Skip to content

Working with uv

uv is an extremely fast Python package and project manager, written in Rust. We will use uv as the single tool that replaces pip, virtualenv, pyenv, and more. The main tasks for which we will use uv are:

  • run and install Python versions
  • installing and managing a virtual environment
  • build all the packages in the workspace or monorepo
  • publish all the packages to PyPI
  • run scripts and apps

Installing uv

On macOS and Linux you can install uv using curl:

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

If you need more specific information on installing and upgrading uv, please refer to the official documentation.

Installing a Python version

The CGSE is guaranteed to work with Python 3.9.x. We will gradually include higher versions of Python, but currently these have not been tested. So, we will for the moment stick with Python 3.9.20. Install this version as follows:

$ uv python install 3.9.20

pyenv

When you are using pyenv to manage your Python versions, make sure you also have the same Python version installed with pyenv and uv. Otherwise you will run into the following error. This is a known issue with uv.

pyenv: version `3.9.20' is not installed (set by /Users/rik/github/cgse/libs/cgse-common/.python-version)

You can check which Python versions are installed already on your system:

$ uv python list --only-installed

cpython-3.12.8-macos-aarch64-none     /Users/rik/Library/Application Support/uv/python/cpython-3.12.8-macos-aarch64-none/bin/python3.12
cpython-3.10.16-macos-aarch64-none    /Users/rik/Library/Application Support/uv/python/cpython-3.10.16-macos-aarch64-none/bin/python3.10
cpython-3.9.21-macos-aarch64-none     /Users/rik/Library/Application Support/uv/python/cpython-3.9.21-macos-aarch64-none/bin/python3.9
cpython-3.9.20-macos-aarch64-none     /Users/rik/Library/Application Support/uv/python/cpython-3.9.20-macos-aarch64-none/bin/python3.9
cpython-3.9.6-macos-aarch64-none      /Library/Developer/CommandLineTools/usr/bin/python3 -> ../../Library/Frameworks/Python3.framework/Versions/3.9/bin/python3
cpython-3.8.17-macos-aarch64-none     /Users/rik/Library/Application Support/uv/python/cpython-3.8.17-macos-aarch64-none/bin/python3.8

Create a virtual environment

Pin a Python version

You can pin a python version with the command:

$ uv python pin 3.9.20

uv will search for a pinned version in the parent folders up to the root folder or your home directory.

You can create a virtual environment with uv for the specific Python version as follows. The '--python' is optional and uv will use the default (pinned) Python version when creating a venv without this option. We are working in a monorepo or what uv calls a workspace. There will be only one virtual environment at the root of the monorepo, despite the fact that we have several individual packages in our workspace. Don't worry, uv will always use the virtual environment at the root and keep it up-to-date with the project your are currently working in.

When creating a virtual environment make sure you are in the root folder, e.g. ~/github/cgse.

$ cd ~/github/cgse
$ uv venv --python 3.9.20
If you want to name your virtual environment, use the optional argument --prompt <venv name> in the above command, otherwise the virtual environment will get the same name as the project, i.e. cgse.

Now, navigate to the package you will be working in and update the projects' environment, assuming you are going to work in cgse-core, this will be:

$ cd ~/github/cgse/libs/cgse-core
$ uv sync

Your package(s) from the workspace should be installed as an editable install. You can check this with the command:

$ uv pip list -v
Using Python 3.9.20 environment at: /Users/rik/github/cgse/.venv
Package           Version     Editable project location
----------------- ----------- ---------------------------------------
apscheduler       3.11.0
cgse-common       0.4.0       /Users/rik/github/cgse/libs/cgse-common
cgse-core         0.4.0       /Users/rik/github/cgse/libs/cgse-core
...

To install any other project as an editable package:

$ uv pip install -e <project location>

Note

If you don't want to use the uv commands, you can activate the virtual environment and use the original pip and python commands as you are used to, but I would recommend you try to get used to uv for a while to experience its benefits.

$ source ~/github/cgse/.venv/bin/activate

Info

In a workspace, maintaining a virtual environment per package might be a hassle and most of the time that is not needed. A good approach is to always use the virtual environment at the workspace root. This venv which will be automatically created if you run a command or if you use uv sync in the package folder. With uv sync you can make sure the virtual environment is up-to-date and contains only those dependencies that are required for the package you are in. So, each time you switch to another package and want to run a comand or a test for that package, use 

$ uv sync

Building and publishing all packages

We have chosen for one and the same version number for all packages in the cgse monorepo. That means that whenever we make a change to one of the packages and want to release that change, all packages shall be rebuild and published.

Inline

When working in a workspace, keep in mind that the commands uv run and uv sync by default work on the workspace root. That means that when you run the uv run pip install <package> command, the .venv at the workspace root will be updated or created if it didn't exist. Similar for the uv sync command, there is only one uv.lock file at the root of the workspace.

Fortunately, with uv, that is done in a few commands.

When you are in the monorepo root folder, you can build all packages at once. They will be placed in the dist folder of the root package. Before building, make sure you update the version in the pyproject.toml of the root package and then bump the versions. Before building, clean up the dist folder, then you can do a default uv publish afterwards.

$ cd <monorepo root>
$ uv run bump.py
$ rm -r dist
$ uv build --all-packages

Publish all packages in the root dist folder to PyPI. The UV_PUBLISH_TOKEN can be defined in a (read protected) ~/. setenv.bash file:

$ uv publish --token $UV_PUBLISH_TOKEN

The above command will publish all package to PyPI. If you don't want the token to be in a shell variable, you can omit the --token in the command above. You will then be asked for a username, use __token__ as the username and then provide the token as a password.