Polars timezone not respected in graph_objects.scatter

[kevinleahy-switchdin]

latest version of plotly // python 3.12

from datetime import datetime, timedelta
from zoneinfo import ZoneInfo

import pandas as pd
import polars as pl

daterange = [datetime(2023, 1, 1, tzinfo=ZoneInfo("Australia/Brisbane")) + timedelta(hours=i) for i in range(48)]

df_1 = pl.DataFrame(
    {
        "timestamp": daterange,
        "line1": [0.1] * 24 + [0.2] * 24,
    }
)
df_2 = pl.DataFrame(
    {
        "timestamp": daterange,
        "line2": [0.0] * 24 + [0.4] * 24,
    }
)
fig = px.line(df_1, x="timestamp", y="line1")

fig.add_trace(
    go.Scatter(
        x=df_2["timestamp"],
        y=df_2["line2"],
        name="line2",
    )
)
fig.show()

"line1" (using px.line )renders correctly with local time on the X-axis. But "line2" (using go.Scatter) gets converted to UTC.

This goes away if df_2 uses pandas or I pass in df_2['timestamp'].to_list() instead.

Probably related to #5166 - is this just because polars isn't supported, but since we've not got type hints in plotly my IDE wasn't complaining?

[MarcoGorelli]

thanks for the report - yeah iirc going through plotly directly (as opposed to plotly.express) will just convert to numpy, so it would get an array of timestamps at the corresponding utc times?