Nomos Test Execution Guide

This skill orchestrates test execution for the Nomos monorepo, providing command guidance and verification workflows following the project's testing standards.

When to Use This Skill

•Running the test suite (unit, integration, or all tests)
•Debugging test failures (hermetic vs integration)
•Validating test coverage
•Executing mandatory verification checklists from AGENTS.md files
•Regenerating golden files for parser/compiler tests
•Running race detector
•Testing specific modules independently

Testing Standards Overview

•80% minimum coverage required for all modules
•100% coverage for critical business logic paths
•Table-driven tests as standard pattern
•Integration tests require //go:build integration tag
•Test isolation using t.TempDir()
•Hermetic tests - no network/external deps in unit tests

Quick Command Reference

Basic Test Commands

bash

# Run all tests (unit only, fast)
make test

# Run with race detector
make test-race

# Run unit tests only (explicit)
make test-unit

# Run integration tests only
make test-integration

# Run all tests (unit + integration)
make test-all

# Test specific module
make test-module MODULE=libs/parser
make test-module MODULE=libs/compiler
make test-module MODULE=apps/command-line

# Generate coverage reports
make test-coverage

Build Commands (Required Before Some Tests)

bash

# Build all applications
make build

# Build CLI specifically
make build-cli

# Build test binaries (required for integration tests)
make build-test

Development Commands

bash

# Format code before testing
make fmt

# Tidy dependencies
make mod-tidy

# Lint (requires golangci-lint)
make lint

# Full verification workflow
make fmt && make test && make lint

Test Execution Workflows

Workflow 1: Basic Unit Testing

When: Regular development, fast feedback loop

bash

# 1. Run unit tests
make test

# 2. If failures, run specific module
make test-module MODULE=libs/parser

# 3. Check race conditions if suspicious
make test-race

Expected output:

code

ok      github.com/autonomous-bits/nomos/libs/parser    0.123s  coverage: 92.1% of statements
ok      github.com/autonomous-bits/nomos/libs/compiler  0.456s  coverage: 88.5% of statements

Workflow 2: Pre-Commit Verification

When: Before committing code

bash

# 1. Format code
make fmt

# 2. Run all unit tests
make test

# 3. Run race detector
make test-race

# 4. Lint if available
make lint

Workflow 3: Full Integration Testing

When: Before PR, release preparation, debugging complex issues

bash

# 1. Build test binaries first (critical!)
make build-test

# 2. Run integration tests
make test-integration

# 3. Run all tests (unit + integration)
make test-all

Common mistake: Forgetting make build-test causes integration tests to fail with "binary not found" errors.

Workflow 4: Module-Specific Testing

When: Working on specific module, faster iteration

bash

# Parser module
make test-module MODULE=libs/parser

# Compiler module
make test-module MODULE=libs/compiler

# CLI module
make test-module MODULE=apps/command-line

Workflow 5: Coverage Validation

When: Ensuring coverage requirements met

bash

# 1. Generate coverage report
make test-coverage

# 2. Open HTML report
open coverage.html  # macOS
xdg-open coverage.html  # Linux
start coverage.html  # Windows

# 3. Verify minimum 80% coverage
# Look for files below threshold

Check coverage manually:

bash

go test -coverprofile=coverage.out ./...
go tool cover -func=coverage.out | grep total
# Example output: total: (statements) 88.2%

Debugging Test Failures

Failure Type 1: Unit Test Failures

Symptom: Test fails in regular make test run

Debug steps:

bash

# 1. Run specific test verbosely
cd libs/parser
go test -v -run TestParsing

# 2. Check for table-driven test case
# Look at test output to identify failing case

# 3. Run with race detector
go test -race -run TestParsing

# 4. Check test helper usage
# Ensure helpers marked with t.Helper()

Failure Type 2: Integration Test Failures

Symptom: Test fails only with make test-integration

Debug steps:

bash

# 1. Verify build tags present
grep -r "//go:build integration" libs/compiler/test/

# 2. Ensure test binary built
make build-test
ls -la bin/

# 3. Run integration test verbosely
cd libs/compiler
go test -v -tags=integration ./test/...

# 4. Check for external dependencies
# - Network calls?
# - File system operations?
# - Provider binary execution?

Failure Type 3: Golden File Mismatches (Parser)

Symptom: "output mismatch" errors comparing actual vs golden files

Debug steps:

bash

# 1. Review actual output
cat testdata/fixtures/test.csl
cat testdata/golden/test.csl.json

# 2. If intentional change, regenerate golden files
cd libs/parser
rm testdata/golden/test.csl.json
go test  # Will regenerate

# 3. Review diff before committing
git diff testdata/golden/

# 4. Run tests again to verify
go test

Failure Type 4: Hermetic Test Violations

Symptom: Tests pass locally but fail in CI

Debug steps:

bash

# 1. Check for hermetic violations:
# - Network calls in unit tests
# - Hardcoded paths
# - Time-dependent behavior
# - External dependencies

# 2. Verify test uses t.TempDir()
grep -A 10 "func Test" libs/compiler/*_test.go | grep TempDir

# 3. Check for missing build tags
# Unit tests should NOT have //go:build tags
# Integration tests MUST have //go:build integration

# 4. Isolate test
go test -run TestSpecific -count=1 -v

Failure Type 5: Race Detector Warnings

Symptom: make test-race reports data races

Debug steps:

bash

# 1. Run race detector with failing test
go test -race -run TestConcurrent -v

# 2. Examine race report
# Shows: goroutine stack traces where race occurs

# 3. Common causes:
# - Missing mutex protection
# - Shared map access
# - Provider caching without locks

# 4. Fix and re-run
make test-race

Module-Specific Test Patterns

Parser Module (libs/parser)

