Add --import-json to combine results from separate runs (fixes #830)

[leno23]

Summary

Adds a --import-json FILE option to load benchmark results from a previous --export-json export and include them in the summary and relative speed comparison of the current run.

This supports the common workflow of benchmarking a baseline, applying a system change, and then benchmarking again — without manually comparing two separate hyperfine outputs.

Example:

hyperfine --export-json baseline.json "dracut --debug --force"
# apply patch...
hyperfine --import-json baseline.json "dracut --debug --force"

Changes

• Add --import-json CLI flag and Options::import_json
• Deserialize JSON exports via load_benchmark_results() (restores display command when omitted)
• Prepend imported results in Scheduler before running new benchmarks
• Add unit and integration tests

Fixes #830

Test plan

• cargo test
• Integration test: export JSON, import it in a second run, verify relative comparison includes imported command
• Integration test: fail early when import file does not exist

Made with Cursor

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 591af06162

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve explicit reference position when importing results

Prepending imported results here shifts all newly measured benchmarks to the right, but print_relative_speed_comparison still assumes self.results[0] is the --reference benchmark when --reference is set. In runs that combine --import-json and --reference, the first imported entry is treated as the reference instead of the command passed via --reference, which produces incorrect relative-speed ratios and summary text for that workflow.

Useful? React with 👍 / 👎.