The three benchmark scripts have no tests:
- `scripts/run.sh` — handles complex JSON parsing, floating-point arithmetic, fixture generation
- `scripts/compare.sh` — performs percentage calculations, threshold comparisons, regression detection
- `scripts/format-release.sh` — formats benchmark results for release notes
These scripts handle numeric data and reporting. Regressions in the scripts could corrupt benchmark results silently. The recent changelog (v2.2.1 through v2.2.4) shows multiple corrections to script behavior, confirming they are prone to edge case issues.
Expected
- Shell unit tests (e.g., with bats-core) for critical functions
- Test cases for edge conditions: zero baseline, missing data, empty results
- Validation of JSON output format
The three benchmark scripts have no tests:
These scripts handle numeric data and reporting. Regressions in the scripts could corrupt benchmark results silently. The recent changelog (v2.2.1 through v2.2.4) shows multiple corrections to script behavior, confirming they are prone to edge case issues.
Expected