17 commits

Author	SHA1	Message	Date
Charlie Marsh	4d5041dc00	Use files instead of junctions on Windows (#11269) ... Instead of using junctions, we can just write files that contain (as the file contents) the target path. This requires a little more finesse in that, as readers, we need to know where to expect these. But it also means we get to avoid junctions, which have led to a variety of confusing behaviors. Further, `replace_symlink` should now be on atomic on Windows. Closes #11263.	2025-02-13 16:17:49 -06:00
Charlie Marsh	dce7b9da13	Store unsupported tags in wheel filename (#10665) ... ## Summary We can retain the small-size advantage of our new tags by moving the "unknown tag" case into `WheelTagLarge`. This ensures that we can still represent unknown tags, but avoid paying the cost for them.	2025-01-17 04:41:53 +00:00
Charlie Marsh	ed34d37e3c	Avoid failing when deserializing unknown tags (#10655) ... ## Summary Closes https://github.com/astral-sh/uv/issues/10654#issuecomment-2594022975.	2025-01-15 18:03:28 -05:00
Charlie Marsh	d20a48a5b4	Use memchr for wheel parsing (#10620) ... ## Summary Before: ``` wheelname_parsing/numpy-compatible time: [106.90 ns 107.86 ns 108.86 ns] thrpt: [402.97 MiB/s 406.71 MiB/s 410.39 MiB/s] change: time: [-4.5360% -3.7694% -2.9179%] (p = 0.00 < 0.05) thrpt: [+3.0056% +3.9170% +4.7515%] Performance has improved. wheelname_parsing/flyte-short-incompatible time: [81.439 ns 82.209 ns 83.015 ns] thrpt: [390.59 MiB/s 394.42 MiB/s 398.15 MiB/s] change: time: [+6.2344% +7.5385% +8.8928%] (p = 0.00 < 0.05) thrpt: [-8.1666% -7.0101% -5.8685%] Performance has regressed. wheelname_parsing/flyte-short-compatible time: [78.909 ns 79.456 ns 80.031 ns] thrpt: [357.49 MiB/s 360.08 MiB/s 362.57 MiB/s] change: time: [+3.2653% +4.1733% +5.1062%] (p = 0.00 < 0.05) thrpt: [-4.8582% -4.0061% -3.1620%] Performance has regressed. Found 1 outliers among 100 measurements (1.00%) 1 (1.00%) high mild wheelname_parsing/flyte-long-incompatible time: [353.35 ns 357.49 ns 361.77 ns] thrpt: [340.06 MiB/s 344.13 MiB/s 348.17 MiB/s] change: time: [+1.4846% +2.0228% +2.6504%] (p = 0.00 < 0.05) thrpt: [-2.5820% -1.9827% -1.4629%] Performance has regressed. Found 10 outliers among 100 measurements (10.00%) 5 (5.00%) high mild 5 (5.00%) high severe wheelname_parsing/flyte-long-compatible time: [256.47 ns 258.12 ns 260.17 ns] thrpt: [417.88 MiB/s 421.20 MiB/s 423.90 MiB/s] change: time: [+0.4079% +1.8252% +3.6270%] (p = 0.02 < 0.05) thrpt: [-3.5001% -1.7925% -0.4063%] Change within noise threshold. Found 1 outliers among 100 measurements (1.00%) 1 (1.00%) high severe ``` After: ``` wheelname_parsing_fastest/numpy-compatible time: [61.500 ns 61.904 ns 62.350 ns] thrpt: [703.60 MiB/s 708.66 MiB/s 713.32 MiB/s] change: time: [+0.9879% +1.4542% +1.9311%] (p = 0.00 < 0.05) thrpt: [-1.8945% -1.4334% -0.9782%] Change within noise threshold. Found 8 outliers among 100 measurements (8.00%) 6 (6.00%) high mild 2 (2.00%) high severe wheelname_parsing_fastest/flyte-short-incompatible time: [49.341 ns 49.538 ns 49.769 ns] thrpt: [651.50 MiB/s 654.54 MiB/s 657.16 MiB/s] change: time: [+5.8750% +6.3656% +6.8338%] (p = 0.00 < 0.05) thrpt: [-6.3967% -5.9847% -5.5490%] Performance has regressed. Found 17 outliers among 100 measurements (17.00%) 5 (5.00%) low severe 1 (1.00%) low mild 6 (6.00%) high mild 5 (5.00%) high severe wheelname_parsing_fastest/flyte-short-compatible time: [49.425 ns 49.789 ns 50.193 ns] thrpt: [570.01 MiB/s 574.63 MiB/s 578.86 MiB/s] change: time: [+5.1267% +5.7418% +6.3476%] (p = 0.00 < 0.05) thrpt: [-5.9687% -5.4300% -4.8767%] Performance has regressed. Found 7 outliers among 100 measurements (7.00%) 7 (7.00%) high mild wheelname_parsing_fastest/flyte-long-incompatible time: [295.81 ns 298.01 ns 301.04 ns] thrpt: [408.66 MiB/s 412.82 MiB/s 415.89 MiB/s] change: time: [+0.5553% +1.0842% +1.7059%] (p = 0.00 < 0.05) thrpt: [-1.6772% -1.0726% -0.5523%] Change within noise threshold. Found 4 outliers among 100 measurements (4.00%) 3 (3.00%) high mild 1 (1.00%) high severe wheelname_parsing_fastest/flyte-long-compatible time: [214.80 ns 216.10 ns 217.50 ns] thrpt: [499.87 MiB/s 503.10 MiB/s 506.15 MiB/s] change: time: [+0.9003% +1.3207% +1.8268%] (p = 0.00 < 0.05) thrpt: [-1.7940% -1.3035% -0.8923%] Change within noise threshold. Found 8 outliers among 100 measurements (8.00%) 7 (7.00%) high mild 1 (1.00%) high severe ``` (Ignore the percent changes; they're not relative to one another.) So it's like >40% faster for the small case and >15% faster for the large case.	2025-01-15 11:39:25 -05:00
