Previously we were using the XDG data dir to avoid symlinks, but there's no
particular guarantee that that's not going to be a symlink too. Using the
canonicalized temp dir by default is also slightly nicer for a couple reasons:
It's sometimes faster (an in-memory tempfs on e.g. Arch), and it makes
overriding `$TMPDIR` or `%TMP%` sufficient to control where tests put temp
files, without needing to override `UV_INTERNAL__TEST_DIR` too.
Replaces https://github.com/astral-sh/uv/pull/12320
Switches to Depot for the large Windows runner we use for `cargo test`.
The runtime goes from 8m 20s -> 6m 44s (total) and 7m 18s -> 4m 41s
(test run) which are 20% and 35% speedups respectively.
A few things got marginally slower, like Python installs went from 11s
-> 38s, the Rust cache went from 15s -> 30s, and drive setup went from
7s -> 20s.
When using the standard Windows runners (as opposed to the _larger_
GitHub runners), an undocumented `D:` drive is available and performant.
We can save some money on by using this on a standard runner instead of
a larger runner with an ReFS drive. Switching to the `D:` drive was not
acceptable for `cargo test` >25m runtime.
Inspired by https://github.com/pypa/pip/pull/13129
See https://github.com/actions/runner-images/issues/8755
Timings (grain of salt — GitHub is super noisy):
- clippy: 2m 18s -> 2m 11s
- build binary: 2m 3s -> 2m 35s
- trampoline check (x86-64): 2m 32s -> 1m 50s (other architectures
similar)
- trampoline test (x86-64): 4m 12s -> 6m 7s
- trampoline test (i686): 6m 44s -> 5m 35s
As suggested by @samypr100 on #6680:
https://github.com/astral-sh/uv/pull/6680#issuecomment-2313607984
## Summary
Instead of using `UV_INTERNAL__TEST_DIR`, it simply exports `TEMP` when
running Windows jobs.
## Test Plan
I'm going to run this manually under ProcMon on my Windows machine and
see where uv writes temp files, hopefully to the dev drive and not
`%(LOCAL)APPDATA%` or something.
I'm going to commit a dummy code change and look at build time changes
in CI.
Zanie Blue