This fixes an obscure cache collision in Python interpreter queries,
which we believe to be the root cause of CI flakes we've been seeing
where a project environment is invalidated and recreated.
This work follows from the logs in [this CI
run](https://github.com/astral-sh/uv/actions/runs/15934322410/job/44950599993?pr=14326)
which captured one of the flakes with tracing enabled. There, we can see
that the project environment is invalidated because the Python
interpreter in the environment has a different version than expected:
```
DEBUG Checking for Python environment at `.venv`
TRACE Cached interpreter info for Python 3.12.9, skipping probing: .venv/bin/python3
DEBUG The interpreter in the project environment has different version (3.12.9) than it was created with (3.9.21)
```
(this message is updated to reflect #14329)
The flow is roughly:
- We create an environment with 3.12.9
- We query the environment, and cache the interpreter version for
`.venv/bin/python`
- We create an environment for 3.9.12, replacing the existing one
- We query the environment, and read the cached information
The Python cache entries are keyed by the absolute path to the
interpreter, and rely on the modification time (ctime, nsec resolution)
of the canonicalized path to determine if the cache entry should be
invalidated. The key is a hex representation of a u64 sea hasher output
— which is very unlikely to collide.
After an audit of the Python query caching logic, we determined that the
most likely cause of a collision in cache entries is that the
modification times of underlying interpreters are identical. This seems
pretty feasible, especially if the file system does not support
nanosecond precision — though it appears that the GitHub runners do
support it.
The fix here is to include the canonicalized path in the cache key,
which ensures we're looking at the modification time of the _same_
underlying interpreter.
This will "invalidate" all existing interpreter cache entries but that's
not a big deal.
This should also have the effect of reducing cache churn for
interpreters in virtual environments. Now, when you change Python
versions, we won't invalidate the previous cache entry so if you change
_back_ to the old version we can re-use our cached information.
It's a bit speculative, since we don't have a deterministic reproduction
in CI, but this is the strongest candidate given the logs and should
increase correctness regardless.
Closes https://github.com/astral-sh/uv/issues/14160
Closes https://github.com/astral-sh/uv/issues/13744
Closes https://github.com/astral-sh/uv/issues/13745
Once it's confirmed the flakes are resolved, we should revert
- https://github.com/astral-sh/uv/pull/14275
- #13817
We were checking whether a path was an executable in a virtual
environment or the base directory of a virtual environment in multiple
places in the codebase. This PR consolidates this logic into one place.
Closes#13947.
> NOTE: The PRs that were merged into this feature branch have all been
independently reviewed. But it's also useful to see all of the changes
in their final form. I've added comments to significant changes
throughout the PR to aid discussion.
This PR introduces transparent Python version upgrades to uv, allowing
for a smoother experience when upgrading to new patch versions.
Previously, upgrading Python patch versions required manual updates to
each virtual environment. Now, virtual environments can transparently
upgrade to newer patch versions.
Due to significant changes in how uv installs and executes managed
Python executables, this functionality is initially available behind a
`--preview` flag. Once an installation has been made upgradeable through
`--preview`, subsequent operations (like `uv venv -p 3.10` or patch
upgrades) will work without requiring the flag again. This is
accomplished by checking for the existence of a minor version symlink
directory (or junction on Windows).
### Features
* New `uv python upgrade` command to upgrade installed Python versions
to the latest available patch release:
```
# Upgrade specific minor version
uv python upgrade 3.12 --preview
# Upgrade all installed minor versions
uv python upgrade --preview
```
* Transparent upgrades also occur when installing newer patch versions:
```
uv python install 3.10.8 --preview
# Automatically upgrades existing 3.10 environments
uv python install 3.10.18
```
* Support for transparently upgradeable Python `bin` installations via
`--preview` flag
```
uv python install 3.13 --preview
# Automatically upgrades the `bin` installation if there is a newer patch version available
uv python upgrade 3.13 --preview
```
* Virtual environments can still be tied to a patch version if desired
(ignoring patch upgrades):
```
uv venv -p 3.10.8
```
### Implementation
Transparent upgrades are implemented using:
* Minor version symlink directories (Unix) or junctions (Windows)
* On Windows, trampolines simulate paths with junctions
* Symlink directory naming follows Python build standalone format: e.g.,
`cpython-3.10-macos-aarch64-none`
* Upgrades are scoped to the minor version key (as represented in the
naming format: implementation-minor version+variant-os-arch-libc)
* If the context does not provide a patch version request and the
interpreter is from a managed CPython installation, the `Interpreter`
used by `uv python run` will use the full symlink directory executable
path when available, enabling transparently upgradeable environments
created with the `venv` module (`uv run python -m venv`)
New types:
* `PythonMinorVersionLink`: in a sense, the core type for this PR, this
is a representation of a minor version symlink directory (or junction on
Windows) that points to the highest installed managed CPython patch
version for a minor version key.
