Strange behaviour of st.dataframe at rendering datetime variables on Heroku

Hi,

I’m using the Heroku to deploy my code. On my local machine everything is fine. But on Heroku, the st.dataframe (also st.write) does show a strange behavior rendering the datetime column of my dataframe. It starts the dates from one day before the first date in the dataset. Interesting that when I use the df.iloc[0] the result is correct. Check the “Data Negócio” column at below image:
Screenshot from 2020-08-15 17-08-22

1 Like

Shot in the dark here, but could this be a timezone issue? (Is the Heroku instance in a different timezone from your personal computer? Might there be some dependence there?)

Thank you for using Streamlit! :balloon:

2 Likes

Thanks very much for your attention.
In fact, I was suspect of it. For this reason I checked both st.dataframe(df) and st.dataframe(df.iloc[0]) together, as in the image on my question. As you can see, the results are different. So it seems that the Heroku isn’t responsible for this issue. Maybe st.dataframe treats a Dataframe and a Series in a different way.