Python dev setup

Python development on OSX/Linux

uv

These days (2025) I use uv.

brew install uv
cd $proj_dir
uv init myapp && cd myapp
uv python list
uv python install 3.12.11
uv python pin 3.12.11        # update .python_version
uv run main.py
uv add pytest
uv lock --upgrade-package pytest  # uv lock -U  to update all deps
uv sync  # to update .venv
uv export --format requirements-txt > requirements.txt

uv run pip freeze --local

pyenv + poetry

(updated: oct 2024)

brew install pipx
pipx install poetry
cd $new_project
pyenv local 3.12.7
poetry init
poetry env show
poetry add fastapi uvicorn
poetry add pytest pytest-cov --group test

pyenv + buildin virtualenv

pyenv install 3.12.1
pyenv local 3.12.1
python -m venv venv
. ./venv/bin/activate
pip install --upgrade pip
pip install fastapi
pip install 'uvicorn[standard]'
pip freeze > requirements.txt

Python development on Windows

(updated: dec 2022)

See win dev setup if you need to setup your box first.

Proxy setup for pip

Assuming the userID=1234566 and password=s3cr3t0, type the following in the terminal:

$env:http_proxy="http://123456:s3cr3t0@proxytr.company.com:8080"
$env:https_proxy="http://123456:s3cr3t0@proxytr.company.com:8080"

(NB: http:// for $env:https IS correct)

Always work inside an isolated environment

It is very hard to manage different versions of dependencies for multiple projects in one place. We need to use virtual environments. In Python 3, virtualenv was added to core library as venv module.

cd new_project
python -m venv .\myevn
.\myenv\Scripts\activate
pip install requests  # installs into .\myenv\Lib\site-packages

For ML projects, it might be better to use [mini]conda as a replacement for pip:

scoop install miniconda3
conda install -n root -c pscondaenvs pscondaenvs
conda init powershell

# relaunch terminal

conda info         # check current env
conda list         # which packages are available
conda info --envs  # list of existing environments
conda create --name myenv
conda activate myenv
conda install -n myenv numpy scikit-learn
conda deactivate

# Add conda forge
conda config --set auto_update_conda False
conda config --add channels conda-forge
conda search psycopg2

conda install --file requirements.txt

Useful packages for maintaining code quality

• pytest: most popular testing framework (including test discovery and test runner)
• pytest-xdist pytest-cov: add parallel test runner and test coverage reporting
• flake8: linting and PEP 8 style conformance
• isort: make imports more consistent
• black: uncompromising Python code formatter