A python package and project manager written in Rust
A drop in replacement for pip, pip-tools, virtualenv
- (python) Virtual environment using pip and venv
- uv VS pip virtualenv
  - pip only handles package management and virtualenv handles only environment creation → uv combines both functionalities in a single tool
  - uv → often 10~100x faster, uses significantly less memory, better conflict resolution
- Virtual environment
What it does
- Manages pyproject.toml, lockfiles (uv.lock), and project-specific virtual environments automatically.
- Downloads and installs Python interpreters on demand
- Installs and runs Python CLI tools in isolated environments
- Uses global cache

mac install & usage

brew install uv 
uv version

uv init explore-uv
cd explore-uv
tree -a
.
├── .gitignore
├── .python-version
├── README.md
├── hello.py
└── pyproject.toml

git is automatically initialized and main git-related files like .gitignore + empty README.md are generated
.python-version→ contains the Python version used for the proj
pyproject.toml → the main configuration file for project metadata and dependencies
- if u wanna change the python version, check if version is ok in pyproject.toml, then go .python-version & change, then uv sync
- then this creates a new virtual env in the proj directory → uv pip install -e.
hello.py → simple file to help u get started quickly

Adding dependencies

uv add requests
uv add requests=2.1.2

uv combines the environment creation & dependency installation
- the 1st time you run the add command, UV creates a new virtual environment in the current working directory and installs the specified dependencies
- on subsequent runs, UR reuses the existing virtual env and only install/update the newly requested packages
add
- when u add, UV uses a modern dependency resolver that analyzes the entire dependency graph to find a compatible set of package versions that satisfy all requirements
- automatically generates + updates uv.lock file
  - records exact versions of all dependencies & sub-dependencies
  - u can maintain requirements.txt too
    - uv export -o requirements.txt → generates a requirements.txt from a uv lock file
- updates the pyproject.toml

uv remove scikit-learn

uv run hello.py

run
- use when u wanna execute single file
- run ensures that script is executed in the virtual env UV created

`pip` to `uv` migration

pip freeze > requirements.txt
uv init.
uv pip install -r requirements.txt

caching
- pip → caches downloaded wheel files + source distributions, but unpacks and copies the actual package files into each individual virtual environment (consuming more disk space and time)
- uv→ uses global cache mechanism - if u install the same version of a package in multiple virtual envs on the same machine, uv links to the cached version rather than copying the files over and over
tool consolidation
- pip & virtualenv → multistep → 1st make virtualenv then use pip to manage packages
- uv → single unified tool → use the uv CLI to create envs and manage packages