<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>weflowly</title>
  
  <subtitle>weflowly</subtitle>
  <link href="https://weflowly.com/atom.xml" rel="self"/>
  
  <link href="https://weflowly.com/"/>
  <updated>2026-03-20T14:11:15.449Z</updated>
  <id>https://weflowly.com/</id>
  
  <author>
    <name>weflowly</name>
    
  </author>
  
  <generator uri="https://hexo.io/">Hexo</generator>
  
  <entry>
    <title>在 Next.js 应用中集成 Zig WebAssembly 模块对 Algolia 搜索结果进行实时客户端重排序</title>
    <link href="https://weflowly.com/3772291175/"/>
    <id>https://weflowly.com/3772291175/</id>
    <published>2023-11-15T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;项目初期，Algolia 提供了无与伦比的搜索性能，毫秒级的响应速度让用户体验极为流畅。我们的技术栈是经典的组合：后端使用 Ruby on Rails 管理数据和业务逻辑，并负责将数据同步至 Algolia；前端则采用 Next.js</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="Next.js" scheme="https://weflowly.com/tags/Next-js/"/>
    
    <category term="Ruby" scheme="https://weflowly.com/tags/Ruby/"/>
    
    <category term="Zig" scheme="https://weflowly.com/tags/Zig/"/>
    
    <category term="Algolia" scheme="https://weflowly.com/tags/Algolia/"/>
    
    <category term="WebAssembly" scheme="https://weflowly.com/tags/WebAssembly/"/>
    
  </entry>
  
  <entry>
    <title>构建基于网关与Delta Lake的LLM应用端到端身份感知数据访问架构</title>
    <link href="https://weflowly.com/8836674285/"/>
    <id>https://weflowly.com/8836674285/</id>
    <published>2023-11-15T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;在构建企业级检索增强生成（RAG）应用时，一个无法回避的核心问题是：如何在一个共享的、包含敏感信息的数据源（如Delta</summary>
        
      
    
    
    
    <category term="架构与设计" scheme="https://weflowly.com/categories/%E6%9E%B6%E6%9E%84%E4%B8%8E%E8%AE%BE%E8%AE%A1/"/>
    
    
    <category term="LangChain" scheme="https://weflowly.com/tags/LangChain/"/>
    
    <category term="IAM" scheme="https://weflowly.com/tags/IAM/"/>
    
    <category term="Delta Lake" scheme="https://weflowly.com/tags/Delta-Lake/"/>
    
    <category term="网关与代理" scheme="https://weflowly.com/tags/%E7%BD%91%E5%85%B3%E4%B8%8E%E4%BB%A3%E7%90%86/"/>
    
  </entry>
  
  <entry>
    <title>使用Go与Python在Azure Functions上为iOS应用构建TDD驱动的混合语言处理管道</title>
    <link href="https://weflowly.com/1657794845/"/>
    <id>https://weflowly.com/1657794845/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;我们的iOS应用面临一个棘手的挑战：用户需要上传大型音频文件进行分析，这个分析过程计算密集且耗时。最初采用单一的Python Azure</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="iOS 开发" scheme="https://weflowly.com/tags/iOS-%E5%BC%80%E5%8F%91/"/>
    
    <category term="Python" scheme="https://weflowly.com/tags/Python/"/>
    
    <category term="Go" scheme="https://weflowly.com/tags/Go/"/>
    
    <category term="Azure Functions" scheme="https://weflowly.com/tags/Azure-Functions/"/>
    
    <category term="TDD" scheme="https://weflowly.com/tags/TDD/"/>
    
  </entry>
  
  <entry>
    <title>构建基于 API Gateway 的声明式 Saga 事务协调器</title>
    <link href="https://weflowly.com/1125855315/"/>
    <id>https://weflowly.com/1125855315/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;h3 id=&quot;一、问题的定义：微服务架构中的一致性难题&quot;&gt;&lt;a href=&quot;#一、问题的定义：微服务架构中的一致性难题&quot; class=&quot;headerlink&quot;</summary>
        
      
    
    
    
