Managing your Streamlit dependencies using conda

Community Cloud
1

Unofficially streamlit was supporting conda dependencies via the conda.txt file and custom conda channels via the conda_channles.txt file. If you were using this feature you should migrate to the new supported environment.yml configuration file.

Example of migrating from conda.txt and conda_channles.txt to environment.yml:

conda.txt

numpy==1.18.1
scikit-learn==0.22.2.post1

conda_channles.txt

conda-forge

The two above files should be converted to one environment.yml file:

environment.yml

channles:
  - conda-forge
dependencies:
  - numpy==1.18.1
  - scikit-learn==0.22.2.post1

For more information about the environment.yml please consult the conda documentation.

4 Likes
3

Thanks for your solution, now my app is working! :v:

1 Like
4

The ability to download from conda-forge opens so many new possibilities :heart: will test ASAP!

ah and welcome to the community @gmerticariu :slight_smile:

5

I have as well dependencies issue but I’m using requirements.txt file

this is what the file contain:

streamlit~=0.62.0
pytube~=10.7.2
pathlib2~=2.3.5
os-sys~=2.1.4

image
image1920×1080 283 KB

6

Try deleting and recreating the app. This should fix your issue.

8

What is the error message in the log?

11

I am having the same problem when deploying an app with environment.yml and get the exact same error: SpecNotFound: Invalid name, try the format: user/package

Any solution to this?