[liquidsec]

Prepares the 4.3.0 release. After this merges, merging dev into stable publishes it to PyPI and crates.io.

Changes in this PR

• Version 4.2.0 → 4.3.0 in Cargo.toml, pyproject.toml and the lockfiles. 4.3.0 is free on both PyPI and crates.io.
• Windows and macOS wheel jobs now install Python 3.11–3.15 explicitly. They used to install only 3.x and rely on maturin --find-interpreter finding whatever else was on the runner, which is why 4.2.0's macOS wheels covered only 3.11–3.14. Linux needs no change: the manylinux and musllinux images already ship CPython 3.15.

Verified locally

• fmt, clippy, cargo test, pytest and ruff all pass.
• cargo publish --dry-run packages cleanly.
• maturin build and maturin sdist succeed. The sdist metadata shows Version: 4.3.0, License: AGPL-3.0, Requires-Python: >=3.11.

What's in 4.3.0 (dev since 4.2.0)

Breaking-ish

• Requires Python 3.11+. Python 3.9 and 3.10 are dropped (build(deps): bump pyo3 to 0.30 for Python 3.15, drop 3.9 and 3.10 #103). Installs on 3.9 and 3.10 keep resolving to 4.2.0.
• Relicensed from GPL-3.0 to AGPL-3.0 (docs: relicense from GPL-3.0 to AGPL-3.0 #93).

Features and performance

• Python 3.15 support: pyo3 0.27 → 0.30 (build(deps): bump pyo3 to 0.30 for Python 3.15, drop 3.9 and 3.10 #103). Unblocks Drop Python 3.10, add Python 3.15 bbot#3474.
• Faster IP tree (Faster IP Tree #64): a tree-bitmap IP index, about 4.5–14x faster for IP inserts and lookups in CI's benchmark. Its correctness is checked against the previous implementation by a randomized differential test, and it contains no unsafe code.

CI

• The benchmark now compares each PR against its target branch (ci(benchmark): compare PR against its target branch #102), and skips the PR comment on fork and dependabot PRs (ci(benchmark): skip PR comment on dependabot and fork PRs #101).
• Dependabot targets dev for GitHub Actions updates (ci: target dev for github-actions dependabot updates #99). Lint fixes for the latest ruff and clippy (fix: get CI green (latest ruff + clippy lint fixes) #98).

[github-actions]

RadixTarget Benchmark Results

100000 networks per table, 100000 lookups, median of 5 runs (seed 1337).

Benchmark	dev (ms)	release/4.3.0 (ms)	Change
IPv4 insert	20.16	20.77	+3.0%
IPv4 lookup	18.35	18.33	-0.1%
IPv6 insert	46.09	49.67	+7.8%
IPv6 lookup	40.76	40.12	-1.6%

Change is in time taken: negative is faster.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100%. Comparing base (9fc51ef) to head (0133701).
⚠️ Report is 28 commits behind head on dev.

Additional details and impacted files

@@         Coverage Diff         @@
##            dev   #104   +/-   ##
===================================
  Coverage   100%   100%           
===================================
  Files         3      3           
  Lines       604    605    +1     
===================================
+ Hits        604    605    +1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.