    <category term="分布式架构" scheme="https://weflowly.com/categories/%E5%88%86%E5%B8%83%E5%BC%8F%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="分布式事务" scheme="https://weflowly.com/tags/%E5%88%86%E5%B8%83%E5%BC%8F%E4%BA%8B%E5%8A%A1/"/>
    
    <category term="API Gateway" scheme="https://weflowly.com/tags/API-Gateway/"/>
    
    <category term="RESTful API" scheme="https://weflowly.com/tags/RESTful-API/"/>
    
    <category term="UnoCSS" scheme="https://weflowly.com/tags/UnoCSS/"/>
    
  </entry>
  
  <entry>
    <title>利用Elixir和LevelDB构建从Fluentd到数据仓库的弹性日志摄取层</title>
    <link href="https://weflowly.com/2004215401/"/>
    <id>https://weflowly.com/2004215401/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;在处理分布式系统中海量、突发性的日志流时，核心挑战在于构建一个既能满足实时观测需求，又能保证数据最终无损落盘至数据仓库的摄取层。单纯依赖消息队列（如Kafka）和流处理框架（如Spark/Flink）的组合虽然功能强大，但在某些场景下显得过于笨重，运维成本高昂。尤其是在需要</summary>
        
      
    
    
    
    <category term="数据工程" scheme="https://weflowly.com/categories/%E6%95%B0%E6%8D%AE%E5%B7%A5%E7%A8%8B/"/>
    
    
    <category term="Elixir" scheme="https://weflowly.com/tags/Elixir/"/>
    
    <category term="LevelDB" scheme="https://weflowly.com/tags/LevelDB/"/>
    
    <category term="Fluentd" scheme="https://weflowly.com/tags/Fluentd/"/>
    
    <category term="Solid.js" scheme="https://weflowly.com/tags/Solid-js/"/>
    
    <category term="Data Warehouse" scheme="https://weflowly.com/tags/Data-Warehouse/"/>
    
  </entry>
  
  <entry>
    <title>构建基于向量检索与消息队列的移动端CI/CD日志分析架构</title>
    <link href="https://weflowly.com/2273497050/"/>
    <id>https://weflowly.com/2273497050/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;当团队的移动端CI/CD流水线每天执行上千次构建时，日志系统就从一个辅助工具变成了瓶颈本身。传统的基于文本的搜索，无论是&lt;code&gt;grep&lt;/code&gt;还是Elasticsearch，在面对语义层面相似但具体错误信息不同的问题时，都显得力不从心。一个新入职的工程师可能会花</summary>
        
      
    
    
    
    <category term="架构设计" scheme="https://weflowly.com/categories/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/"/>
    
    
    <category term="Pinecone" scheme="https://weflowly.com/tags/Pinecone/"/>
    
    <category term="消息队列" scheme="https://weflowly.com/tags/%E6%B6%88%E6%81%AF%E9%98%9F%E5%88%97/"/>
    
    <category term="CI/CD for Mobile" scheme="https://weflowly.com/tags/CI-CD-for-Mobile/"/>
    
    <category term="Next.js" scheme="https://weflowly.com/tags/Next-js/"/>
    
    <category term="Micro-frontends" scheme="https://weflowly.com/tags/Micro-frontends/"/>
    
  </entry>
  
  <entry>
    <title>为 LangChain 应用构建基于 Firestore 与 InfluxDB 的双核可观测性系统</title>
    <link href="https://weflowly.com/2311131515/"/>
    <id>https://weflowly.com/2311131515/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;任何运行在生产环境的 LangChain 应用，尤其是涉及 Agent 或复杂 Chain 的系统，很快就会变成一个难以理解的黑盒。当用户抱怨响应缓慢时，瓶颈是 LLM 调用、工具执行还是数据解析？当运营部门质询成本时，哪个用户的哪类请求消耗了最多的</summary>
        
      
    
    
    
    <category term="可观测性" scheme="https://weflowly.com/categories/%E5%8F%AF%E8%A7%82%E6%B5%8B%E6%80%A7/"/>
    
    
    <category term="UnoCSS" scheme="https://weflowly.com/tags/UnoCSS/"/>
    
    <category term="LangChain" scheme="https://weflowly.com/tags/LangChain/"/>
    
    <category term="可观测性" scheme="https://weflowly.com/tags/%E5%8F%AF%E8%A7%82%E6%B5%8B%E6%80%A7/"/>
    
    <category term="Firestore" scheme="https://weflowly.com/tags/Firestore/"/>
    
    <category term="InfluxDB" scheme="https://weflowly.com/tags/InfluxDB/"/>
    
  </entry>
  
  <entry>
    <title>基于 gRPC-Go 与 Paxos 构建分布式 Emotion 原子化 CSS 编译服务</title>
    <link href="https://weflowly.com/2354812769/"/>
    <id>https://weflowly.com/2354812769/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;在大型微前端架构中，当数十个团队并行开发并部署组件时，确保构建产物的一致性成为一个棘手的挑战。我们遇到的一个具体问题源于 CSS-in-JS 库 &lt;code&gt;Emotion&lt;/code&gt;。&lt;code&gt;Emotion&lt;/code&gt;</summary>
        
      
    
    
    
    <category term="分布式系统" scheme="https://weflowly.com/categories/%E5%88%86%E5%B8%83%E5%BC%8F%E7%B3%BB%E7%BB%9F/"/>
    
    
    <category term="Emotion" scheme="https://weflowly.com/tags/Emotion/"/>
    
    <category term="Paxos 算法" scheme="https://weflowly.com/tags/Paxos-%E7%AE%97%E6%B3%95/"/>
    
    <category term="gRPC-Go" scheme="https://weflowly.com/tags/gRPC-Go/"/>
    
    <category term="分布式一致性" scheme="https://weflowly.com/tags/%E5%88%86%E5%B8%83%E5%BC%8F%E4%B8%80%E8%87%B4%E6%80%A7/"/>
    
    <category term="内部开发者平台" scheme="https://weflowly.com/tags/%E5%86%85%E9%83%A8%E5%BC%80%E5%8F%91%E8%80%85%E5%B9%B3%E5%8F%B0/"/>
    
  </entry>
  
  <entry>
    <title>在 Vercel Functions 中利用 Babel 动态转译实现一个轻量级 GraphQL Feature Store</title>
    <link href="https://weflowly.com/2963706721/"/>
    <id>https://weflowly.com/2963706721/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;团队最近的一个项目需要在边缘节点上进行实时用户意图预测，这是一个典型的 ML 应用场景。传统的做法是部署一个独立的、通常是基于 Python 的模型服务，并连接到一个中心化的 Feature Store。但在我们这个场景下，整个业务栈是构建在 Vercel 上的</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="Babel" scheme="https://weflowly.com/tags/Babel/"/>
    
    <category term="Vercel Functions" scheme="https://weflowly.com/tags/Vercel-Functions/"/>
    
    <category term="GraphQL" scheme="https://weflowly.com/tags/GraphQL/"/>
    
    <category term="Feature Store" scheme="https://weflowly.com/tags/Feature-Store/"/>
    
  </entry>
  
  <entry>
    <title>为Elixir与Angular混合型Monorepo构建增量式Jenkins CI/CD流水线</title>
    <link href="https://weflowly.com/3542883442/"/>
    <id>https://weflowly.com/3542883442/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;我们的项目迁移到 Monorepo 架构后面临的第一个、也是最尖锐的痛点，就是 CI/CD 流水线的执行效率。最初的 &lt;code&gt;Jenkinsfile&lt;/code&gt; 简单粗暴，任何一次提交，无论改动的是 Elixir 后端的某个 Phoenix Context，还是</summary>
        
      
    
    
    
    <category term="DevOps" scheme="https://weflowly.com/categories/DevOps/"/>
    
    
    <category term="Elixir" scheme="https://weflowly.com/tags/Elixir/"/>
    
    <category term="Jenkins" scheme="https://weflowly.com/tags/Jenkins/"/>
    
    <category term="Angular" scheme="https://weflowly.com/tags/Angular/"/>
    
    <category term="Apollo Client" scheme="https://weflowly.com/tags/Apollo-Client/"/>
    