Charlie Marsh	a7fe84aa03	Avoid allocating when stripping source distribution extension (#10625)	2025-01-15 03:29:08 +00:00
Charlie Marsh	279043f864	Include build tag in rendered wheel filenames (#10599) ... ## Summary I don't think this had an impact in practice, but it is "wrong" to omit these. Confirmed that the cache (for example) now includes the build tag (as in, `mkl_fft-1.3.8-72-cp310-cp310-manylinux2014_x86_64`).	2025-01-14 15:01:45 +00:00
Charlie Marsh	24a5920739	Reduce WheelFilename to 48 bytes (#10583) ... ## Summary Based on some advice from @konstin.	2025-01-14 14:49:17 +00:00
Charlie Marsh	faa4481ccc	Shrink size of platform tag enum (#10546) ... ## Summary Reduces it from 56 bytes to 16 bytes.	2025-01-14 03:13:59 +00:00
Charlie Marsh	5c91217488	Use structured wheel tags everywhere (#10542) ... ## Summary This PR extends the thinking in #10525 to platform tags, and then uses the structured tag enums everywhere, rather than passing around strings. I think this is a big improvement! It means we're no longer doing ad hoc tag parsing all over the place.	2025-01-14 01:39:39 +00:00
Charlie Marsh	e0e8ba582a	Use structured types for parsing and formatting language and ABI tags (#10525) ... ## Summary I need to be able to do non-lexicographic comparisons between tags (e.g., so I can sort `cp313` as greater than `cp39`). It ended up being easiest to just create structured types for all the tags we support, with `FromStr` and `Display` implementations. We don't currently store these in `Tags` or in `WheelFilename`. We may want to, since they're really small (and `Copy`), but I need to benchmark to determine whether parsing these in `WheelFilename` is prohibitively slow.	2025-01-13 19:49:43 -05:00
Charlie Marsh	9e948b7363	Remove resolved build tag TODO (#10526)	2025-01-11 23:12:11 +00:00
Charlie Marsh	3aaa9594be	Add test coverage for build tag prioritization (#9680) ... ## Summary See: https://github.com/astral-sh/uv/pull/3781 and https://github.com/astral-sh/uv/pull/9677.	2024-12-06 09:32:19 -05:00
konsti	566c178276	uv build: Catch version mismatch between sdist and wheel (#9633) ... When building a wheel from a source distribution or both a source distribution and a wheel, the versions in their filenames must be the same. By inspecting the filenames, we also assert that the filenames from the build a valid (we don't enforce normalization though, just that uv can parse them). Note that we're not yet checking that also the `pyproject.toml` version, if declared, and METADATA version matches.	2024-12-04 15:21:01 +01:00
Charlie Marsh	d08bfee718	Remove separate test files in favor of same-file mod tests (#9199) ... ## 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.	2024-11-18 20:11:46 +00:00
Charlie Marsh	29e1b15473	Add uv tree --outdated (#8893) ... ## Summary Similar to `pip list --outdated`, but for `uv tree`. ## Test Plan Looks like: ``` foo v0.1.0 └── flask v2.0.0 (latest: v3.0.3) ├── click v8.1.7 ├── itsdangerous v2.2.0 ├── jinja2 v3.1.4 │ └── markupsafe v3.0.2 └── werkzeug v3.1.2 └── markupsafe v3.0.2 ``` With `(latest: v3.0.3)` in bold cyan.	2024-11-07 14:10:46 -06:00
Amos Wenger	715f28fd39	chore: Move all integration tests to a single binary (#8093) ... 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).	2024-10-11 16:41:35 +02:00
Charlie Marsh	14507a1793	Add uv- prefix to all internal crates (#7853) ... ## Summary Brings more consistency to the repo and ensures that all crates automatically show up in `--verbose` logging.	2024-10-01 20:15:32 -04:00