[astral-automations-bot]

Summary

We decouple the build backend's parsing of build-system.requires from uv's supported installation sources. The build frontend resolves and installs these requirements before invoking the backend, so the backend should not enforce uv's Git transport or archive-format restrictions. We retain general requirement and URL parsing while removing distribution classification from this field, allowing the standalone uv-build binary to omit Git parsing and archive-extension inference.

This intentionally broadens acceptance. The backend now accepts Mercurial references such as hg+https://example.com/repo and archive download URLs without a recognized filename extension. It also accepts strings such as git+banana://example.com/repo and git+://example.com/repo that parse as URLs but fail uv's Git URL validation. Acceptance by the backend does not establish that a requirement is installable; the frontend owns that check. uv's direct-build validation and project-dependency parsing retain their existing URL checks.

In the Linux x86_64 comparison against main at e37c1f0c6, using Rust 1.99.0 and the stripped minimal-size profile, uv-build shrank from 3,075,808 to 3,029,232 bytes (46,576 bytes, 1.51%). This subsumes the standalone-binary size saving from astral-sh#2753; the two savings are not additive.

[astral-sh-bot]

uv test inventory changes

This PR changes the tests when compared with the main base revision.

• Added tests: 1
• Removed tests: 0
• Changed suites: 1

uv-build-backend: +1 / -0

Added:

• uv-build-backend::metadata::tests::build_system_frontend_url

Removed: none

[codspeed]

Merging this PR will improve performance by 8%

⚠️ 8 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• discover_workspace_from_all_members
• discover_workspace_from_all_members_with_excludes
• unpack_sdist_many_files
• install_wheel_many_files
• resolve_warm_jupyter
• resolve_warm_jupyter_universal
• run_python_version_synthetic_workspace
• resolve_warm_airflow

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 3 improved benchmarks
✅ 66 untouched benchmarks
⏩ 54 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡ ⚠️	Simulation	discover_workspace_from_all_members	152.9 ms	134.8 ms	+13.39%
⚡	WallTime	install_wheel_many_files	14.1 ms	13.4 ms	+5.41%
⚡	WallTime	discover_workspace_from_all_members	8.2 ms	7.8 ms	+5.38%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing charlie/codex-uv-build-requirement-urls (bcbfc13) with main (3ea045b)}

Footnotes

1. 54 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