    <category term="CI/CD" scheme="https://weflowly.com/tags/CI-CD/"/>
    
    <category term="Monorepo" scheme="https://weflowly.com/tags/Monorepo/"/>
    
  </entry>
  
  <entry>
    <title>构建基于 Serverless 与 Python 的可观测数据仓库ETL管道：SkyWalking 与 OpenTelemetry 集成实践</title>
    <link href="https://weflowly.com/4024044773/"/>
    <id>https://weflowly.com/4024044773/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;数据仓库里的一条脏数据，反向追溯其ETL链路花了我们整整两天。问题出在一个由多个AWS Lambda函数组成的Serverless数据管道上，每个函数处理一个阶段：验证、扩充、加载。当数据量达到每日千万级别时，通过CloudWatch</summary>
        
      
    
    
    
    <category term="可观测性" scheme="https://weflowly.com/categories/%E5%8F%AF%E8%A7%82%E6%B5%8B%E6%80%A7/"/>
    
    
    <category term="Python" scheme="https://weflowly.com/tags/Python/"/>
    
    <category term="Data Warehouse" scheme="https://weflowly.com/tags/Data-Warehouse/"/>
    
    <category term="SkyWalking" scheme="https://weflowly.com/tags/SkyWalking/"/>
    
    <category term="Serverless" scheme="https://weflowly.com/tags/Serverless/"/>
    
    <category term="OpenTelemetry" scheme="https://weflowly.com/tags/OpenTelemetry/"/>
    
  </entry>
  
  <entry>
    <title>基于Tornado与LevelDB构建低延迟推荐系统实时特征节点及Loki可观测性实践</title>
    <link href="https://weflowly.com/4029139727/"/>
    <id>https://weflowly.com/4029139727/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;推荐服务对P99响应延迟的要求被压缩到了20ms以内，但链路分析显示，对远程集中式缓存（如Redis</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="LevelDB" scheme="https://weflowly.com/tags/LevelDB/"/>
    
    <category term="Tornado" scheme="https://weflowly.com/tags/Tornado/"/>
    
    <category term="推荐系统" scheme="https://weflowly.com/tags/%E6%8E%A8%E8%8D%90%E7%B3%BB%E7%BB%9F/"/>
    
    <category term="Loki" scheme="https://weflowly.com/tags/Loki/"/>
    
    <category term="实时特征存储" scheme="https://weflowly.com/tags/%E5%AE%9E%E6%97%B6%E7%89%B9%E5%BE%81%E5%AD%98%E5%82%A8/"/>
    
  </entry>
  
  <entry>
    <title>整合 Spring Boot Kafka 与 SciPy 构建异构实时特征计算架构</title>
    <link href="https://weflowly.com/4130562963/"/>
    <id>https://weflowly.com/4130562963/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;一个常见的需求是为下游的风控或推荐系统提供实时的用户行为特征，例如“用户过去5分钟内的点击速率”、“交易金额的滚动标准差”等。当核心业务系统由Spring</summary>
        
      
    
    
    
    <category term="数据工程" scheme="https://weflowly.com/categories/%E6%95%B0%E6%8D%AE%E5%B7%A5%E7%A8%8B/"/>
    
    
    <category term="Ruby" scheme="https://weflowly.com/tags/Ruby/"/>
    
    <category term="Kafka" scheme="https://weflowly.com/tags/Kafka/"/>
    
    <category term="Spring Boot" scheme="https://weflowly.com/tags/Spring-Boot/"/>
    
    <category term="SciPy" scheme="https://weflowly.com/tags/SciPy/"/>
    
    <category term="架构设计" scheme="https://weflowly.com/tags/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/"/>
    
  </entry>
  
  <entry>
    <title>构建后端驱动的动态UI渲染引擎 Java定义结构与Chakra UI实现渲染</title>
    <link href="https://weflowly.com/4252096943/"/>
    <id>https://weflowly.com/4252096943/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;运营和产品团队最常提的需求之一，就是希望能“动态调整”页面布局。今天想在首页加一个营销卡片，明天想把A/B测试的两个区块对调一下。如果每次这种变更都需要前端走完整的开发、测试、发版流程，响应速度根本无法满足业务需求。这个痛点催生了一个长期的技术思考：能否将UI的“结构”与“</summary>
        
      
    
    
    
    <category term="全栈架构" scheme="https://weflowly.com/categories/%E5%85%A8%E6%A0%88%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="架构设计" scheme="https://weflowly.com/tags/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/"/>
    
