Skip to content

Latest commit

 

History

History
35 lines (25 loc) · 1.37 KB

File metadata and controls

35 lines (25 loc) · 1.37 KB

Selecting Benchmark Candidates

Evaluate repositories against these criteria before creating a new benchmark.

Required Criteria

Criteria Description
Standardized interface CLI (stdin/stdout/args), REST API, gRPC, or file processor. No libraries or GUIs.
Adaptable test suite Tests that can verify behavior through external interface only.
Deterministic behavior Same input → same output, or existing tests handle variability.
Self-contained No external dependencies like databases, authentication services, cloud APIs.
Size: 50k - 200k LOC Large enough to be meaningful, not too large to be impractical.

Bonus

  • Alternative implementation exists in target language
  • Well-documented
  • Actively maintained

Quick Checklist

□ Interface type: CLI / REST API / gRPC / File processor
□ Has good coverage of implementation-agnostic tests (or easily convertible)?
□ Is behavior deterministic (or tests handle variability)?
□ No database, auth, or external service dependencies?
□ LOC between 50k - 200k?

Examples

Good candidate: jq (C) - CLI interface, comprehensive test suite, deterministic JSON processing, self-contained, ~50k LOC

Poor candidate: React - Library (no external interface), requires browser environment, massive ecosystem