Sourced from astral-sh/setup-uv's releases.
v5.2.1 🌈 Support toml spec 1.0.0
v5.2.0 introduced TOML parsing using
@iarna/toml
because we already found out inastral-sh/ruff-action
that toml has missing features.As it turns out
@iarna/toml
also is not fully TOML spec (1.0.0) compliant.We now use smol-toml
🐛 Bug fixes
- Support toml spec 1.0.0
@eifinger
(#245)v5.0.0 🎄 Merry Christmas - Help fastly and users by default
Changes
This christmans 🎄 release is a bit early bit still full of presents 🎁 Since we are changing some of the defaults this can lead to breaking changes, thus the major version increase.
Here are the highlights:
Default to enable-cache: true on GitHub hosted runners
Did you know that that Fastly, the company hosting PyPI, theoretically has to pay $12.5 million per month and so far have served more than 2.41 exabytes of data?
This is why they asked us to turn on caching by default. After weighting the pros and cons we decided to automatically upload the cache to the GitHub Actions cache when running on GitHub hosted runners. You can still disable that with
enable-cache: false
.I remember when I first got into actions and didn't understand all the magic. I was baffled that some actions did something behind the scenes to make everything faster. I hope with this change we help a lot of users who are don't want to or are afraid to understand what
enable-cache
does.Add **/requirements*.txt to default cache-dependency-glob
If caching is enabled we automatically searched for a
uv.lock
file and when this changed we knew we had to refresh the cache. A lot of projects don't use this but rather the good oldrequirements.txt
. We now automatically search for bothuv.lock
andrequirements*.txt
(this means alsorequirements-test.txt
,requirements-dev.txt
, ...) files. You can change this withcache-dependency-glob
Auto activate venv when python-version is set
Some workflows install packages on the fly. This automatically works when using a python version that is already present on the runner. But if uv installs the version, e.g. because it is a free-threaded version or an old one, it is a standalone-build and installing packages "into the system" is not possible.
We now automatically create a new virtual environment with
uv venv
and activate it for the rest of the workflow ifpython-version
is used. This means you can now do- name: Install uv uses: astral-sh/setup-uv@auto-environment with: python-version: 3.13t - run: uv pip install -i https://pypi.anaconda.org/scientific-python-nightly-wheels/simple cython
🚨 Breaking changes
- Default to enable-cache: true on GitHub hosted runners
@eifinger
(#193)
... (truncated)
b5f58b2
Support toml spec 1.0.0 (#245)4e3dbec
Add venv/bin as absolute path to PATH (#241)2487ffc
Bump @octokit/rest
from 21.0.2 to 21.1.0 (#229)118b721
chore: update known checksums for 0.5.20 (#238)d942048
chore: update known checksums for 0.5.19 (#237)77cc1ae
Bump typescript from 5.7.2 to 5.7.3 (#230)169ed2a
Bump @types/node
from 22.10.5 to 22.10.6 (#236)9fffe05
Add documentation for new inputs uv-file and pyproject-file (#235)5ce9ee0
Detect required-version from config file (#233)d577e74
chore: update known checksums for 0.5.18 (#232)