Bar chart with timeseries index not displaying depending on proximity of the times

Found an issue causing a bar chart to display nothing when displaying timeseries data. It seems to be caused by having stacked data at two points which are very close in time (e.g. 12:00:00 and 12:00:01) and then one point much further away (e.g. 17:00:00).

import pandas as pd
import plotly.express as px

times = ['2020-12-18 02:15:00', '2020-12-18 02:15:00', '2020-12-18 02:15:00', '2020-12-18 02:15:01']  
#times = ['2020-12-18 02:15:00', '2020-12-18 02:15:00', '2020-12-18 02:15:00', '2020-12-18 02:15:00'] <-- works
vals = [10, 50, 20, 350, 250]

# Uncomment to cause bar chart to display no data
# times.append('2020-12-18 13:00:00')
# vals.append(1)
  
df = pd.DataFrame(list(zip(times, vals)), 
               columns =['DT', 'Val']) 

df['DT'] = pd.to_datetime(df['DT'])

df = df.set_index('DT')
fig = px.bar(df, x=df.index, y=df['Val'])

fig.show()

• The code unchanged displays the figure fine
  
• If you uncomment the two append lines, adding in a data point which is further away, then the figure is empty
  
• If you uncomment the two append lines and changes the initial times list so that they are all :00, the figure displays fine
  

Therefore it seems to be caused by having points at 02:15:00 / 02:15:01 and then a point much later at 13:00:00

Thanks

首先运行该 issue 的 pandas 和 Plotly Express 复现，先使用相近的时间戳，然后添加 13:00:00 这一数据点。在两种情况下比较生成的 figure 数据或渲染结果；当柱状图能够在失败的情况下显示所提供的时间序列数据，同时不破坏正常工作的示例时，即视为完成。