每日新聞摘要

Kotlin Weekly #530 rounds up a wide set of items for Kotlin developers, opening with a farewell to Hadi Hariri after nearly seventeen years at JetBrains. The issue links to news on experimental companion blocks and extensions, the State of Kotlin in 2026 report, and a new Agentic Gradle Plugin. It also highlights an article on why “30 characters max” is usually the wrong way to think about length, plus videos on name-based destructuring, Kotlin/Wasm AI apps, and Swift Export. The issue closes with library updates spanning Compose Multiplatform, compiler plugins, and tooling around bundletool.

Kotlin Weekly #530 汇总了本周 Kotlin 生态中的多项内容，开头先向在 JetBrains 任职近十七年的 Hadi Hariri 告别。新闻部分介绍了实验中的 companion blocks 和 companion extensions、2026 Kotlin 现状报告，以及新的 Agentic Gradle Plugin。文章聚焦于为什么“最多 30 个字符”通常是错误的思路；视频则覆盖 name-based destructuring、Kotlin/Wasm AI 应用和 Swift Export。最后还收录了多个库与工具更新，涉及 Compose Multiplatform、编译器插件以及 bundletool 相关集成。

This article introduces cloud storage as a way for photographers to protect their work against drive failure, theft, and ransomware. It frames the topic as a practical guide to evaluating cloud storage options, with attention to backup features, security, and cost. The excerpt does not provide product recommendations or performance claims, but signals that the full piece compares key considerations photographers should review before choosing a storage solution.

文章指出，云存储可以帮助摄影师保护作品，避免硬盘故障、盗窃和勒索软件造成的损失。它将主题定义为一份完整指南，重点在于比较不同云存储方案的备份功能、安全性和成本。摘录没有给出具体产品推荐或性能数据，只表明全文会帮助读者在选择存储方案前评估几个关键因素。

This article takes an honest look at how much obfuscation can really help a banking app. It centers on techniques such as R8, string encryption, and Frida detection, framing them as ways to increase the cost of reverse engineering rather than to provide absolute protection. The piece also emphasizes that obfuscation has clear limits: it may slow attackers down, but it does not eliminate the underlying security risks or guarantee that an app cannot be analyzed.

本文从务实角度审视混淆技术对银行 App 的实际作用，重点围绕 R8、字符串加密和 Frida 检测等手段展开。作者指出，这些做法的目标是提高逆向工程的成本，而不是提供绝对保护。文章同时强调，混淆有其局限性：它也许能拖慢攻击者，但无法消除根本风险，也不能保证应用程序不会被分析。

A five-minute Android Gradle build can turn even a tiny edit into a long interruption. The article points to the common pain of slow builds in multi-module Android apps and frames the problem as one that should be diagnosed rather than solved with a simplistic Gradle tweak. From the excerpt alone, the main takeaway is that build slowness is a practical workflow issue that deserves investigation into its causes before attempting changes.

五分钟的 Android Gradle 构建，就可能让一次小改动变成长时间中断。文章指出，在多模块 Android 应用中，构建缓慢是常见痛点，而真正的问题通常不是靠一个简单的 Gradle 调整就能解决，而是需要先诊断其原因。就现有内容来看，构建变慢是一个会影响开发流程的实际问题，应该先厘清成因再考虑改动。

This entry is only a short teaser that redirects readers to the full article on MobileByteSensei. From the title alone, it highlights an Android debugging pitfall: using adb install -r does not necessarily kill the running app, so any screenshot or observed state may be misleading if the process remains alive. No further technical details are provided in the excerpt.

这段内容只是一个导流性的简短条目，主要引导读者前往 MobileByteSensei 阅读完整文章。仅从标题可知，作者想提醒 Android 开发者：使用 adb install -r 并不一定会杀掉正在运行的 App，因此如果进程仍然存活，看到的截图或状态可能并不真实。节选本身没有提供更多技术细节。

The article describes how a large-scale Android app replaced its local storage foundation with Jetpack DataStore and then verified the outcome with numbers. It emphasizes the migration process and the importance of measuring results instead of relying on assumptions. The piece centers on maintaining performance while changing the storage layer, but the provided excerpt does not include detailed implementation steps or metrics.

