Skip to content

Throughput metrics (LTFC + Deployment Frequency)

Repeatable delivery metrics sourced from local git. Built to iterate on the throughput calcs and the per-active-contributor benchmark.

Why these sources are trustworthy

  • LTFC (lead time for changes) uses git author dates — immune to rebase (rebase rewrites committer dates only). Verify author ≈ committer dates on the target repo.
  • DF: a merge to the trunk counts as a deployment only if merges deploy. Confirm the deploy trigger before labelling merge frequency as deployment frequency.
  • Cycle time (PR created_on → merge) needs a forge API and is out of the default git-only path.

Pure-Python, stdlib only (subprocess) — no pip install required.

Files

File Role
metrics.py Pure calc engine (no IO). All formulas + classification live here.
test_metrics.py unittest tests for the engine.
collect.py IO layer: enumerate PR merges from git, print/emit records.

Run

# from this scripts/ directory
python -m unittest test_metrics.py

# current 30-day window
python collect.py --since 2026-05-17 --until 2026-06-16

# by day-count instead of an end date
python collect.py --since 2026-05-17 --days 30

# dump raw records for offline analysis
python collect.py --since 2026-04-17 --json /tmp/records.json

Definitions (tunable in metrics.py)

  • Micro change: files<=1 and churn<=5 and commits<=1 (trivial one-liner).
  • Meaningful change: not micro.
  • Genuinely-active contributor: >=3 PRs OR >=5 active days in the window (require_both=True to AND them).
  • Benchmark: 1 good change per 2 active contributors per work-day (contributors_per_change=2 → 0.5 meaningful PRs / active contributor / work-day).
  • AUTHOR_ALIASES: start empty; extend per target project.