* `PythonInstallationMinorVersionKey`: provides a view into a
`PythonInstallationKey` that excludes the patch and prerelease. This is
used for grouping installations by minor version key (e.g., to find the
highest available patch installation for that minor version key) and for
minor version directory naming.
### Compatibility
* Supports virtual environments created with:
* `uv venv`
* `uv run python -m venv` (using managed Python that was installed or
upgraded with `--preview`)
* Virtual environments created within these environments
* Existing virtual environments from before these changes continue to
work but aren't transparently upgradeable without being recreated
* Supports both standard Python (`python3.10`) and freethreaded Python
(`python3.10t`)
* Support for transparently upgrades is currently only available for
managed CPython installations
Closes#7287Closes#7325Closes#7892Closes#9031Closes#12977
---------
Co-authored-by: Zanie Blue <contact@zanie.dev>
Surprisingly, we weren't locking during `uv sync` so far, so running `uv
sync` in parallel could cause errors in filesystem races.
I've also added locks to `uv add` and `uv remove` which concurrently
modify `pyproject.toml`. These locks only apply after we determined the
interpreter, so they don't actually prevent computing the same thing
twice when running `uv add` in parallel.
All other subcommands that I checked were already locking (with no claim
to exhaustiveness)
Fixes#12751
# Test Plan
I don't have CI-sized reproducer for this.
```toml
[project]
name = "debug"
version = "0.1.0"
requires-python = ">=3.12"
dependencies = [
"boto3>=1.38.30",
"fastapi>=0.115.12",
"numba>=0.61.2",
"polars>=1.30.0",
"protobuf>=6.31.1",
"pyarrow>=20.0.0",
"pydantic>=2.11.5",
"requests>=2.32.3",
"urllib3>=2.4.0",
"scikit-learn>=1.6.1",
"jupyter>=1.1.1",
]
[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"
```
```
rm -rf .venv && parallel -n0 "uv sync -q" ::: {1..10}
```
This includes some initial work on adding Pyodide support (issue
#12729). It is enough to get
```
uv pip compile -p /path/to/pyodide --extra-index-url file:/path/to/simple-index
```
to work which should already be quite useful.
## Test Plan
* added a unit test for `pyodide_platform`
* integration tested manually with:
```
cargo run pip install \
-p /home/rchatham/Documents/programming/tmp/pyodide-venv-test/.pyodide-xbuildenv-0.29.3/0.27.4/xbuildenv/pyodide-root/dist/python \
--extra-index-url file:/home/rchatham/Documents/programming/tmp/pyodide-venv-test/.pyodide-xbuildenv-0.29.3/0.27.4/xbuildenv/pyodide-root/package_index \
--index-strategy unsafe-best-match --target blah --no-build \
numpy pydantic
```
---------
Co-authored-by: konsti <konstin@mailbox.org>
Co-authored-by: Zanie Blue <contact@zanie.dev>
Rustfmt introduces a lot of formatting changes in the 2024 edition. To
not break everything all at once, we split out the set of formatting
changes compatible with both the 2021 and 2024 edition by first
formatting with the 2024 style, and then again with the currently used
2021 style.
Notable changes are the formatting of derive macro attributes and lines
with overly long strings and adding trailing semicolons after statements
consistently.
