GrottoCenter / grottocenter-api / 25185951719

87%

push

github

test(perf): parallelize test suite and fix stability (5x speedup)

Replace HTTP-hitting property tests with conventional integration tests
to eliminate socket hang ups and connection corruption. Add a parallel
test runner that shards by domain folder, each shard against its own
database cloned from a pre-seeded PostgreSQL template. The template
rebuilds automatically when fixtures, models, or SQL migrations change.

Stability:
- Replace 33 HTTP property tests with ~70 conventional it() tests
- Reuse Sails HTTP server in supertest (no ephemeral listen/close cycles)
- Fix Cave update after hook (broken Waterline association restore)
- Fix rate limiter tests to reuse connections via supertest.agent()
- Fix blacklist test clock skew (backdate revocation instead of sleeping)
- Fix RecentChangeWindow flaky spy assertion (calledOnce → called)
- Set global Mocha timeout to 10s (was 2s default)

Performance:
- Parallel test runner with folder-based sharding (auto-detected CPUs)
- Template database snapshot with mtime-based staleness detection
- Move auth brute-force delay and argon2 cost to Sails config
- config/env/test.js overrides: delay=0, minimal argon2 cost
- Batch 4 sequential bootstrap SQL queries into one round-trip
- Extract shared fixture order to test/fixtureOrder.js

Runner UX:
- Live progress with percentage, failure count, and boot status
- Press q to abort mid-run with partial summary
- Top 3 slowest tests printed after results
- Shard tmp file paths shown in results table

Housekeeping:
- Add glob and pg to devDependencies (were transitive only)
- Extract DEFAULT_URL to shared test/test-config.js
- Wrap shard cleanup in try/finally to prevent orphaned databases
- Document sequential coverage in README

New commands:
- npm run test              — parallel with auto-snapshot (default)
- npm run test:sequential   — single-process fallback
- npm run test:snapshot     — manually rebuild the template

2948 of 3564 branches covered (82.72%)

Branch coverage included in aggregate %.

4 of 5 new or added lines in 2 files covered. (80.0%)

6149 of 6974 relevant lines covered (88.17%)

51.75 hits per line