Plotly paper formatting is not uniform

Question

I am using the annotation system in Plotly to annotate the axis of a graph. I am specifying coordinates using the "paper" setting in plotly. However, I have noticed that paper coordinates are not always uniform across a figure, at least when arrows are turned off. Does anyone know why this might be happening? Here I provide code to show you:

def test():
    
    fig = go.Figure()

    ylabels = [0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
    annotations = []
    for i, label in enumerate(ylabels):
        print(label/100)
        annotations.append(
            dict(
                yref = 'paper',
                y = label/100, x = 0,
                text = ylabels[1],
                showarrow = False,
                yshift = 0,
            )
        )
    fig.update_layout(
        annotations = annotations,
        yaxis = dict(
            showticklabels = False,
            tickvals = ylabels,
            range = [0, 100]
        ),
        xaxis = dict(
            showticklabels = False
        )
    )
    fig.show()

Non-uniform spacing when arrows are off

For some reason, this has to do with turning off the arrows. When the arrows are on, they are evenly spaced across the x-axis.

Uniform spacing when arrows are on

Answer 1

The issue seems to be resolved by setting yanchor equal to either top, middle, or bottom, see the Plotly documentation.

import plotly.graph_objects as go

fig = go.Figure()

ylabels = [0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100]

annotations = []
for i, label in enumerate(ylabels):
    annotations.append(
        dict(
            yref='paper',
            y=label / 100,
            x=0,
            text=ylabels[1],
            showarrow=False,
            yshift=0,
            yanchor='middle' # default is 'auto'
        )
    )
    
fig.update_layout(
    annotations=annotations,
    yaxis=dict(
        showticklabels=False,
        tickvals=ylabels,
        range=[0, 100]
    ),
    xaxis=dict(
        showticklabels=False
    )
)

fig.show()