When removing a Python interpreter underneath an existing venv, uv
currently shows a not found error:
```
error: Failed to inspect Python interpreter from active virtual environment at `.venv/bin/python3`
Caused by: Python interpreter not found at `/home/konsti/projects/uv/.venv/bin/python3`
```
This is unintuitive, as the file for the Python interpreter does exist,
it is a broken symlink that needs to be replaced with `uv venv`.
I've been encountering those occasionally, and I expect users that
switch between versions a lot will, too, especially when they also use
pyenv or a similar Python manager.
The new error hints at this solution:
```
error: Failed to inspect Python interpreter from active virtual environment at `.venv/bin/python3`
Caused by: Broken symlink at `.venv/bin/python3`, was the underlying Python interpreter removed?
hint: To recreate the virtual environment, run `uv venv`
```
When working on #13025 I noticed this message was lacking versions,
which seems frustrating if you're debugging things.
I refactored the general `matches_interpreter` utilities that were added
in https://github.com/astral-sh/uv/pull/12884 into a more purpose-fit
function that returns an `Option` with the versions if there's a
mismatch.
In an attempt to avoid reporting shims as their resolved
`sys.executable` path, we report the queried executable path instead.
This seems more correct for this command, broadly? Interestingly, it
changes the reported paths for Homebrew Python
<img width="1430" alt="Screenshot 2025-04-02 at 11 05 18 AM"
src="https://github.com/user-attachments/assets/0e1600e8-fb07-40c7-a6d6-56eaeb4b9293"
/>
Closes#9979
## Summary
This ended up being more involved than expected. The gist is that we
setup all the packages we want to reinstall upfront (they're passed in
on the command-line); but at that point, we don't have names for all the
packages that the user has specified. (Consider, e.g., `uv pip install
.` -- we don't have a name for `.`, so we can't add it to the list of
`Reinstall` packages.)
Now, `Reinstall` also accepts paths, so we can augment `Reinstall` based
on the user-provided paths.
Closes#12038.
## Summary
In. https://github.com/astral-sh/uv/issues/11857, we had a case of a
user that was seeing incorrect resolution results after upgrading to a
newer version of macOS, since we retained cache information about the
interpreter. This PR adds the OS name and version to the cache key for
the interpreter. This seems to be extremely cheap, and it's nice to make
this robust so that users don't run into the same confusion in the
future.
Closes https://github.com/astral-sh/uv/issues/11857.
Revert #11601 for now
We run Python interpreter discovery with `-I` (#2500) which means these
environments variables are ignored when determining `sys.path`. Unless
we decide to remove the `-I` flag from the `sys.path` query, we
shouldn't release these changes to interpreter discovery caching.
We want to use `sys.path` for package discovery (#2500, #9849). For
that, we need to know the correct value of `sys.path`. `sys.path` is a
runtime-changeable value, which gets influenced from a lot of different
sources: Environment variables, CLI arguments, `.pth` files with
scripting, `sys.path.append()` at runtime, a distributor patching
Python, etc. We cannot capture them all accurately, especially since
it's possible to change `sys.path` mid-execution. Instead, we do a best
effort attempt at matching the user's expectation.
The assumption is that package installation generally happens in venv
site-packages, system/user site-packages (including pypy shipping
packages with std), and `PYTHONPATH`. Specifically, we reuse
`PYTHONPATH` as dedicated way for users to tell uv to include specific
directories in package discovery.
A common way to influence `sys.path` that is not using venvs is setting
`PYTHONPATH`. To support this we're capturing `PYTHONPATH` as part of
the cache invalidation, i.e. we refresh the interpreter metadata if it
changed. For completeness, we're also capturing other environment
variables documented as influencing `sys.path` or other fields in the
interpreter info.
This PR does not include reading registry values for `sys.path`
additions on Windows as documented in
https://docs.python.org/3.11/using/windows.html#finding-modules. It
notably also does not include parsing of python CLI arguments, we only
consider their environment variable versions for package installation
and listing. We could try parsing CLI flags in `uv run python`, but we'd
still miss them when Python is launched indirectly through a script, and
it's more consistent to only consider uv's own arguments and environment
variables, similar to uv's behavior in other places.
