The reasons I switched from poetry to uv

poetry is an amazing Python packaging and dependency management tool. I used it for every tool I wrote in private and at work. For writing tools I am also using a cookiecutter template.

Those are the reasons I switched.

Speeeeeed

Although I never had any performance issues with poetry, getting something faster is always a good thing. However, it was not the main factor in the decision to switch since initializing projects is probably not the most critical thing. Still, uv is noticeably a lot faster then poetry.

Python version management

Being able to install and switch Python versions on the fly is a very useful feature.

E.g. to configure your project with another Python 3.12:

uv sync --python 3.12

Or to run a standalone script with a certain version:

uv run --python 3.10 --no-project something.py

Additionally you will not need another tool or containerization to run tests for more than one version.

For instance, uv integrates seamless with nox:

import nox

PYTHON_VERSIONS = ["3.11", "3.12", "3.13"]

nox.options.default_venv_backend = "uv"


@nox.session(python=PYTHON_VERSIONS)
def tests(session):
    session.run_install(
        "uv",
        "sync",
        env={"UV_PROJECT_ENVIRONMENT": session.virtualenv.location},
    )
    session.run(
        "pytest",
        *session.posargs,
    )

Prototyping

Running a Python script for testing something is a very common task. uv simplifies doing that by being able to run a script with certain dependencies from the CLI:

uv run --with requests api_call.py

Alternatively, dependencies can be specified as inline metadata:

# /// script
# dependencies = [
#   "requests<3",
#   "rich",
# ]
# ///

uv has an interface for handling those dependencies for a script for you using:

uv add --script api_call.py 'requests<3' 'rich'

pyproject.toml syntax

While that might be opinionated, uv is a bit more in sync with the Python Packaging User Guide regarding how the pyproject.toml should look like.