每日新聞摘要

📝 今日焦點 Android 15 升級與 edge-to-edge 強制改動：將 compileSdk/targetSdk 設為 35 並更新 AGP 與第三方函式庫；系統欄透明化要求使用 WindowInsetsCompat 重構版面以避免 UI 被遮擋；前景服務須正確宣告服務類型並遵守新的逾時/可見性規範；MediaProjection 要求更明確的使用者同意。影響：這是必做的相容性與使用者體驗工作，建議早期優先處理並跨裝置完整測試。 🔧 效能最佳化 建立自訂 Coroutine Dispatcher 以降低延遲：對延遲敏感的任務（音訊、輸入、即時更新）可透過專用或綁定執行緒減少競爭與抖動，但須衡量資源與複雜度；Jetpack Compose 效能升級（摘要截短）值得關注，可能影響重組與繪製成本；UI 元件選擇對效能有實務影響：少量靜態 chip 用 ChipGroup 簡單；大量或頻繁變更者用 RecyclerView 以重用與擴充提升效能。 🏗 架構設計 KMP 多模組與集中化設定：建議以 Convention Plugin 統一共用設定以降低重複與維護成本，適用於大型多模組專案；從 CocoaPods 遷移到 SwiftPM 的實務步驟與風險管理（盤點 Pod、確認 SwiftPM 套件、漸進遷移、CI 驗證）可降低原生相依中斷風險；採用純 Kotlin ViewModel 可提升 KMP 共享邏輯的可測試性與重用性，但需在 DI 與平台端接合層保持輕量；MVVM 與 LifecycleService 有助於面對配置變更與背景工作的可預期性。 ✅ 測試與品質 測試工具選擇與常見錯誤：Fakt（編譯器外掛）於編譯期生成跨平臺、無執行時負擔的假實作，適合需跨目標或偏向 state-based 驗證的專案；傳統 runtime mocking 會導致測試變慢且不支援 Native/Wasm；注意常見 coroutine/Flow 陷阱（文章列出 15 類錯誤）並套用修正建議以避免崩潰；生產審核顯示 Compose 中的重組與 lambda 穩定性問題會在視覺正常時仍引發錯誤，應用 Layout Inspector 與實際運行驗證修復效果。 🚀 語言演進 Kotlin 使用習慣與工具鏈：scope 函式應以「清晰表意」為選擇準則，避免為簡寫犧牲可讀性；Jetpack Compose 作為宣告式 UI 已改變 Android UI 開發範式，需評估學習成本與與現有 XML 的互通策略；工具鏈層面，KSP 在 Kotlin 2.0 的變動暴露出限制，凸顯編譯期解決方案（如 Fakt）與穩健工具支援的重要性。

📝 Top highlights Android 15 migration and mandatory edge-to-edge: set compileSdk/targetSdk to 35, upgrade AGP and audit third-party libraries; WindowInsets handling is now required to avoid UI clipping; foreground services must declare correct types and follow new timeout/visibility rules; MediaProjection demands clearer user consent. Impact: these are mandatory compatibility and UX tasks—prioritize edge-to-edge work early and test across device profiles. 🔧 Performance optimizations Build a custom Coroutine Dispatcher for low-latency workloads: isolating latency-sensitive tasks on dedicated/pinned threads reduces contention and jitter but increases resource use and complexity; watch Jetpack Compose performance upgrades (full details in original article) for changes to recomposition and rendering costs; UI container trade-offs matter—use ChipGroup for a few static chips, RecyclerView for large, dynamic lists to gain recycling and scalability. 🏗 Architecture KMP multi-module and centralized configuration: use a Convention Plugin to consolidate shared settings and reduce duplication in large KMP projects; migrating native iOS/macOS dependencies from CocoaPods to SwiftPM requires inventorying pods, validating SwiftPM availability, adjusting KMP/Gradle, and performing incremental migration with CI to mitigate risk; pure Kotlin ViewModels enable cross-platform business logic reuse but require lightweight platform adapters and careful DI design; MVVM and LifecycleService help retain state across configuration changes and manage background work predictably. ✅ Testing & quality Testing strategy and common pitfalls: Fakt’s compiler-plugin approach generates readable, cross-platform fakes at compile time (no runtime reflection), making it a strong choice when you prefer state-based verification or need MPP support; runtime mocking adds overhead and lacks portability; study the listed 15 coroutine/Flow mistakes and apply fixes to avoid crashes; production audits reveal Compose recomposition and unstable lambda issues that can be diagnosed with Layout Inspector and fixed pragmatically. 🚀 Language and ecosystem Kotlin best practices and tooling trends: choose scope functions only when they clarify intent to preserve readability; Jetpack Compose represents a paradigm shift away from XML that affects APIs, state management, and migration decisions; tooling fragility (e.g., KSP changes with Kotlin 2.0) underscores the value of compile-time solutions and robust toolchain support.