## Summary
For pyx, we can allow uploads that bypass the registry and send the file
directly to S3. This is an opt-in feature, enabled via the `--direct`
flag.
<!--
Thank you for contributing to uv! To help us out with reviewing, please
consider the following:
- Does this pull request include a summary of the change? (See below.)
- Does this pull request include a descriptive title?
- Does this pull request include references to any relevant issues?
-->
## Summary
<!-- What's the purpose of the change? What does it do, and why? -->
This allows users to set the HTTP, HTTPS, and no proxy variables via the
configuration files like ~pyproject.toml~ and uv.toml.
Users can set like so:
`uv.toml`
```toml
https-proxy = "http://my_cool_proxy:10500"
http-proxy = "http://my_cool_proxy:10500"
no-proxy = [
"dontproxyme.com",
"localhost",
]
```
Resolves https://github.com/astral-sh/uv/issues/9472
## Test Plan
<!-- How was it tested? -->
It also adds a new integration test for the proxy support in
`uv-client`.
This was tested on some of our developer machines with our proxy setup
using `~/.config/uv/uv.toml` with values like:
```toml
https-proxy = "http://my_cool_proxy:10500"
http-proxy = "http://my_cool_proxy:10500"
no-proxy = [
"dontproxyme.com",
"localhost",
]
```
---------
Signed-off-by: Eli Uriegas <eliuriegas@meta.com>
Co-authored-by: Zanie Blue <contact@zanie.dev>
## Summary
Fixes https://github.com/astral-sh/uv/issues/17232
When `uv build` fails (e.g., due to missing `__init__.py`), partial
distribution
files were being left in the `dist/` directory.
### Changes
- Use `NamedTempFile` to write build output to a temporary file first
- Only persist the file to the final location on successful build
- If build fails, the temporary file is automatically cleaned up
- Added `Error::Persist` variant for handling persistence failures
## Test Plan
Added `no_partial_files_on_build_failure` test that verifies:
1. `build_source_dist` fails when `__init__.py` is missing
2. `build_wheel` fails when `__init__.py` is missing
3. The `dist/` directory remains empty after both failures (no partial
files)
---------
Co-authored-by: Hayashi Reo <reo@wantedly.com>
Formats the generated JSON schema with prettier so it doesn't fail CI
lints immediately after update.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Previously, we had a retry count both on the top level error type, and
on an error variant, and had a conversion step in between. When
reviewing #17274, I noticed we can simplify that.
## Summary
Fixes#17266.
The retry count was getting dropped by
`ErrorKind::from_retry_middleware` and `<Error as
From<ErrorKind>>::from` so we were doing more retries than we should
have.
## Test Plan
Added a testcase for the specific error path in the issue. Added an
expect to the other retry test too.
## Summary
Gates four tests in `python_find.rs` that call `python_install()` behind
the `python-managed` feature flag. These tests attempt to download
Python versions from the network (free-threaded and pre-release
versions) which fail in offline build environments.
Fixes#16431
## Test Plan
Verified that the gated tests match the pattern used elsewhere in the
codebase where the entire `python_install` module is already gated
behind `#[cfg(feature = "python-managed")]`.
Since #16923, `-` stdin paths are suddenly only supported on the
`RequirementsSource::Extensionless`. However, parsing of cli arguments
using `from_requirements_txt`, `from_constraints_txt`
`from_overrides_txt` would always output a
`RequirementsSource::RequirementsTxt`. Resulting in the error:
```
$ cat overrides.txt | cargo run --bin uv --profile dev --manifest-path ./uv/crates/uv/Cargo.toml pip install 'numpy' --overrides=-
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.22s
Running `./uv/target/debug/uv pip install 'numpy' --overrides=-`
error: File not found: `-`
```
In this PR, I've added a small check in those for the `-` paths to use
`RequirementsSource::ExtensionLess`.
I'm not too sure about this change though, as it would also implicitly
start allowing PEP 723 scripts as input to overrides/constraints. I
don't see the direct issue in that, but then maybe we should explicitly
handle it so that an `--overrides=script.py` would also be supported.
@zanieb what do you think?
Relates to #17227
## Summary
Partially address #16709.
Previously, if cornered, `pip compile` would fail when the requested
python interpreter couldn't be found (more details in the issue and
comments), and now in those cases it will download it.
## Test Plan
Added an integration test for this case.
This also removes the file-specific targets from prettier execution
which means we're including `.json`, `.css`, and `.html` files, which
seems like an improvement.
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
Implement #16653 by making `uv sync --output-format=json` output
information about package changes.
## Test Plan
Additional tests to test the cases where there is no known package
version _may_ be beneficial but as the information used is the same as
the information used by the dry run logging now, I don't think that's
strictly necessary as those cases are tested.
---------
Co-authored-by: Liam <liam@scalzulli.com>
We were using slightly different retry code in multiple places, this PR
unifies it.
Also fixes retry undercounting in publish if the retry middleware was
involved.
---------
Co-authored-by: Tomasz Kramkowski <tom@astral.sh>
<!--
Thank you for contributing to uv! To help us out with reviewing, please
consider the following:
- Does this pull request include a summary of the change? (See below.)
- Does this pull request include a descriptive title?
- Does this pull request include references to any relevant issues?
-->
## Summary
improve code comments clarity
<!-- What's the purpose of the change? What does it do, and why? -->
## Test Plan
<!-- How was it tested? -->
Signed-off-by: stringscut <stringscut@outlook.jp>
## Summary
Right now, when we return a `Dist` from a lockfile, we concatenate all
hashes for all distributions for a given package. In the case of
https://github.com/astral-sh/uv/issues/17143, I think that means we'll
return the SHA256 from the sdist, plus the SHA512 from the wheel. If the
wheel was previously installed (i.e., it's in the cache), and we
computed the SHA256 at that point in time, then `Hashed::has_digests`
would return `true` because we have _at least_ one SHA256. We now limit
the hashes to the distribution that we expect to install.
Closes https://github.com/astral-sh/uv/issues/17143.
