Python Package Magic with `uv`

There’s a lot of tools we use when dealing with or setting up Python projects. The basic builtins include `venv`, `pip`, and so on. You’re probably familiar with them all.

But there’s a whole menagerie of tools which, though helpful, never quite solve all the problems you encounter.

Enter `uv` and its magic: it bills itself as a replacement for:

pip, pip-tools, pipx, poetry, pyenv, twine, virtualenv, and more

source.

And I don’t think they are overselling themselves from my limited experience.

It usually takes awhile to convert me to becoming a tool user, but `uv` just worked out of the box, and was able to do some quick magic that turned me into a believer:

1. I could select the version of Python I wanted. This helped me experiment with pre-release libraries of lesser known machine learning packages such as PaddlePaddle’s 3.0rc1.

   > uv init --python 3.14 test
   Initialized project `test` at `/Users/wolf/Development/test`
   > cd test
   > uv add flask
   Using CPython 3.14.0a5
   Creating virtual environment at: .venv
   Resolved 9 packages in 431ms
         Built markupsafe==3.0.2
   Prepared 7 packages in 6.15s
   Installed 7 packages in 3ms
    + blinker==1.9.0
    + click==8.1.8
    + flask==3.1.0
    + itsdangerous==2.2.0
    + jinja2==3.1.6
    + markupsafe==3.0.2
    + werkzeug==3.1.3

2. I could run python scripts, snippets, or whole commands with throwaway virtual environments. With whatever version of Python I wanted—even the unreleased 3.14 to get me ready for some conversations at PyCon 2025.

   > uvx --python python3.14 pycowsay "Hello, from Python 3.14"
   
     -----------------------
   < Hello, from Python 3.14 >
     -----------------------
      \   ^__^
       \  (oo)\_______
          (__)\       )\/\
              ||----w |
              ||     ||

3. It doesn’t fight the system or make you learn many new things: it has a subcommand that replicates the `pip` interface.

   > uv pip install flask
   Resolved 9 packages in 135ms
   Installed 9 packages in 8ms
    + blinker==1.8.2
    + click==8.1.8
    + flask==3.0.3
    + importlib-metadata==8.5.0
    + itsdangerous==2.2.0
    + jinja2==3.1.6
    + markupsafe==2.1.5
    + werkzeug==3.0.6
    + zipp==3.20.2

4. It is, truly, blazingly fast. And, yes, they of course they use Rust. How else do you go fast these days? And stay cool?

So, give it a whirl yourself and let me know what you think. I guarantee you’ll be surprised, and a little (or a lot!) happier programming with Python!