environment.yml instead of requirements.txt?

[serenalotreck]

I'm using dygiepp within a project I'm working on, and I was wondering if there's a reason why there's a requirements.txt file installed with pip, instead of a environment.yml file that installs packages with conda, since a conda environment is built. I've been having issues where when I try and install new packages with conda that I use in other code in the project, and I get dependency conflicts that pip's dependency resolver can't fix. I'm not totally sure if having the original environment be built with a conda environment.yml would fix that, but I've always understood that using pip in a conda environment is something you shouldn't so unless you can't avoid it. Wondering what your thoughts are?

[dwadden]

I think you're right, pip within conda is discouraged. I'd definitely accept a PR that relies solely on conda using an environment.yml file. Alternatively, you can just force pip to install your packge with the --no-deps flag, which is the go-to strategy to ensure high-quality, reproducible research 😅 .