test

Run just test-all and fix failures to 100% completion.

Strategy

1. Run the full gate: just test-all 2>&1 | grep -E "[0-9]+ passed|[0-9]+ failed|^FAILED |^error:|Recipe .* failed|warnings summary" | tail -20 — use timeout: 600000 on the Bash call (the suite takes 5–6 min; the default 2-min Bash timeout kills it).
2. Iterate per-module: when fixing, use just test <module> (e.g. just test routers) for faster turnaround. Plain | tail -20 works fine here — output is short.
3. Stop when green: don't re-run the full suite to "confirm" — once it passes, you're done.

What just test-all actually runs

In order: just test (pytest) → just test-dbt quickbooks → just lint fix → just lint → just format → just typecheck → just cf-lint-all → just lint-actions.

That matters for reading the output: only the first stage produces a pytest summary. A dbt build failure, a ruff/basedpyright failure, or a CloudFormation/workflow lint failure will not produce a failed count — it surfaces as a recipe failure line instead. So a green pytest count alone is not proof the gate passed.

For the code-quality half without the ~6-minute test run, use just test-code — that's the combination the git hooks enforce.

Why the grep

With plain | tail -N the pytest summary scrolls away behind the later stages' output. Each alternative anchors something specific:

• [0-9]+ passed — the summary count (9785 passed,) but NOT test names like test_passed_validation
• [0-9]+ failed — the failure count (1 failed,) but NOT names like test_logs_failed_revocation
• ^FAILED — line-start anchor catches failure headers without matching PASSED lines whose name contains failed
• ^error: and Recipe .* failed — just's recipe-failure lines, which is how the non-pytest stages report
• warnings summary — pytest's warnings-section header

Success = a [0-9]+ passed line, no failed/FAILED, AND no recipe-failure line. All three conditions, not just the first.