这篇文章介绍了如何把一个大型 Android 应用的本地存储基础迁移到 Jetpack DataStore，并用数据验证迁移后的结果。作者强调，这类改造不能只靠感觉判断，而应通过实际测量确认影响。文章核心在于替换存储层时尽量保持性能；但当前提供的内容未包含具体实现步骤或量化指标。

The article is introduced as a comparison between Dart and Kotlin for programmers exploring application development, with the title asking which language is better in 2026. However, the provided content is only a short Medium preview line and does not include the article’s actual analysis, recommendations, or conclusions. Based on the excerpt alone, it only establishes the topic and framing of the comparison, not any factual claims about either language’s capabilities or relative merits.

这篇文章以 Dart 与 Kotlin 的对比为主题，标题直接提出“2026 年谁更好”的问题，面向正在探索应用开发的程序员。不过，现有内容只有 Medium 的简短预览片段，没有提供完整分析、建议或结论。因此，依据这段文字只能确认文章主题和比较框架，无法推断两种语言的能力差异或优劣判断。

This guide explains how Kotlin ADK helps developers build AI agents by abstracting lifecycle, communication, state management, and LLM interaction. It first defines what an AI agent is, then walks through project setup, a simple stateless GreeterAgent, and a stateful content summarizer that stores summaries through tool methods. The article also highlights Kotlin coroutines for concurrent requests, discusses testing and debugging strategies, and outlines deployment options such as JVM servers, Docker, serverless functions, and edge devices. It closes by comparing Kotlin ADK with other agent frameworks and emphasizing Kotlin’s fit for production AI applications.

本文介绍 Kotlin ADK 如何通过抽象生命周期、通信、状态管理和 LLM 交互，帮助开发者构建 AI Agent。文章先解释 AI agent 的概念，再逐步说明项目初始化、一个简单的无状态 GreeterAgent，以及一个可通过工具方法保存摘要的有状态内容摘要器。文中还强调 Kotlin 协程适合处理并发请求，并整理了测试、调试与部署方案，包括 JVM 服务器、Docker、无服务器函数和边缘设备。最后，作者比较了 Kotlin ADK 与其他框架，并指出 Kotlin 在生产级 AI 应用中的适配性。

This article describes how the author built Muffle, an Android app that changes sound profiles based on location without draining the battery. Instead of polling GPS in a background loop, the implementation uses Google Play Services’ GeofencingClient with Geofence objects, a GeofencingRequest, and PendingIntent-triggered callbacks. That architecture keeps the app asleep until a boundary crossing occurs. The author also explains practical limits: indoor GPS is imprecise, requiring larger geofence radii, and Android 12+ background restrictions plus battery optimization can make receivers unreliable. To improve robustness, the author ended up using a foreground service and suggests a hybrid approach with cell tower checks. The main takeaway is to rely on system APIs, design for process death, and keep automation simple, invisible, and battery-conscious.

本文介绍作者打造 Muffle 的过程：这是一款根据位置自动切换音量设置的 Android 应用。为了避免后台轮询 GPS 带来的耗电，作者改用 Google Play Services 的 GeofencingClient，结合 Geofence、GeofencingRequest 和 PendingIntent，在跨越地理围栏时才唤醒应用。文章也说明了几个现实限制：室内和城市环境中的 GPS 漂移很明显，导致地理围栏半径从 50 米不得不调到 150 米；而 Android 12 之后更严格的后台限制和电池优化，也会让 BroadcastReceiver 不够稳定。最终作者使用前台服务提升可靠性，并建议以系统 API 为基础，设计好进程被终止后的恢复机制，必要时加入多重验证来减少误判。

The article argues that offline-first inventory apps are more complicated than they initially appear. What seems like a simple quantity becomes harder to manage once transfers and physical counts overlap, especially in a local-first Android app. The piece frames these challenges as notes from building such an inventory system, highlighting how offline behavior can complicate inventory tracking even when the core idea looks straightforward.

文章指出，离线优先的库存应用往往比表面看起来更复杂。原本看似简单的数量，一旦调拨和实物盘点发生重叠，就会变得更难管理，尤其是在 local-first 的 Android 应用中。作者以构建这类库存系统的笔记形式，强调即使核心概念很直观，离线场景仍会让库存追踪变得棘手。