Test organization:

code

testdata/
  fixtures/     # Input .csl files
  golden/       # Expected AST outputs
  errors/       # Error test cases

Common commands:

bash

cd libs/parser

# Run all tests
go test ./...

# Run benchmarks
go test -bench=. -benchmem

# Regenerate golden files (after intentional changes)
rm testdata/golden/*.json
go test

# Check coverage
go test -coverprofile=coverage.out ./...
go tool cover -html=coverage.out -o coverage.html

Expected behavior:

•Golden files auto-generate if missing (review before commit!)
•Benchmarks should meet performance goals (see AGENTS.md)
•Coverage should exceed 80%

Compiler Module (libs/compiler)

Test organization:

code

testdata/          # Test fixtures
test/              # Integration tests
testutil/          # Fake providers, test helpers

Common commands:

bash

cd libs/compiler

# Unit tests only
go test -run "^Test" ./...

# Integration tests (hermetic)
make build-test  # Critical first step
go test -tags=integration ./test/...

# All tests
go test ./... && go test -tags=integration ./test/...

# Test provider resolution specifically
go test -v -run TestProvider

Expected behavior:

•Hermetic tests use fake providers from testutil/
•Integration tests require real provider binaries
•Provider process cleanup verified (no zombie processes)

CLI Module (apps/command-line)

Test organization:

code

testdata/          # Test configs
test/              # Integration tests (CLI invocation)
internal/          # Unit tests alongside code

Common commands:

bash

cd apps/command-line

# Build CLI first
make build-cli

# Unit tests
go test ./internal/...

# Integration tests (invoke CLI binary)
make build-test
go test -tags=integration ./test/...

# Test specific command
go test -v -run TestInit ./internal/initcmd/

Expected behavior:

•Integration tests invoke actual CLI binary
•Exit codes verified
•Stdout/stderr captured and validated

Verification Checklists (Mandatory)

Each module has verification requirements in AGENTS.md. Follow these before completing tasks:

Universal Checklist (All Modules)

bash

# ✅ Build Verification
make build
# All code must compile without errors

# ✅ Unit Test Verification
make test
# All existing tests must pass

# ✅ Race Detector
make test-race
# No data races reported

# ✅ Linting (if golangci-lint installed)
make lint
# No errors (warnings acceptable if documented)

# ✅ Coverage Check
make test-coverage
# Minimum 80% overall, 100% for critical paths

Parser Module Checklist

bash

cd libs/parser

# ✅ Build
go build ./...

# ✅ Unit Tests
go test ./...
# All tests pass

# ✅ Race Detector
go test -race ./...

# ✅ Golden Files
git diff testdata/golden/
# Review any changes to golden files

# ✅ Benchmarks
go test -bench=. -run=^$
# Performance meets goals

# ✅ Linting
go vet ./...
golangci-lint run

# ✅ Coverage
go test -coverprofile=coverage.out ./...
go tool cover -func=coverage.out | grep total
# Minimum 80%

Compiler Module Checklist

bash

cd libs/compiler

# ✅ Build
go build ./...

# ✅ Unit Tests
go test ./...

# ✅ Integration Tests
go test -tags=integration ./test/...

# ✅ Race Detector
go test -race ./...

# ✅ Linting
go vet ./...
golangci-lint run

# ✅ Coverage
go test -coverprofile=coverage.out ./...
# Check: >= 80%

CLI Module Checklist

bash

cd apps/command-line

# ✅ Build CLI
make build-cli
./nomos --help
# Binary executes and shows help

# ✅ Unit Tests
go test ./internal/...

# ✅ Integration Tests
make build-test
go test -tags=integration ./test/...

# ✅ Race Detector
go test -race ./...

# ✅ Exit Codes
go test -v -run TestExitCode ./test/...
# Verify: 0 (success), 1 (errors), 2 (usage)

# ✅ Linting
go vet ./...
golangci-lint run

# ✅ Coverage
go test -coverprofile=coverage.out ./...

Common Issues and Solutions

Issue: "Test binary not found"

Solution:

bash

make build-test

Integration tests require pre-built binaries.

Issue: "go.work out of sync"

Solution:

bash

make work-sync
make mod-tidy

Issue: Tests pass locally, fail in CI

Causes:

•Missing integration build tags
•Hermetic test violations (network calls)
•Platform-specific assumptions
•Uncommitted test fixtures

Solution:

bash

# Check build tags
grep -r "//go:build integration" .

# Check for network calls in unit tests
grep -r "http\." *_test.go

# Test in clean environment
git clean -xdf
make test

Issue: Coverage below 80%

Solution:

bash

# Identify uncovered code
go test -coverprofile=coverage.out ./...
go tool cover -func=coverage.out | grep -v "100.0%"

# Add tests for uncovered functions
# Focus on critical paths first (must be 100%)

Issue: Race detector warnings

Solution:

bash

# Run with race detector
go test -race -v ./...

# Fix identified races:
# - Add mutex protection
# - Use channels for coordination
# - Copy data instead of sharing

Best Practices

•
Always run tests before committing:
bash
```
make fmt && make test && make test-race
```
•
Test specific module during development:
bash
```
make test-module MODULE=libs/parser
```
•
Build test binaries before integration tests:
bash
```
make build-test
```
•
Check coverage periodically:
bash
```
make test-coverage
```

•

Run full suite before PR:

bash

make build-test && make test-all && make test-race

•
Review golden file changes carefully:
bash
```
git diff testdata/golden/
```

Reference Documentation

For complete testing guidelines, see: