Breakwater

Backtest / #1028

perf: don't renormalize extras

pypa/packaging#1028 · merged 2026-01-06 · base 65092ce350 → head b20096ccce · ran on Docker

PASS

No behaviour divergence found by any probe.

Performance · Requirement

No realistic input family measured slower.

Measured in backtest packaging-v2 (same commits, run bt-1028-74aaa2, image bw-pair:1777b4e053ae8975). The semantic results above are from a later run; perf doesn't use the semantic planner, so it was not re-measured.

head slowerhead faster1×input size (characters)F2F2: 1.01× at ~35 charsF2: 1.01× at ~55 charsF2: 1.00× at ~95 charsF2: 1.00× at ~95 charsF2: 1.00× at ~95 charsF2: 1.00× at ~95 charsF2: 1.00× at ~95 chars
  • F2
Head/base cachegrind instructions per call, by input size (log-log). Each line is one input family Nemotron proposed; points above 1× are where the "faster" code does more work.
FamilyInputsRatio range Slope base → headVerdict
F2The diff removes a double canonicalize_name call for the "extra" key in _normalize and skips normalisation when the value is empty. A requirement whose marker claimed speedup · plan1.00× – 1.01×0.00 → 0.00neutral

Receipt

TargetBackendImageRun AuditAdjudication
packaging.requirements:Requirementlocal Dockerbw-pair:5bb20e4196737a91bt-1028-3e02dc$0.021$0.000

Nemotron cost on Nebius Token Factory. Backend: where the untrusted code ran. local Docker (an image cited by its content-addressed tag). Every image is content-addressed (repo at each commit + harness), so every card re-runs from the image and seed 0. Raw data: audit.