benchmark-results-publisher

You can update README tables safely by publishing benchmark result JSONs through a validated conversion process. This approach integrates with existing docs workflows to prevent raw pyperf data from polluting your documentation.

Raw pyperf JSON files pollute documentation because they contain verbose performance data unsuitable for readme tables. Validating summary files as *_results.json prevents this raw data from entering your docs.

Convert benchmark JSONs into documentation-ready formats by running the update_docs.sh script. This automates the conversion of result data into safe, readme-friendly tables and figures while excluding gcp_runs/ and output/ artifacts.

Yes, you can include Pillow benchmark results for public docs when available. The publishing process ensures Pillow results are preserved specifically for website documentation alongside MacBook RGB results.

Exclude gcp_runs/ and output/ artifacts when publishing benchmark results to docs. Additionally, avoid raw pyperf files (*.pyperf.json) and only validate summary files matching the *_results.json naming convention.

Yes, benchmark-results-publisher integrates with existing README/docs workflows to keep benchmarks up-to-date. It automates the safe conversion of result JSONs into documentation-ready tables without disrupting your current documentation process.