A user reported a homebrew Python that would raise an exception in the
interpreter probing script because `platform.mac_ver()` returned `('',
('', '', ''), '')` on his installation due to
https://github.com/Homebrew/homebrew-core/issues/206778
This is easy enough to catch and show a proper error message instead of
the Python backtrace.
## Summary
This is attempting to solve the same problem surfaced in #11208 and
#11209. However, those PRs only worked for our own managed Pythons. In
Gentoo, for example, they disable the managed Pythons, which led to
failures in the test suite, because the "base Python" returned after
creating a virtual environment would differ from the "base Python" that
you get after _querying_ an existing virtual environment.
The fix here is to apply our same base Python normalization and
discovery logic, to non-standalone / non-managed Pythons. We continue to
use `sys._base_executable` for such Pythons when creating the
virtualenv, but when _caching_, we perform this second discovery step.
Closes https://github.com/astral-sh/uv/issues/11237.
## Summary
I'm not sure that this has much of an effect in practice, but currently,
when we return a virtual environment, the `sys_base_executable ` of the
parent ends up being retained as `sys_base_executable` of the created
environment. But these can be, like, subtly different? If you have a
symlink to a Python, then for the symlink, `sys_base_executable` will be
equal to `sys_executable`. But when you create a virtual environment for
that interpreter, we'll set `home` to the resolved symlink, and so
`sys_base_executable` will be the resolved symlink too, in general.
Anyway, this means that we should now have a consistent value between
(1) returning `Virtualenv` from the creation routine and (2) querying
the created interpreter.
## Summary
It turns out that we were returning slightly different interpreter paths
on repeated `uv run --with` commands. This likely didn't affect many (or
any?) users, but it does affect our test suite, since in the test suite,
we use a symlinked interpreter.
The issue is that on first invocation, we create the virtual
environment, and that returns the path to the `python` executable in the
environment. On second invocation, we return the `python3` executable,
since that gets priority during discovery. This on its own is
potentially ok. The issue is that these resolve to different
`sys._base_executable` values in these flows... The latter gets the
correct value (since it's read from the `home` key), but the former gets
the incorrect value (since it's just the `base_executable` of the
executable that created the virtualenv, which is the symlink).
We now use the same logic to determine the "cached interpreter" as in
virtual environment creation, to ensure consistency between those paths.
See https://github.com/astral-sh/uv/issues/4204 for motivation
This doesn't really reach the user experience I'd expect — i.e., we end
up saying a virtual environment "does not exist" which is a little
silly. However, I think improving the error messaging on interpreter
queries in general should be solved separately. I did one small
"general" change in
89e11d0222
— otherwise we don't show the message at all.
---------
Co-authored-by: konsti <konstin@mailbox.org>
## Summary
This PR reimplements
[`sysconfigpatcher`](https://github.com/bluss/sysconfigpatcher) in Rust
and applies it to our Python installations at install-time, ensuring
that the `sysconfig` data is more likely to be correct.
For now, we only rewrite prefixes (i.e., any path that starts with
`/install` gets rewritten to the correct absolute path for the current
machine).
Unlike `sysconfigpatcher`, this PR does not yet do any of the following:
- Patch `pkginfo` files.
- Change `clang` references to `cc`.
A few things that we should do as follow-ups, in my opinion:
1. Rewrite
[`AR`](c1ebf8ab92/src/sysconfigpatcher.py (L61)).
2. Remove `-isysroot`, which we already do for newer builds.
## Summary
These were moved as part of a broader refactor to create a single
integration test module. That "single integration test module" did
indeed have a big impact on compile times, which is great! But we aren't
seeing any benefit from moving these tests into their own files (despite
the claim in [this blog
post](https://matklad.github.io/2021/02/27/delete-cargo-integration-tests.html),
I see the same compilation pattern regardless of where the tests are
located). Plus, we don't have many of these, and same-file tests is such
a strong Rust convention.
This updates the surrounding code to use the new ResolverEnvironment
type. In some cases, this simplifies caller code by removing case
analysis. There *shouldn't* be any behavior changes here. Some test
snapshots were updated to account for some minor tweaks to error
messages.
I didn't split this up into separate commits because it would have been
too difficult/costly.
As per
https://matklad.github.io/2021/02/27/delete-cargo-integration-tests.html
Before that, there were 91 separate integration tests binary.
(As discussed on Discord — I've done the `uv` crate, there's still a few
more commits coming before this is mergeable, and I want to see how it
performs in CI and locally).
## Summary
Closes#7977. Makes `PythonDownloadRequest` account for the prerelease
part if allowed. Also stores the prerelease in `PythonInstallationKey`
directly as a `Prerelease` rather than a string.
## Test Plan
Correctly picks the relevant prerelease (rather than picking the most
recent one):
```
λ cargo run python install 3.13.0rc2
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.17s
Running `target/debug/uv python install 3.13.0rc2`
Searching for Python versions matching: Python 3.13rc2
cpython-3.13.0rc2-macos-aarch64-none ------------------------------ 457.81 KiB/14.73 MiB ^C
λ cargo run python install 3.13.0rc3
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.17s
Running `target/debug/uv python install 3.13.0rc3`
Searching for Python versions matching: Python 3.13rc3
Found existing installation for Python 3.13rc3: cpython-3.13.0rc3-macos-aarch64-none
```
Closes#7118
This only really affects managed interpreters, as we exclude alternative
Python implementations from the search path during the
`VersionRequest::executable_names` part of discovery.
## Summary
This PR adds a more flexible cache invalidation abstraction for uv, and
uses that new abstraction to improve support for dynamic metadata.
Specifically, instead of relying solely on a timestamp, we now pass
around a `CacheInfo` struct which (as of now) contains
`Option<Timestamp>` and `Option<Commit>`. The `CacheInfo` is saved in
`dist-info` as `uv_cache.json`, so we can test already-installed
distributions for cache validity (along with testing _cached_
distributions for cache validity).
Beyond the defaults (`pyproject.toml`, `setup.py`, and `setup.cfg`
changes), users can also specify additional cache keys, and it's easy
for us to extend support in the future. Right now, cache keys can either
be instructions to include the current commit (for `setuptools_scm` and
similar) or file paths (for `hatch-requirements-txt` and similar):
```toml
[tool.uv]
cache-keys = [{ file = "requirements.txt" }, { git = true }]
```
This change should be fully backwards compatible.
Closes https://github.com/astral-sh/uv/issues/6964.
Closes https://github.com/astral-sh/uv/issues/6255.
Closes https://github.com/astral-sh/uv/issues/6860.
## Summary
This is similar to https://github.com/astral-sh/uv/pull/6171 but more
expansive... _Anywhere_ that we test requirements for platform
compatibility, we _need_ to respect the resolver-friendly markers. In
fixing the motivating issue (#6621), I also realized that we had a bunch
of bugs here around `pip install` with `--python-platform` and
`--python-version`, because we always performed our `satisfy` and `Plan`
operations on the interpreter's markers, not the adjusted markers!
Closes https://github.com/astral-sh/uv/issues/6621.
## Summary
resolves https://github.com/astral-sh/uv/issues/5915, not entirely sure
if `manylinux_compatible` should be a separate field in the JSON
returned by the interpreter or there's some way to use the existing
`platform` for it.
## Test Plan
ran the below
```
rm -rf .venv
target/debug/uv venv
# commenting out the line below triggers the change..
# target/debug/uv pip install no-manylinux
target/debug/uv pip install cryptography --no-cache
```
is there an easy way to add this into the existing snapshot-based test
suite? looking around to see if there's a way that doesn't involve
something implementation-dependent like mocks.
~update: i think the output does differ between these two, so probably
we can use that.~ i lied - that "building..." output seems to be
discarded.
samypr100
Zanie Blue
John Mumm
konsti
Hood Chatham
Charlie Marsh
Geoffrey Thomas
Sede Soukossi
Aarni Koskela
Andrew Gallant
Amos Wenger
trag1c
Chan Kang