## Summary
We serialize the index to the lockfile without the username, so if we
compare based on `==` and the user _includes_ the username in their
`pyproject.toml`, the check will always fail.
Closes https://github.com/astral-sh/uv/issues/16076.
## Summary
This PR enables users to mark a URL as an S3 endpoint, at which point uv
will sign requests to that URL by detecting credentials from the
standard AWS environment variables, configuration files, etc.
Signing is handled by the
[reqsign](https://docs.rs/reqsign/latest/reqsign/) crate, which we can
also use in the future to sign requests for other providers.
## Summary
This PR adds support for pyx to `uv auth login`, `uv auth logout`, and
`uv auth token`. These are generic uv commands that can be used to store
credentials for arbitrary indexes and other URLs, but we include a
fast-path for pyx that initiates the appropriate login or logout flow.
## 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.
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).
In #2976 I made some changes that led to regressions:
- We stopped tracking URLs that we had not seen credentials for in the
cache
- This means the cache no longer returns a value to indicate we've seen
a realm before
- We stopped seeding the cache with URLs
- Combined with the above, this means we no longer had a list of
locations that we would never attempt to fetch credentials for
- We added caching of credentials found on requests
- Previously the cache was only populated from the seed or credentials
found in the netrc or keyring
- This meant that the cache was populated for locations that we
previously did not cache, i.e. GitHub artifacts(?)
Unfortunately this unveiled problems with the granularity of our cache.
We cache credentials per realm (roughly the hostname) but some realms
have mixed authentication modes i.e. different credentials per URL or
URLs that do not require credentials. Applying credentials to a URL that
does not require it can lead to a failed request, as seen in #3123 where
GitHub throws a 401 when receiving credentials.
To resolve this, the cache is expanded to supporting caching at two
levels:
- URL, cached URL must be a prefix of the request URL
- Realm, exact match required
When we don't have URL-level credentials cached, we attempt the request
without authentication first. On failure, we'll search for realm-level
credentials or fetch credentials from external services. This avoids
providing credentials to new URLs unless we know we need them.
Closes https://github.com/astral-sh/uv/issues/3123
konsti
Charlie Marsh
Amos Wenger
Zanie Blue