Differences between Plotly and Dash (texttemplate, insidetextanchor)

Dash Python
1

I’ve noticed that a Plotly figure embedded in Dash does not honor all parameters as if it were rendered as a pure Plotly figure. For example, the below code returns different plots between the two platforms

note x is a pandas Series object of floats, y is a pandas Series object of strings (labels). This is essentially a budget plot displaying percent of budget expended.

fig = go.Figure(
        
        data=[go.Bar(x=x, y=y,
                    name='Discretionary Spending',
                    orientation='h',
                    marker_color=(
                        ['#FF6666' if a > 1 else '#00CC66' for a in x]
                    ),
                    text=x, 
                    textposition='auto',
                    texttemplate='%{x:.0%}',
                    textfont=dict(size=10),
                    insidetextanchor='start',
                    )],
        
        layout=go.Layout(
            title=go.layout.Title(text="Discretionary Spending"),
            xaxis=dict(showgrid=False, 
                    visible=False),
            paper_bgcolor='white',
            plot_bgcolor='white',
            height=650
        )
    )

When run as a python script, the plot looks like this (displayed in my browser):

image

When embedded in a dash dcc.Graph, it looks like this:

image

Any thoughts on why parameters like insidetextanchor and texttemplate are not honored in the Dash environment?

2

it’s likely a version difference between the underlying plotly.js library- plotly (the python library) and dash both contain their own version of plotly.js. you can try upgrading and/or looking at the CHANGELOG files to see exactly version of plotly.js you are using in each library.

3

It was really that simple–updated Dash to 1.8.0 and Plotly to 1.4.5 and everything is looking the same. Thanks for the tip!

Really been enjoying playing around with Dash and excited to explore what it can do. Thanks for all your work on it!

1 Like