从 linux 到 windows 的 Conda 环境

Question

I am working on an image processing / computer vision / deep learning app which requires the use of old versions for some Python packages. I am new to the world of machine learning and software,it's been 6 months that I made a career change and I am not aware of the best practices when it comes to python projects that work for different platforms (Windows, Linux, MacOS). So today, I am working on a linux, I did some conda installs and some pip installs and I am exporting my environment to a.yml file. How to make sure that creating a virtual environment on a Windows Machine using conda and my exported.yml file is going to work with no problems?

Answer 1

first you create the virtual environment eg

pipenv shell

then you write a batch script checking for python packages in that environment:

echo %time% Check if python packages are installed...
echo ------------------------------------------------------------------------
pip install -r [your .yml file]
echo ------------------------------------------------------------------------

Answer 2

You just have to make sure your .yml file doesn't include platform dependent packages by adding the --from-history flag when creating the file. This exports only packages you intentionally installed.

On Windows conda will take care of the virtual environment by installing dependencies for the specified packages in the .yml file that are Windows compatible.