performance-check

Use a structured checklist covering N+1 queries, queries in loops, missing pagination, goroutine leaks, and unbounded caches. Then validate with go test -bench for benchmarks and pprof for CPU and memory profiling.

Use GORM's Preload to eager load associations, reducing N+1 queries to two queries total. Also replace per-item queries inside loops with a single Where id IN clause, and add indexes on frequently filtered columns.

Apply React.memo to components that re-render unnecessarily when parent state changes, and useMemo for expensive computations like date formatting. For lists over 100 items, use virtualization with react-window instead of memoization alone.

Compare runtime.NumGoroutine before and after test execution to spot leaks. Prevent them by passing context to producer goroutines, calling cancel via defer, and building with the -race flag to catch race conditions.

Apache Bench (ab -n 1000 -c 10) handles simple load tests, while k6 scripts support virtual users, duration control, and response time assertions. Both verify endpoints meet latency targets like sub-500ms responses.

Skip the full review for simple configuration changes that do not touch queries, APIs, rendering, caching, or algorithms. The checklist targets changes with real performance impact, not cosmetic or config-only edits.