Skip to content

Performance: benchmark cold and cached extraction - #96

Merged
LukasNiessen merged 4 commits into
mainfrom
perf/46-extraction-benchmark
Aug 24, 2026
Merged

LukasNiessen merged 4 commits into
mainfrom
perf/46-extraction-benchmark

Conversation

@LukasNiessen

@LukasNiessen LukasNiessen commented Aug 24, 2026 •

Copy link
Copy Markdown
Collaborator

Closes #46

What changed

  • generate a deterministic 240-file, 2,880-internal-import benchmark project outside measured windows
  • measure public enumeration, loading, parse, resolution/classification, normalization, projection, cold rule, and cached rule stages
  • record per-stage allocation calls/requested bytes/peak live bytes with zero-live-byte gates
  • pin graph counts and a cross-run digest, re-normalize independently, and exercise one cold plus twenty cached fluent checks
  • write zig-out/benchmark/results.json, enforce evidence-backed regression budgets, and preserve artifacts even when a budget fails
  • document D050, benchmark usage, ownership scope, CI evidence, and budget policy

Hosted evidence

Local QA

  • zig fmt --check build.zig build.zig.zon src benchmark
  • python -B -m unittest scripts.test_install_zig (5 tests)
  • python -B scripts/check_ci.py (10 immutable actions, 6 Zig archives)
  • zig test benchmark/contracts.zig (3 tests)
  • zig build benchmark-check (5.21 s total, 5,966,099 peak tracked bytes, zero final tracked bytes)
  • zig build test (complete 529-test suite)

@LukasNiessen
LukasNiessen merged commit 5ee8c17 into main Aug 24, 2026
5 checks passed
@LukasNiessen
LukasNiessen deleted the perf/46-extraction-benchmark branch August 24, 2026 12:16
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.

Performance: benchmark and improve cold and cached extraction

1 participant