Investigate Bumpy Road semantic discrepancy with CodeScene

[coderabbitai]

Summary

Investigate the semantic difference between Whitaker's bumpy_road_function lint and CodeScene's Bumpy Road diagnostic.

Whitaker does run the lint. The non-finding is due to rule semantics, not rule absence.

The discrepancy was identified while assessing leynos/rstest-bdd#657, specifically:

• crates/rstest-bdd-macros/src/codegen/tracking/mod.rs
• TrackedFeaturePath::relative_to_manifest

CodeScene reports this function because it contains two nested conditional blocks. Whitaker does not report it.

Current Whitaker behaviour

Whitaker builds a weighted, per-line complexity signal. It reports only when it finds at least two separated bumps after smoothing.

Each bump must meet the current defaults:

• Weighted smoothed score of at least 2.5.
• Span of at least 2 contiguous lines.
• Survival through a centred moving-average window of 3.

TrackedFeaturePath::relative_to_manifest does not meet these conditions:

• The prefix-validation if has a predicate contribution of 0.5 and a body-depth contribution of 1.0.
• The common-prefix loop's if has the same maximum local contribution.
• The while let condition is a let pattern. Whitaker excludes let predicates from scoring.
• The let Some(text) = ... else construct does not create a scored if cluster.

The repository has no [bumpy_road_function] override in dylint.toml.

Expected investigation

Determine whether Whitaker should provide a mode, configuration, or supplementary diagnostic that detects CodeScene-style Bumpy Road cases. CodeScene treats two nested conditional blocks as a finding even when they do not form two sustained, separated weighted bumps.

Document the intended relationship between the two models if their semantics must remain different.

Affected area

• crates/bumpy_road_function/
• Bumpy Road configuration and user documentation
• The score, smoothing, and bump-segmentation logic

Acceptance criteria

• Reproduce and document the outcome for TrackedFeaturePath::relative_to_manifest, or an equivalent minimal Rust fixture.
• State whether semantic alignment with CodeScene is a project goal.
• If alignment is a goal, define an implementation approach and tests for this two-nested-conditional case.
• If alignment is not a goal, document the difference, including the score threshold, minimum span, smoothing window, and excluded let predicates.
• Update user-facing configuration or documentation so users can select or understand the applicable Bumpy Road model.

Source

• Requester: @leynos
• Source PR: Make a .feature file edit rebuild the scenario binary (10.3.3) rstest-bdd#657