How do you conda install a library in an environment created with virtualenv?
Question:
I’m working on a (python) project where the choice was to create a virtual environment using virtualenv. However, one of the project dependencies can’t be installed through pip on macOS due to this bug: https://github.com/streamlit/streamlit/issues/283
The workaround is to conda install one of the dependencies to bypass the gcc compiler.
How do you conda install something in a virtual environment not created with conda?
Answers:
I think the easiest approach would be to create a conda env by it’s own.
1) Create a requirement.txt file by doing pip freeze > requirements.txt inside your virtualenv environment
2) Create conda env: conda create --name myenv
3) Activate your environment: source activate myenv
4) Install your dependencies: conda install --file requirements.txt
5) Install missing dependecy: conda install YOUR_MISSING_DEPENDENCY
In the accepted answer (upvoted) you can also change point 1) to use conda-installed packages (compatible with subsequent conda install, and excluding pip-installed packages that would be unavailable in conda channels, identified by "pypi" in their extended version names that only conda displays):
conda list --export | grep -v pypi > requirements.txt
And if you still want to use pip, the correct syntax that gets you packages versions list in a format compatible with pip install is now:
pip list --format=freeze > requirements.txt
I’m working on a (python) project where the choice was to create a virtual environment using virtualenv. However, one of the project dependencies can’t be installed through pip on macOS due to this bug: https://github.com/streamlit/streamlit/issues/283
The workaround is to conda install one of the dependencies to bypass the gcc compiler.
How do you conda install something in a virtual environment not created with conda?
I think the easiest approach would be to create a conda env by it’s own.
1) Create a requirement.txt file by doing pip freeze > requirements.txt inside your virtualenv environment
2) Create conda env: conda create --name myenv
3) Activate your environment: source activate myenv
4) Install your dependencies: conda install --file requirements.txt
5) Install missing dependecy: conda install YOUR_MISSING_DEPENDENCY
In the accepted answer (upvoted) you can also change point 1) to use conda-installed packages (compatible with subsequent conda install, and excluding pip-installed packages that would be unavailable in conda channels, identified by "pypi" in their extended version names that only conda displays):
conda list --export | grep -v pypi > requirements.txt
And if you still want to use pip, the correct syntax that gets you packages versions list in a format compatible with pip install is now:
pip list --format=freeze > requirements.txt