Issue deploying incompatibility pandas and numpy

Deployment
,
1

Hello,

I try to deploy a small web app.
Nothing complicated apparently, since only pandas and streamlit were in my requirements at first.

  1. First, I had issues between latest verison of numpy and pandas.
    So I added in my requirements an older version of numpy. it worked well until I tried to deploy the app
  2. Deployment issue : when I try to deploy the app, it always fails.
    I tried to change my requirements :
    a) No numpy but then incompatibility beteen numpy and pandas (issue 1)
    b) Older numpy but impossible deployment (conflict maybe between the old numpy and the numpy streamlit wants to install with pandas ?)
    c) Older version of streamlit, pandas but did not work either

I tried to solve the issue with some posts I have read but nothing worked
For instance, I have read that I should install pandas like this :
pip install --no-cache-dir --no-binary :all: pandas
But did not work.
I also tried in my requirements.txt to not fix the version of streamlit, someone wrote that it worked by just writing “streamlit” in requirements.txt. Did not work either.

My Git Hub repository is :

Thanks in advance for your help !

2
  1. As of today, there should be no incompatibility between latest numpy and pandas and as far as I can tell they work well along with your own code. You may be misinterpreting whatever issues you are having.
  2. Old versions of packages may not work with latest python. Python 3.12 is just about one year old; the pandas and numpy in your requirements are older than that.
3

Hello, Thanks a lot for your feedback. I do not understand then what should I do :melting_face:. Do I need to install the latest versions of pandas and python to have no more incompatibility issues between pandas and numpy ?

4

That is what I would recommend, unless you have a very good reason to do otherwise.

2 Likes
5

It seems to work indeed. Thanks a lot !