    <category term="Java" scheme="https://weflowly.com/tags/Java/"/>
    
    <category term="Chakra UI" scheme="https://weflowly.com/tags/Chakra-UI/"/>
    
    <category term="JavaScript" scheme="https://weflowly.com/tags/JavaScript/"/>
    
    <category term="Sass/SCSS" scheme="https://weflowly.com/tags/Sass-SCSS/"/>
    
  </entry>
  
  <entry>
    <title>基于esbuild插件与容器编排实现前端应用的动态WAF规则注入</title>
    <link href="https://weflowly.com/5015173620/"/>
    <id>https://weflowly.com/5015173620/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;通用型WAF（Web Application Firewall）规则，例如允许所有对 &lt;code&gt;/api/*&lt;/code&gt;</summary>
        
      
    
    
    
    <category term="DevSecOps" scheme="https://weflowly.com/categories/DevSecOps/"/>
    
    
    <category term="WAF" scheme="https://weflowly.com/tags/WAF/"/>
    
    <category term="esbuild" scheme="https://weflowly.com/tags/esbuild/"/>
    
    <category term="容器编排" scheme="https://weflowly.com/tags/%E5%AE%B9%E5%99%A8%E7%BC%96%E6%8E%92/"/>
    
    <category term="Redis" scheme="https://weflowly.com/tags/Redis/"/>
    
    <category term="Frontend" scheme="https://weflowly.com/tags/Frontend/"/>
    
  </entry>
  
  <entry>
    <title>使用 C#、Redis Streams 和 Datadog 构建可观测的 LLM 推理任务管道</title>
    <link href="https://weflowly.com/4784022026/"/>
    <id>https://weflowly.com/4784022026/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;一个看似简单的同步 LLM 调用是许多生产事故的开端。&lt;/p&gt;
&lt;pre class=&quot;line-numbers language-csharp&quot; data-language=&quot;csharp&quot;&gt;&lt;code class=&quot;language-csharp&quot;&gt;&lt;span</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="可观测性" scheme="https://weflowly.com/tags/%E5%8F%AF%E8%A7%82%E6%B5%8B%E6%80%A7/"/>
    
    <category term="LLM" scheme="https://weflowly.com/tags/LLM/"/>
    
    <category term="C#" scheme="https://weflowly.com/tags/C/"/>
    
    <category term="Redis Streams" scheme="https://weflowly.com/tags/Redis-Streams/"/>
    
    <category term="Datadog" scheme="https://weflowly.com/tags/Datadog/"/>
    
    <category term="微服务" scheme="https://weflowly.com/tags/%E5%BE%AE%E6%9C%8D%E5%8A%A1/"/>
    
  </entry>
  
  <entry>
    <title>在 GKE 上利用 Consul Session 实现 CDC 数据处理管道的幂等消费与分片协调</title>
    <link href="https://weflowly.com/5622061643/"/>
    <id>https://weflowly.com/5622061643/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;h3 id=&quot;技术痛点：从一个失控的数据管道开始&quot;&gt;&lt;a href=&quot;#技术痛点：从一个失控的数据管道开始&quot; class=&quot;headerlink&quot;</summary>
        
      
    
    
    
    <category term="云原生" scheme="https://weflowly.com/categories/%E4%BA%91%E5%8E%9F%E7%94%9F/"/>
    
    
    <category term="Go" scheme="https://weflowly.com/tags/Go/"/>
    
    <category term="GCP GKE" scheme="https://weflowly.com/tags/GCP-GKE/"/>
    
    <category term="Consul" scheme="https://weflowly.com/tags/Consul/"/>
    
    <category term="数据处理" scheme="https://weflowly.com/tags/%E6%95%B0%E6%8D%AE%E5%A4%84%E7%90%86/"/>
    
    <category term="分布式锁" scheme="https://weflowly.com/tags/%E5%88%86%E5%B8%83%E5%BC%8F%E9%94%81/"/>
    
