oxlint-plugin-react-doctor@0.5.8

Patch Changes

• #903 627f9ca Thanks @rayhanadev! - Add a per-file content-addressed lint cache so repeat scans re-lint only the files whose content changed. On a warm scan the oxlint pass partitions the file list by content hash: unchanged files replay their cached raw diagnostics, and only changed files are re-linted. The five cross-file rules (no-barrel-import, nextjs-missing-metadata, nextjs-no-use-search-params-without-suspense, no-mutating-reducer-state, rn-prefer-expo-image) — whose verdict for a file can depend on other files — always run fresh in a never-cached sidecar pass, so a dependency change can never serve a stale verdict. Output is byte-identical with the cache on or off (the design invariant), so the score, JSON report, and inspect()/diagnose() return values are unchanged.

  The cache is on by default and content-hashed (so it survives CI re-clones), and is automatically bypassed in audit mode, when an extends lint config is adopted, or when user plugins are configured. Disable it with REACT_DOCTOR_NO_FILE_CACHE=1; the existing REACT_DOCTOR_NO_CACHE=1 now disables both the whole-repo scan cache and this per-file cache. A cross-file-rules guard test fails if a future rule starts reading other files without being carved into the always-fresh sidecar. The CLI reports cache effectiveness on its Sentry run event as lintCacheHitRatio.

  oxlint-plugin-react-doctor now exports CROSS_FILE_RULE_IDS, the canonical set of rules whose verdict can depend on other files.