Pip Installing Dependencies Into Conda Envs

I prefer to use Conda because [once installed](https://docs.conda.io/en/latest/miniconda.html) it works essentially the same everywhere. Including Windows. ## Step 1: Create a Virtual Environment ```bash conda create -n credible python=3.10 ``` After agreeing to install some dependencies, it places them in an out-of-the way location. Here's where they are on my work, personal laptops, and a 2018-era Windows laptop I sometimes test on: - `/Users/drg/opt/miniconda3/envs/` - `/Users/danny/opt/miniconda3/envs/` - `c:\Users\danielandaudrey\.conda\envs` Get the list of current envs: ```bash $ conda env list base * /Users/drg/opt/miniconda3 credible /Users/drg/opt/miniconda3/envs/credible system /Users/drg/opt/miniconda3/envs/system that /Users/drg/opt/miniconda3/envs/that just /Users/drg/opt/miniconda3/envs/just works /Users/drg/opt/miniconda3/envs/works everywhere /Users/drg/opt/miniconda3/envs/everywhere ``` ## Step 2: Activate a Virtual Environment ```bash conda activate credible ``` This prefixes the CLI shell with `(credible)`, exactly like `venv` or `virtualenv`. ```bash (credible) ``` ## Step 3: Install Dependencies Using Pip Into a Conda Env Just like many any other virtual environments for Python, use pip to install/uninstall dependencies ```bash (credible) python -m pip install typer ``` That's right. I'm using pip to manage packages in a conda env. Conda works great with pip. Conda has worked great with pip for years. ## Questions Because I know you have them. ### 1. What about conda install for dependencies? I rarely use `conda install` on MacOS or Linux. On Windows, `conda install` is sometimes useful because it handles non-python binaries that otherwise require Docker, [Chocolately](https://chocolatey.org/), or other tools. ### 2. Why not pyenv-win instead of conda? Reasons: 1. I always struggled to get pyenv-win to work on Windows 2. Didn't handle non-python binaries well, requiring use of Chocalatey 3. Conda just works ### 3. Did you ever teach classes with Conda? Yes. Conda was awesome when I used to teach. Conda is a universal installation system that works for everyone regardless of operating system. Once people got it installed we got to teaching instead of dealing with setup issues. ### 4. What about Docker instead of Conda? My classes focused more on teaching Python and Django rather than Docker. Docker is a whole different world. ### 5. What does Audrey use? [Audrey](https://audrey.feldroy.com) uses pyenv and loves it. Taught people to use pyenv and pyenv-win. Updates: - 2022-03-12 [Suggested by David R. Pugh](https://twitter.com/TheSandyCoder/status/1502577229607415808): Changed to use `python -m pip` to ensure the dependency is installed on the right place.  _[Source of Image](https://rna.colostate.edu/2020/lib/exe/detail.php?id=wiki%3Asoftwareinstall&media=wiki:python-conda.png)_