Skip to content

feat: QJL sketch dedup and semantic lint grouping#873

Open
MaTriXy wants to merge 1 commit intortk-ai:developfrom
MaTriXy:turboquant-enhancements-v2
Open

feat: QJL sketch dedup and semantic lint grouping#873
MaTriXy wants to merge 1 commit intortk-ai:developfrom
MaTriXy:turboquant-enhancements-v2

Conversation

@MaTriXy
Copy link

@MaTriXy MaTriXy commented Mar 26, 2026

Summary

  • QJL-inspired line sketch module (src/core/sketch.rs) — 64-bit fuzzy line fingerprinting with Hamming similarity
  • Sketch-enhanced log dedup — two-layer approach (normalize + sketch threshold 0.85) with exact-match fallback
  • Semantic lint sub-grouping — ESLint rules with >5 violations clustered by message similarity

Files changed

  • src/core/sketch.rs — LineSketch module (new)
  • src/core/mod.rs — added pub mod sketch
  • src/cmds/system/log_cmd.rs — sketch-enhanced fuzzy dedup
  • src/cmds/js/lint_cmd.rs — semantic sub-grouping
  • docs/turboquant-enhancements.md — documentation (new)

Test plan

  • cargo build clean
  • Benchmark sketch dedup on 10K+ line logs
  • Verify lint grouping on real ESLint JSON with >50 violations

Apply TurboQuant-inspired improvements:
- QJL-inspired 64-bit line sketch module (src/core/sketch.rs)
- Sketch-enhanced log dedup (threshold 0.85) with exact-match fallback
- Semantic sub-grouping for ESLint violations (>5 per rule)
@CLAassistant
Copy link

CLAassistant commented Mar 26, 2026

CLA assistant check
All committers have signed the CLA.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants