Test Fix

Test Fix

Systematic approach to diagnosing and fixing failing tests.

Purpose

Quickly identify, diagnose, and resolve test failures in the Rust codebase.

Process

Step 1: Identify Failing Tests

# Run tests to see failures
cargo test --all

# Or run specific test suite
cargo test --test integration_memory

Capture:

• Test name
• Failure message
• Stack trace
• Module/file location

Step 2: Reproduce Locally

# Run the specific failing test with output
cargo test test_name -- --exact --nocapture

# With debug logging
RUST_LOG=debug cargo test test_name -- --exact --nocapture

# Force single-threaded for race conditions
cargo test test_name -- --exact --nocapture --test-threads=1

Step 3: Diagnose Root Cause

Common Failure Patterns

Pattern 1: Async/Await Issues

error: future cannot be sent between threads safely

Diagnosis: Missing .await or non-Send type across await Fix:

• Add missing .await
• Use Arc<Mutex<T>> instead of Rc<RefCell<T>>
• Ensure all types are Send + Sync

Pattern 2: Database Errors

error: connection refused / timeout

Diagnosis: Database not available or wrong credentials Fix:

• Check environment variables
• Use test database instance
• Verify connection string

Pattern 3: Race Conditions

assertion failed: expected X but got Y (intermittent)

Diagnosis: Concurrent access without proper synchronization Fix:

• Add proper locking (Mutex, RwLock)
• Use sequential test execution
• Add synchronization barriers

Pattern 4: Type Errors

error: expected `Result<X>`, found `Result<Y>`

Diagnosis: Type mismatch after refactoring Fix:

• Update function signatures
• Add type conversions
• Fix error type propagation

Pattern 5: Lifetime Issues

error: borrowed value does not live long enough

Diagnosis: Reference outlives owner Fix:

• Clone data when needed
• Adjust lifetime parameters
• Use Arc for shared ownership

Step 4: Implement Fix

Example: Fix Missing Await

// BEFORE (failing)
#[tokio::test]
async fn test_episode_creation() {
    let id = memory.start_episode("test", ctx); // Missing .await
    assert!(!id.is_empty());
}

// AFTER (fixed)
#[tokio::test]
async fn test_episode_creation() {
    let id = memory.start_episode("test", ctx).await.unwrap();
    assert!(!id.is_empty());
}

Example: Fix Race Condition

// BEFORE (flaky)
#[tokio::test]
async fn test_concurrent_writes() {
    let memory = create_test_memory().await;

    tokio::join!(
        memory.write_pattern(pattern1),
        memory.write_pattern(pattern2),
    );

    let count = memory.get_pattern_count().await;
    assert_eq!(count, 2); // Sometimes fails!
}

// AFTER (fixed)
#[tokio::test]
async fn test_concurrent_writes() {
    let memory = create_test_memory().await;

    // Use proper synchronization
    memory.write_pattern(pattern1).await.unwrap();
    memory.write_pattern(pattern2).await.unwrap();

    let count = memory.get_pattern_count().await;
    assert_eq!(count, 2); // Reliable
}

Example: Fix Database Connection

// BEFORE (failing)
#[tokio::test]
async fn test_turso_connection() {
    let client = TursoClient::new("hardcoded_url", "token").await;
    // Fails in CI
}

// AFTER (fixed)
#[tokio::test]
async fn test_turso_connection() {
    use std::env;

    let url = env::var("TEST_TURSO_URL")
        .unwrap_or_else(|_| "file:test.db".to_string());
    let token = env::var("TEST_TURSO_TOKEN")
        .unwrap_or_else(|_| "test_token".to_string());

    let client = TursoClient::new(&url, &token).await;
    assert!(client.is_ok());
}

Step 5: Verify Fix

# Run the fixed test multiple times
for i in {1..10}; do
    cargo test test_name -- --exact || break
done

# Run full test suite
cargo test --all

# Run with different thread counts
cargo test --all -- --test-threads=1
cargo test --all -- --test-threads=4

Step 6: Regression Prevention

Add tests for the bug if not covered:

#[tokio::test]
async fn test_regression_issue_123() {
    // Test for the specific bug that was fixed
    let memory = create_test_memory().await;

    // Reproduce the bug scenario
    let result = memory.problematic_operation().await;

    // Ensure it doesn't happen again
    assert!(result.is_ok());
}

Systematic Debugging Checklist

• Run failing test in isolation
• Check test environment (env vars, file paths)
• Review recent changes (git diff)
• Check for missing .await on async calls
• Verify database connections are valid
• Look for race conditions (run with --test-threads=1)
• Check type compatibility after refactoring
• Review error propagation (? operator, unwrap)
• Verify test cleanup (no leftover state)
• Check for platform-specific issues

Tools

1. Cargo Test with Filters

# Run only tests matching pattern
cargo test storage

# Exclude tests
cargo test -- --skip integration

2. Logging

# Debug level
RUST_LOG=debug cargo test

# Trace level (very verbose)
RUST_LOG=trace cargo test

# Specific module
RUST_LOG=memory_core=debug cargo test

3. Backtrace

# Full backtrace on panic
RUST_BACKTRACE=full cargo test

4. Valgrind (memory issues)

# Check for memory leaks
cargo build --tests
valgrind --leak-check=full ./target/debug/deps/test_binary

Batch Fixing Multiple Tests

When multiple tests fail for the same reason:

# List all failing tests
cargo test 2>&1 | grep "test.*FAILED"

# Fix common issue (e.g., missing await)
# Use sed or manual editing

# Verify all fixed
cargo test --all

Documentation

After fixing, update:

1. Test comment: Explain what the test verifies
2. Code comment: Note why the fix was needed
3. Commit message: Reference issue/test name
4. CHANGELOG: If user-facing impact

Example Commit Message

[tests] fix async/await in episode_creation test

- Added missing .await on start_episode call
- Test was panicking with "cannot call async fn without await"
- Added proper error handling with unwrap
- Verified fix runs 10x without failure

Fixes: test_episode_creation

Performance Testing After Fix

# Ensure fix doesn't slow down tests
time cargo test --all

# Compare before/after
# Before: 2.3s
# After: 2.4s (acceptable)

When to Skip vs Fix

Skip (temporarily) if:

• Test requires external service that's down
• Platform-specific issue being investigated
• Known upstream bug

#[tokio::test]
#[ignore = "Requires external Turso instance"]
async fn test_production_sync() {
    // ...
}

Fix immediately if:

• Logic error in code
• Incorrect assertion
• Missing error handling
• Race condition

Escalation

If unable to fix after reasonable effort:

1. Document findings in issue
2. Add #[ignore] with reason
3. Create detailed bug report
4. Request help from goap-agent