scatter plots with datetime x-axis and more than 1000 data points do not render all points

[cprice404]

I believe this issue is similar to plotly/plotly_express#145 .

I am using plotly 5.15.0 in jupyter lab 3.6.3.

My browsers (have tried firefox and chrome) support webgl.

When I render a scatter plot where the x-axis values are datetime objects,

if the plot has 1000 points, it renders fine:

size = 1000
x = np.random.randint(1690332167190, 1690332667190, size=size)
y = np.random.randint(0, 10000, size=size)
df = pd.DataFrame()
df['x'] = pd.to_datetime(x, unit='ms')
df['y'] = y
fig = px.scatter(df, x='x', y='y')
fig

but if I change size to 1001, the render only shows a few clusters of values on the x-axis:

the hover/mouseover behavior makes it clear that there are a bunch of points that exist in the spaces between what is plotted on the graph:

but the individual points don't render (or are rendered with the wrong x-value, not sure which).

[cprice404]

i presume this has to do with the switch from svg to webgl after 1000 data points, and if I add render_mode=svg, it works properly, but that isn't an ideal solution for graphs where we will have much more than 1000 data points.

[cprice404]

it has something to do with large numeric values on the x-axis because if I normalize the timestamps against the epoch by subtracting 1690330000000 from the x-values:

# x = np.random.randint(1690332167190, 1690332667190, size=size)
x = np.random.randint(2167190, 2667190, size=size)

then it renders fine with 1001 points.

[qchenevier]

It seems to be fixed with the release of v5.19.0: https://github.com/plotly/plotly.py/releases/tag/v5.19.0

[Coding-with-Adam]

I ran the code and it works for me as well. Closing issue.