Skip to content

perf(runner): reuse unchanged file metadata - #1380

Merged
Chemaclass merged 1 commit into
mainfrom
feat/1372-metadata-cache
Oct 2, 2026
Merged

Chemaclass merged 1 commit into
mainfrom
feat/1372-metadata-cache

Conversation

@Chemaclass

Copy link
Copy Markdown
Member

🤔 Background

Related #1372

Header counting discards metadata needed again during execution. Reuse unchanged short-file scans while checking complete contents before replay, so earlier files can still rewrite later ones.

💡 Changes

  • Reuse provider and annotation metadata for unchanged files up to 512 characters in sequential multi-file runs with a header.
  • Preserve discovery and dynamic-file counting, with a fresh scan whenever content changes or exceeds the bound.
  • Cover same-size rewrites, scan-time changes, reordered paths, scan budgets and total-count parity. Update the changelog.
  • Source, standalone, static and formatting gates pass. Two 40-file fixtures improve by 9% and 7%, including freshness checks. Medians use nine alternating pairs against main at 5731afa on macOS Bash 3.2 with the same paths and fixtures; repeated larger-file medians differ by less than 1%.

@Chemaclass Chemaclass added the enhancement New feature or request label Oct 2, 2026
@Chemaclass Chemaclass self-assigned this Oct 2, 2026
@Chemaclass
Chemaclass merged commit cbe767a into main Oct 2, 2026
37 checks passed
@Chemaclass
Chemaclass deleted the feat/1372-metadata-cache branch October 2, 2026 15:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant