# Veza Unified Testing Pipeline (TMT)

> **Philosophy**: Sobriety, Frugality, Accessibility.
> **The Law**: [docs/FRUGALITY.md](../docs/FRUGALITY.md)
> **The Contract**: [docs/BUDGETS.md](../docs/BUDGETS.md)

TMT is the **single entry point** for all Veza tests.
The CI calls `tmt run`, never `go test`, `vitest`, or `cargo test` directly.

## Rules

1.  **Vital tests block everything.** If a tier 1 test fails, the commit is rejected.
2.  **Contractual budgets.** Resource limits from `docs/BUDGETS.md` are enforced by tests.
3.  **No `|| true`.** A test that passes despite regression is worse than no test.
4.  **New frontend tests are legacy by default.** Promote to vital only if it protects a critical invariant.

## Plans

| Plan | Tier | Scope | Blocking | Usage |
|------|------|-------|----------|-------|
| `/vital` | 1 | All components | Yes | Local: `make test-tmt` |
| `/vital-backend` | 1 | Go backend only | Yes | CI parallel job |
| `/vital-frontend` | 1 | Web frontend only | Yes | CI parallel job |
| `/vital-services` | 1 | Rust services only | Yes | CI parallel job |
| `/legacy` | 2 | Slow / integration tests | No (warning) | `tmt --root tmt run plan --name /legacy` |
| `/integration` | 2 | Tests needing infra (DB, Redis) | No (warning) | Requires docker-compose |
| `/nightly` | 3 | E2E Playwright, Storybook | No | Nightly / pre-release |

## Directory Structure

```
tmt/
├── .fmf/version              # FMF tree root
├── plans/
│   ├── vital.fmf             # Tier 1 — all components
│   ├── vital-backend.fmf     # Tier 1 — Go only (CI)
│   ├── vital-frontend.fmf    # Tier 1 — Web only (CI)
│   ├── vital-services.fmf    # Tier 1 — Rust only (CI)
│   ├── legacy.fmf            # Tier 2 — slow / secondary
│   ├── integration.fmf       # Tier 2 — needs infra
│   └── nightly.fmf           # Tier 3 — E2E, storybook
├── tests/
│   ├── backend/              # Go: govulncheck, vet, lint, unit, build, etc.
│   ├── frontend/             # Web: audit, lint, typecheck, unit, build, etc.
│   ├── services/             # Rust: audit, clippy, build, test
│   ├── e2e/                  # Playwright (tier 3)
│   └── storybook/            # Storybook audit (tier 3)
└── README.md
```

## How to Run

```bash
# All vital tests (the standard)
make test-tmt

# Component-specific
make test-tmt-backend
make test-tmt-frontend
make test-tmt-services

# Direct TMT commands
tmt --root tmt run plan --name /vital           # All vital
tmt --root tmt run plan --name /vital-backend   # Backend only
tmt --root tmt run plan --name /integration     # Integration (needs infra)
tmt --root tmt run plan --name /nightly         # E2E + storybook
tmt --root tmt run                              # Everything

# Install TMT
pip install tmt
```

## Test Execution Order

Tests within each component use the `order` attribute for sequencing:

| Order | Phase | Examples |
|-------|-------|---------|
| 10 | Security audits | govulncheck, npm audit, cargo audit |
| 15 | Code generation | Types sync check |
| 20 | Static analysis | vet, lint, format, clippy, core isolation |
| 30 | Type checking | TypeScript typecheck |
| 40 | Unit tests | go test, vitest, cargo test |
| 50 | Build | go build, cargo build, npm run build |
| 60 | Post-build | Bundle size, build perf |

Fast checks fail early. Heavy checks only run if basic hygiene passes.

## Environment Variables

| Variable | Value | Purpose |
|----------|-------|---------|
| `GOMAXPROCS` | 1 | Low-power backend (set in unit.sh) |
| `RUST_BACKTRACE` | 0 | Reduce noise (set in plans) |