  </entry>
  
  <entry>
    <title>构建从 Micronaut gRPC 服务到 SwiftUI 客户端的自动化协议驱动开发管道</title>
    <link href="https://weflowly.com/5476234317/"/>
    <id>https://weflowly.com/5476234317/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;跨语言、跨平台的技术栈协作中，最大的摩擦力源于客户端与服务端之间的契约同步。在 SwiftUI 原生应用与 Micronaut 后端的组合中，gRPC 凭借 Protobuf 提供了强类型的契约定义，但这仅仅是起点。当团队规模扩大、迭代速度加快时，手动管理</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="SwiftUI" scheme="https://weflowly.com/tags/SwiftUI/"/>
    
    <category term="Micronaut" scheme="https://weflowly.com/tags/Micronaut/"/>
    
    <category term="gRPC" scheme="https://weflowly.com/tags/gRPC/"/>
    
    <category term="构建与工具" scheme="https://weflowly.com/tags/%E6%9E%84%E5%BB%BA%E4%B8%8E%E5%B7%A5%E5%85%B7/"/>
    
    <category term="云服务商" scheme="https://weflowly.com/tags/%E4%BA%91%E6%9C%8D%E5%8A%A1%E5%95%86/"/>
    
  </entry>
  
  <entry>
    <title>构建高可用Memcached客户端在Spring Boot与Quarkus间的架构权衡与实现</title>
    <link href="https://weflowly.com/5895127070/"/>
    <id>https://weflowly.com/5895127070/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;定义一个生产级的缓存客户端，其需求远不止&lt;code&gt;get&lt;/code&gt;和&lt;code&gt;set&lt;/code&gt;。在一个复杂的分布式系统中，缓存层往往是性能的基石，也同样是故障的放大器。当一个Memcached节点由于网络分区或自身崩溃而无响应时，单纯的超时机制会导致大量请求线程</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="Spring Boot" scheme="https://weflowly.com/tags/Spring-Boot/"/>
    
    <category term="微服务" scheme="https://weflowly.com/tags/%E5%BE%AE%E6%9C%8D%E5%8A%A1/"/>
    
    <category term="Memcached" scheme="https://weflowly.com/tags/Memcached/"/>
    
    <category term="Quarkus" scheme="https://weflowly.com/tags/Quarkus/"/>
    
    <category term="高可用" scheme="https://weflowly.com/tags/%E9%AB%98%E5%8F%AF%E7%94%A8/"/>
    
  </entry>
  
  <entry>
    <title>ScyllaDB与Trino联邦查询构建实时特征分析平台的架构决策</title>
    <link href="https://weflowly.com/6827062027/"/>
    <id>https://weflowly.com/6827062027/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;我们面临一个具体的工程挑战：一个高并发的AI推荐服务，每秒产生数十万次用户行为事件。这些事件必须以极低的延迟写入数据库，用于实时用户画像的更新与读取，这是典型的OLTP场景。同时，数据科学家和运营团队需要一个交互式仪表盘，对这些最近几小时甚至几天内的原始事件数据进行探索性分</summary>
        
      
    
    
    
    <category term="数据架构" scheme="https://weflowly.com/categories/%E6%95%B0%E6%8D%AE%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="架构设计" scheme="https://weflowly.com/tags/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/"/>
    
    <category term="SwiftUI" scheme="https://weflowly.com/tags/SwiftUI/"/>
    
    <category term="ScyllaDB" scheme="https://weflowly.com/tags/ScyllaDB/"/>
    
    <category term="Trino" scheme="https://weflowly.com/tags/Trino/"/>
    
    <category term="数据工程" scheme="https://weflowly.com/tags/%E6%95%B0%E6%8D%AE%E5%B7%A5%E7%A8%8B/"/>
    
  </entry>
  
  <entry>
    <title>基于 Kong、Valtio 和动态样式注入的微前端架构决策与实现</title>
    <link href="https://weflowly.com/7723048186/"/>
    <id>https://weflowly.com/7723048186/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;企业内部的运营中台项目，其复杂性通常会随着业务线的扩张而指数级增长。当不同团队负责的模块（如订单管理、用户中心、数据报表）被强制捆绑在同一个单体前端应用中时，技术栈统一的优势很快就会被协作成本、部署瓶셔和代码库膨胀等问题所抵消。我们面临的正是这样一个局面：一个庞大的</summary>
        
      
    
    
    
