android-performance-profilinglisted

# Android Performance and Profiling ## Core Principles - Measure first, optimize second. - Fix the highest-impact bottleneck before micro-optimizing. - Prefer repeatable profiling workflows over intuition. - Performance work must preserve correctness and maintainability. - Regressions should be caught with tooling where practical. --- ## What to Measure Focus on the metrics that matter most: | Concern | Typical signal | |---|---| | App startup | cold / warm / hot startup time | | UI smoothness | jank, frame time, skipped frames | | Compose work | unnecessary recomposition, unstable state | | Memory | growth, leaks, churn, GC pressure | | Disk / network | blocking work on main, oversized payloads | | Build/runtime optimization | baseline profile and release performance | --- ## Startup Performance Startup work should be minimal and intentional. Guidelines: - avoid heavy synchronous work in `Application` - defer non-critical initialization - initialize analytics, logging, and background work lazily when possible - do not open database/network just because the app launched unless needed immediately A good startup review asks: - what must happen before first frame? - what can wait until after first draw? - what can wait until the related feature is opened? --- ## Baseline Profiles Use baseline profiles to improve release startup and navigation performance. Guidelines: - generate profiles from critical user journeys - keep them focused on startup and key flows - val