Issue with streamlit as dependency when deploying my package to Bioconda repository

I am opening an issue here for a problem that I don’t think you guys can solve, but you never know. I recently developped a Python tool, and I need to get it onto the Bioconda repository . To do this, I have created a recipe that I must know merge into the bioconda-recipes master branch. At this stage, my pull request is open but does not pass the CI phase because the virtual environment gets killed for no apparent reason. This only happens once I add streamlit as a dependency for my tool. Any ideas on why this keeps happening?

Hi @llegregam, welcome to the community! :wave: :partying_face:

It looks like the reason is provided in the error message: Bash exited with code '137'.

Exit code 137 is a Linux error code emitted when the process runs out of memory:

This issue seems unrelated to Streamlit or your PR. It likely has to do with the RAM allocated to the CI containers. I suspect you’d run into the error even if you switched streamlit with tensorflow or torch.

I tried running the machine locally and added a few big dependencies and it worked. It only gets killed when I add streamlit in the mix…

It’s likely a memory issue when resolving dependencies. Please go through related past issues in the bioconda-recipes repo:

For similar cases where the minimal container is insufficient, there seems to be an extended image:

Thank you for the tip. I tried with the extended container and still no luck… It’s a very weird thing that is going on here ^^’