    <category term="架构设计" scheme="https://weflowly.com/categories/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/"/>
    
    
    <category term="Valtio" scheme="https://weflowly.com/tags/Valtio/"/>
    
    <category term="样式方案" scheme="https://weflowly.com/tags/%E6%A0%B7%E5%BC%8F%E6%96%B9%E6%A1%88/"/>
    
    <category term="Kong" scheme="https://weflowly.com/tags/Kong/"/>
    
    <category term="微前端" scheme="https://weflowly.com/tags/%E5%BE%AE%E5%89%8D%E7%AB%AF/"/>
    
    <category term="架构权衡" scheme="https://weflowly.com/tags/%E6%9E%B6%E6%9E%84%E6%9D%83%E8%A1%A1/"/>
    
  </entry>
  
  <entry>
    <title>构建基于强化学习的动态UI自适应系统：从WebSocket向量化状态到PostCSS原子化执行的全链路追踪实践</title>
    <link href="https://weflowly.com/9278529254/"/>
    <id>https://weflowly.com/9278529254/</id>
    <published>2023-10-27T10:30:00.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;A/B</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="WebSockets" scheme="https://weflowly.com/tags/WebSockets/"/>
    
    <category term="强化学习" scheme="https://weflowly.com/tags/%E5%BC%BA%E5%8C%96%E5%AD%A6%E4%B9%A0/"/>
    
    <category term="Jaeger" scheme="https://weflowly.com/tags/Jaeger/"/>
    
    <category term="PostCSS" scheme="https://weflowly.com/tags/PostCSS/"/>
    
    <category term="向量" scheme="https://weflowly.com/tags/%E5%90%91%E9%87%8F/"/>
    
  </entry>
  
  <entry>
    <title>构建基于Jib与Rome的TDD驱动型多语言微服务本地开发环境</title>
    <link href="https://weflowly.com/8090063457/"/>
    <id>https://weflowly.com/8090063457/</id>
    <published>2023-10-27T10:15:32.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;团队内部的多语言（Polyglot）微服务实践已经有一段时间，一个常见的组合是后端使用稳定的JVM语言（如Java/Kotlin）处理核心业务逻辑，而BFF（Backend for</summary>
        
      
    
    
    
    <category term="架构与设计" scheme="https://weflowly.com/categories/%E6%9E%B6%E6%9E%84%E4%B8%8E%E8%AE%BE%E8%AE%A1/"/>
    
    
    <category term="TDD" scheme="https://weflowly.com/tags/TDD/"/>
    
    <category term="微服务" scheme="https://weflowly.com/tags/%E5%BE%AE%E6%9C%8D%E5%8A%A1/"/>
    
    <category term="NoSQL" scheme="https://weflowly.com/tags/NoSQL/"/>
    
    <category term="Jib" scheme="https://weflowly.com/tags/Jib/"/>
    
    <category term="Rome" scheme="https://weflowly.com/tags/Rome/"/>
    
  </entry>
  
  <entry>
    <title>构建基于 Jenkins API 的 UI 组件库自动化发布平台</title>
    <link href="https://weflowly.com/5148120695/"/>
    <id>https://weflowly.com/5148120695/</id>
    <published>2023-10-27T10:15:30.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;团队维护的 UI 组件库最初依赖开发者手动执行 &lt;code&gt;npm publish&lt;/code&gt;，这是一个脆弱且混乱的流程。版本号管理不规范、changelog</summary>
        
      
    
    
    
    <category term="DevOps 实战" scheme="https://weflowly.com/categories/DevOps-%E5%AE%9E%E6%88%98/"/>
    
    
    <category term="内部开发者平台" scheme="https://weflowly.com/tags/%E5%86%85%E9%83%A8%E5%BC%80%E5%8F%91%E8%80%85%E5%B9%B3%E5%8F%B0/"/>
    
    <category term="Jenkins" scheme="https://weflowly.com/tags/Jenkins/"/>
    
    <category term="UI 组件库" scheme="https://weflowly.com/tags/UI-%E7%BB%84%E4%BB%B6%E5%BA%93/"/>
    
    <category term="Web API" scheme="https://weflowly.com/tags/Web-API/"/>
    
    <category term="自动化" scheme="https://weflowly.com/tags/%E8%87%AA%E5%8A%A8%E5%8C%96/"/>
    
