astral-uv / notes

Install:

curl -LsSf https://astral.sh/uv/install.sh | sh
uv --version

Python

uv python list
uv python install 3.12

By default uv python install will verify that a managed Python version is installed or install the latest version.

However, a project may include a .python-version file specifying a default Python version. If present, uv will install the Python version listed in the file.

Project Init

uv init uv_test
which python 
> /home/moses/.pyenv/shims/python
uv python pin 3.12
> Pinned `.python-version` to `3.12`
uv run which python
> Using Python 3.12.5
> Creating virtualenv at: .venv
> /home/moses/Repos/uv_test/.venv/bin/python
> uv.lock also created with this step
uv run python --version
> 3.12.5

The lockfile is created and updated during uv invocations that use the project environment, i.e., uv sync and uv run. The lockfile may also be explicitly updated using uv lock.

Installing Packages

uv add pandas datadog

if you just run python those packages are not there. uv run hello.py works, while just python hello.py doesn’t.

If you do uv venv ; uv run hello.py it installs the packages then runs it.

After this, when you do source .venv/bin/activate your packages are not available (e.g. python hello.py doesn’t work).

When using uv, the virtual environment does not need to be activated. uv will find a virtual environment (named .venv) in the working directory or any parent directories.

Steps for Project:

1. uv python install so it install the python version specified in .python-version
2. uv run hello.py , before running creates .venv and installs specified packages or run uv sync for standalone.

This doesn’t work:

uv sync
source .venv/bin/activate
flask run -p 3000
python example.py