/radar

Async Testing Patterns

Purpose: Use this when the target code depends on async scheduling, concurrency, streams, channels, or timeouts.

Contents:

• JS / TS async rules
• Python async rules
• Go concurrency rules
• Rust async rules
• Timeout strategy
• Race detection

TypeScript / JavaScript

Rules:

• Prefer async/await plus explicit await.
• Use findBy* or waitFor for eventual UI states.
• Use fake timers for debounce, retries, and scheduled behavior.
• Use Promise.allSettled tests when partial success matters.

Python

Rules:

• Use pytest.mark.asyncio for coroutine tests.
• Use httpx.AsyncClient or framework-native async clients for HTTP boundaries.
• Apply per-test timeout markers to slow integration paths.

Go

Rules:

• Use t.Parallel() only when shared state is fully isolated.
• Prefer channels, contexts, and WaitGroup over sleeps.
• Always run go test -race in CI for concurrent code.

Must-keep commands:

go test -race ./...
go test -race -coverprofile=coverage.out ./...
go test -timeout 60s ./...

Rust

Rules:

• Use #[tokio::test] for async code.
• Use flavor = "multi_thread" only when concurrency itself is under test.
• Use tokio::time::timeout for timeout paths and channel-based assertions for ordering.

Timeout Strategy

Framework notes:

• Vitest: testTimeout, hookTimeout, teardownTimeout
• pytest: timeout = 30, override per test when needed
• Go: CLI -timeout
• Rust nextest: slow-timeout

Race Condition Detection & Prevention

Common Symptoms

Prevention Rules

• Replace sleeps with synchronization.
• Freeze or fake time where possible.
• Keep state local to each test.
• Repeat async fixes 10+ times when validating stability.

Quick Reference