Also, installing the PyPI package in Google Colab would be much faster than the conda package (through condacolab).
Hope this helps.
Update: I found one other advantage; since the opensim dll and exe files are being copied into the opensim package folder using this method, there won't be any interference with other dependencies like casadi and ezc3d. So, all packages can be installed and used in a same Python environment.
Hi, I think opensim-core can be published on PyPI with minimal effort, for those who have to work with official Python and pip.
Here, I created a workflow in GitHub Actions (derived from yours) that clones opensim-core, builds it for Windows (Python 3.12), creates a wheel file, and uploads it (available in artifacts): https://github.com/mrrezaie/testBuildOpenSim/blob/42d9596d642ee87f06c89eee0a4130cd123efd3a/.github/workflows/main.yml#L100-L102
The wheel file can be installed easily using
pip install zzz.whl
, or be published on PyPI for any Python users: https://medium.com/@blackary/publishing-a-python-package-from-github-to-pypi-in-2024-a6fb8635d45dAlso, installing the PyPI package in Google Colab would be much faster than the conda package (through condacolab).
Hope this helps.
Update: I found one other advantage; since the opensim dll and exe files are being copied into the opensim package folder using this method, there won't be any interference with other dependencies like casadi and ezc3d. So, all packages can be installed and used in a same Python environment.