Plotly Express Timeline discrete color feature bug?

I'm using plotly.express.timeline to plot... Timelines. Let me illustrate:

display(plot_df)
# Note that start and end time dates are not used

Data

fig1 = px.timeline(plot_df, x_start='start_time', x_end='end_time', y='start_date', color='category')
fig1.show()

Timeline plot 1

So far so good. The only problem is that the category based colormap is continuous, while in reality category is categorical. So I change the category column type to string and try again:

plot_df['category'] = plot_df['category'].astype(str)
fig2 = px.timeline(plot_df, x_start='start_time', x_end='end_time', y='start_date', color='category')
fig2.show()

Timeline plot 2

Obviously something went wrong... But what?

PS: Depending on the input data, I get different kinds of strange results when converting the category column to string. Another example:

plot_df2 = plot_df[plot_df['start_date'] > pd.to_datetime('2019-05-24')]
fig3 = px.timeline(plot_df2, x_start='start_time', x_end='end_time', y='start_date', color='category')
fig3.show()

Timeline plot 3

I also tried visualizing the plots in a different environment, with similar results.

I can't explain exactly why this is the case, but the first graph assumes the category column is a number and displays the default colour map, while the second graph assumes the category column is a string and therefore the x and y axes are also time series data. The reason for this is that clicking on category 1 in the legend hides the blue graph. Similarly, clicking on category 5 in the legend hides the red graph. So only category 8 is shown. I made my decision by looking at the scale of the y-axis. In summary, what you want to achieve is to change the category column to a string and then set the y-axis type to category.I have also added a setting for any color you wish to specify.

import pandas as pd
import numpy as np
import io

data = '''
end_time start_date start_time category
"2023-03-08 10:00:00" 2019-05-24 "2023-03-08 09:00:00" 1
"2023-03-08 09:00:00" 2019-05-24 "2023-03-08 08:00:00" 1
"2023-03-08 13:00:00" 2019-05-24 "2023-03-08 10:00:00" 5
"2023-03-08 10:00:00" 2019-05-27 "2023-03-08 07:00:00" 1
"2023-03-08 15:00:00" 2019-05-27 "2023-03-08 10:00:00" 1
"2023-03-08 14:00:00" 2019-05-28 "2023-03-08 07:00:00" 8
"2023-03-08 15:00:00" 2019-05-28 "2023-03-08 14:00:00" 5 
'''

plot_df = pd.read_csv(io.StringIO(data), delim_whitespace=True)

plot_df['end_time'] = pd.to_datetime(plot_df['end_time'])
plot_df['start_time'] = pd.to_datetime(plot_df['start_time'])

import plotly.express as px
plot_df['category'] = plot_df['category'].astype(str)

fig1 = px.timeline(plot_df, x_start='start_time', x_end='end_time', y='start_date', color='category', 
                   color_discrete_sequence=["red", "green", "blue"])
fig1.update_yaxes(type='category')
fig1.show()

• second graph(category:8 display)