  </entry>
  
  <entry>
    <title>构建基于 CDC 和 LLM 的 Node.js 实时数据管道同步 Meilisearch 索引</title>
    <link href="https://weflowly.com/6581440939/"/>
    <id>https://weflowly.com/6581440939/</id>
    <published>2023-10-27T10:15:30.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;项目的搜索功能需求从简单的关键词匹配演变成了语义理解。这意味着我们不能再依赖数据库的 &lt;code&gt;LIKE&lt;/code&gt; 查询。引入全文搜索引擎是必然选择，我们看中了 Meilisearch</summary>
        
      
    
    
    
    <category term="后端架构" scheme="https://weflowly.com/categories/%E5%90%8E%E7%AB%AF%E6%9E%B6%E6%9E%84/"/>
    
    
    <category term="Kafka" scheme="https://weflowly.com/tags/Kafka/"/>
    
    <category term="LLM" scheme="https://weflowly.com/tags/LLM/"/>
    
    <category term="Node.js" scheme="https://weflowly.com/tags/Node-js/"/>
    
    <category term="Meilisearch" scheme="https://weflowly.com/tags/Meilisearch/"/>
    
    <category term="ORM" scheme="https://weflowly.com/tags/ORM/"/>
    
    <category term="CDC" scheme="https://weflowly.com/tags/CDC/"/>
    
    <category term="Prisma" scheme="https://weflowly.com/tags/Prisma/"/>
    
  </entry>
  
  <entry>
    <title>构建基于 Flux CD 的 WAF 规则 GitOps 交付流水线与单元测试实践</title>
    <link href="https://weflowly.com/0866352593/"/>
    <id>https://weflowly.com/0866352593/</id>
    <published>2023-10-27T10:15:28.000Z</published>
    <updated>2026-03-20T14:11:15.445Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;一次线上故障的起因，仅仅是 WAF 自定义规则中的一个正则表达式打错了括号。手动变更，没有评审，直接在生产环境的 Web 服务器上 &lt;code&gt;systemctl reload</summary>
        
      
    
    
    
    <category term="DevSecOps" scheme="https://weflowly.com/categories/DevSecOps/"/>
    
    
    <category term="单元测试" scheme="https://weflowly.com/tags/%E5%8D%95%E5%85%83%E6%B5%8B%E8%AF%95/"/>
    
    <category term="Flux CD" scheme="https://weflowly.com/tags/Flux-CD/"/>
    
    <category term="WAF" scheme="https://weflowly.com/tags/WAF/"/>
    
    <category term="GitOps" scheme="https://weflowly.com/tags/GitOps/"/>
    
    <category term="ModSecurity" scheme="https://weflowly.com/tags/ModSecurity/"/>
    
  </entry>
  
  <entry>
    <title>构建自定义Rollup插件为Azure AKS上的多语言AI数据服务注入全链路追踪上下文</title>
    <link href="https://weflowly.com/7750255039/"/>
    <id>https://weflowly.com/7750255039/</id>
    <published>2023-10-27T10:15:28.000Z</published>
    <updated>2026-03-20T14:11:15.449Z</updated>
    
    
      
      
        
        
    <summary type="html">&lt;p&gt;我们新上线的AI数据科学平台遇到了一个典型的“黑盒”困境。用户在前端发起一个复杂的分析请求，这个请求会触发部署在Azure</summary>
        
      
    
    
    
    <category term="可观测性" scheme="https://weflowly.com/categories/%E5%8F%AF%E8%A7%82%E6%B5%8B%E6%80%A7/"/>
    
    
    <category term="OpenTelemetry" scheme="https://weflowly.com/tags/OpenTelemetry/"/>
    
    <category term="Rollup" scheme="https://weflowly.com/tags/Rollup/"/>
    
    <category term="Azure AKS" scheme="https://weflowly.com/tags/Azure-AKS/"/>
    
    <category term="AI" scheme="https://weflowly.com/tags/AI/"/>
    
    <category term="MyBatis" scheme="https://weflowly.com/tags/MyBatis/"/>
    
  </entry>
  
</feed>
