67d8805...
by
Andrew Gallant <email address hidden>
uv-toolchain: use colocated temporary directory (#3240)
Previously, this would use the "system" temporary directory.
Because we rename the resulting directory to its final destination,
and because renaming is implemented via hardlinking, and because
hardlinking doesn't work across mount points, and because /tmp is
commonly on a different mount point than where `uv` is checked out,
we "fix" this by putting the temporary directory somewhere close to
the final destination of the fetched artifact.
There are alternatives we might consider pursuing. For example,
if the `rename` fails, then we should probably do a recursive
directory copy. But this is a quick fix for now and it also
consistent with colocation of other temporary directories in `uv`.
The main downside of this change is that if a user does ^C while
`uv-dev fetch-python` is running, then there is no mechanism for
cleaning up temporary directories.
On `main` you can see a dozen keyring attempts at once. Here, the other
requests wait for the first attempt and only a single keyring call is
performed.
116d47e...
by
Charlie Marsh <email address hidden>
Add alternate manylinux targets to `--python-platform` (#3229)
## Summary
I initially implemented this by allowing arbitrary
`x86_64-manylinux_x_y`, but it makes the Clap, Serde, and Schemars
definitions more complicated, _and_ makes it harder for the user.
Ultimately, manylinux itself only provides images for 2_17 and 2_28, so
this seems like it should be sufficient.
84989a3...
by
Charlie Marsh <email address hidden>
Unroll self-dependencies via extras (#3230)
## Summary
We now recursively expand any self-dependencies via extras, which lets
us detect conflicts sooner and avoid building unnecessary versions of
packages that are excluded via the extra.
Previously, uv-auth would fail to compile due to a missing process
feature. I chose to make all tokio features we use top level features,
so we can share the tokio cache between all test invocations.
8b711d2...
by
Charlie Marsh <email address hidden>
Avoid adding extras when expanding constraints (#3232)
## Summary
See the diff in the tests. If you have a constraint with an extra, we
should respect it, but we shouldn't _add_ the extra to the requirements.