v1.10.0

Fettle v1.10.0: Reproducible Mutation Evidence

Fettle v1.10.0 turns full-repository Python mutation testing from an expensive
experiment into reproducible assurance evidence.

What Changed

• Two independent full calibrations reproduced all 28,723 canonical outcomes:
  14,107 killed, 14,611 survived, five native timeouts, zero suspicious, and
  zero untested mutants.
• Fingerprint-keyed checkpoints can resume compatible full calibrations without
  rerunning terminal mutants.
• Baseline comparison now separates surviving-mutant changes from timeout,
  suspicious, and skipped evidence debt.
• fettle mutation status correctly evaluates retained reports against a
  committed baseline.
• README and active documentation now route users by adoption outcome and state
  current product boundaries explicitly.

Adoption And Upgrade

Install or upgrade the zero-runtime-dependency CLI:

pipx upgrade finefettle
# or: pipx install finefettle

Ruff, Semgrep, and mutation tooling remain explicit, user-controlled tools. The
v1.10.0 wheel supports CLI workflows, rules, and templates; live agent
transports still require the documented Git checkout path in this release.

Evidence

• The accepted repository mutation floor is 49.1 percent for this pinned
  calibration, not a universal quality target.
• Release artifacts include the wheel, source distribution, CycloneDX SBOM,
  and GitHub build-provenance attestations.
• Changed-scope mutation enforcement remains advisory until representative
  runtime and reviewer-actionability criteria pass.

Full Changelog

v1.9.0...v1.10.0