Why is it important?

Sometimes you really need to run some old code. Maybe it is a obscure Fortran 77 program, written by an advisor of your advisor. Maybe it is a Python script from another PhD student who left a few years ago, which uses particular combination of packages no longer working in Python 3.10. With some luck there might be a README file, but you never know for sure how accurate it is.

Needless to say, how painful it could be.

So, what can I do?

Time-traveling still works in only one direction. There is never enough time to write the documentation.  However, not all hope is lost: you can always try to make things easier for your future self (or for another poor soul).

The idea of this session is to overview & compare the tools, which can help with creating the environments for your projects and reconstruct the same environments in the future.

Which tools?

• conda / mamba  or any language-specific package manager
• environment modules or spack
• containerization:  Docker, Charliecloud or Apptainer
• functional package managment: Nix and Guix
• something else?

How to contribute to this session?

• Tell a story how you were trying to run some very old code
• Do you pack your stuff in Docker? You can show the Dockerfile and explain what's inside.
• Do you compile some MPI codes on a few different clusters and also on your PC and laptop? (where there is MPICH insted of IMPI for example) How do you archive it?
• Share your tricks to make python scripts do not break with every python update :)

I (Iliya) have some experience with conda, Docker and Guix, so I can talk about all these things. Something similar (but not limited to) to this.