash import html, Input, Output, State

app = dash.Dash(__name__, hide_all_callbacks=True)

app.layout = html.Div(
    [
        html.Div(
            [
                html.Button("Button 1", id="button1"),
                html.Button("Button 2", id="button2"),
                html.Button("Button 3", id="button3"),
            ]
        ),
        html.Div(0, id="output"),
    ],
    style={"padding": 50},
)


@app.callback(
    Output("output", "children"),
    Input("button1", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
)
def button1_update_output(n_clicks, children):
    return children + 1


@app.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button2", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
)
def button2_update_output(n_clicks, children):
    return children + 1


@app.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button3", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
    hidden=False,
)
def button3_update_output(n_clicks, children):
    return children + 1


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

[T4rk1n]

How are those actually insert when using @callback directly? I think we need a better system for those since we want the same behavior for both @app.callback and @callback.

Think you can refactor these config into a context and use the context instead of passing them around?

[CNFeffery]

How are those actually insert when using @callback directly? I think we need a better system for those since we want the same behavior for both @app.callback and @callback.

Think you can refactor these config into a context and use the context instead of passing them around?

Yes, similarly, for prevent_initial_callbacks, the callback functions defined by dash.callback() cannot obtain the setting values from dash.Dash() too.

[CNFeffery]

Taking the following application as an example, both the prevent_initial_callbacks and hide_all_callbacks parameters can now work properly in the callback functions defined by app.callback() and dash.callback():

import dash
from dash import html, Input, Output, State

app = dash.Dash(__name__, hide_all_callbacks=True, prevent_initial_callbacks=True)

app.layout = html.Div(
    [
        html.Div(
            [
                html.Button("Button 1", id="button1"),
                html.Button("Button 2", id="button2"),
                html.Button("Button 3", id="button3"),
            ]
        ),
        html.Div(0, id="output"),
    ],
    style={"padding": 50},
)


@app.callback(
    Output("output", "children"),
    Input("button1", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
)
def button1_update_output(n_clicks, children):
    return children + 1


@dash.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button2", "n_clicks"),
    State("output", "children"),
)
def button2_update_output(n_clicks, children):
    return children + 1


@app.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button3", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
    hidden=False,
)
def button3_update_output(n_clicks, children):
    return children + 1


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

[T4rk1n]

Nice usage of get_app, but this will not work if the app hasn't been instanced yet which we can't guarantee with dash.callback.

We can defer it all and map the configs values in the _setup_server after this line:

dash/dash/dash.py

Line 1648 in a9a173f

self._callback_list.extend(_callback.GLOBAL_CALLBACK_LIST)

[CNFeffery]

After refactoring, the current hide_all_callbacks parameter can take effect on callback functions defined in various types:

• Callback functions defined based on dash.callback() before the app is instantiated
• Callback functions defined based on dash.callback() after the app is instantiated
• Callback functions defined based on app.callback()

Example application:

import dash
from dash import html, Input, Output, State


@dash.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button2", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
)
def button2_update_output(n_clicks, children):
    return children + 1


app = dash.Dash(__name__, hide_all_callbacks=True)

app.layout = html.Div(
    [
        html.Div(
            [
                html.Button("Button 1", id="button1"),
                html.Button("Button 2", id="button2"),
                html.Button("Button 3", id="button3"),
            ]
        ),
        html.Div(0, id="output"),
    ],
    style={"padding": 50},
)


@app.callback(
    Output("output", "children"),
    Input("button1", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
)
def button1_update_output(n_clicks, children):
    return children + 1


@dash.callback(
    Output("output", "children", allow_duplicate=True),
    Input("button3", "n_clicks"),
    State("output", "children"),
    prevent_initial_call=True,
    hidden=False,
)
def button3_update_output(n_clicks, children):
    return children + 1


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