Keel 0.3 发布：Rust 写的静态类型解释语言补上标准库、Map 和文档站 Keel 0.3 这次更新挺像一个“从实验玩具往可用语言继续推进”的节点。作者把它定位成一门用 Rust 编写、语法接近 Rust、使用体验又想靠近 Python的静态类型解释语言，而且继续强调它在解释执行场景下的速度目标：大约能做到 Python 的 2 到 10 倍，同时维持零注解式的类型推断体验。对...

As geopolitics shift, more countries are wooing semiconductor manufacturers to enhance resilience. At the same time, demand for advanced semiconductors continues to grow.

By Parth Jain, Rakesh Sukumar, Yingwu Zhao, Renzo Sanchez-Silva & Nathan FisherHow we built a living map of our distributed infrastructure to help engineers understand dependencies, troubleshoot...

本文讨论了多种 BPF map 类型的实现及其应用，包括环形缓冲区、性能事件数组、布隆过滤器、队列、栈和 LPM trie。每种 map 解决特定问题，如事件输出、快速排除和数据传递。理解这些 map 的特性有助于在架构决策中选择合适的工具，以提高性能和效率。

BPF程序在内核中执行时无法访问全局变量和调用内核函数，唯一的持久化机制是BPF map。本文分析了BPF map的内核实现，包括hash表和数组的结构、并发模型及适用场景。hash map使用分桶链表和预分配策略，而array map则采用连续内存布局，支持零拷贝。per-CPU变体允许每个CPU独立操作，避免缓存行竞争。理解这些并发模型对优化BPF程序性能至关重要。

本文介绍了libbpf的加载生命周期，包括ELF解析、BPF程序和map的创建与加载。libbpf简化了BPF程序的加载过程，解决了传统bcc模式的依赖和延迟问题。文章详细阐述了libbpf的API设计、Skeleton的类型安全封装，以及ring buffer消费者的实现，强调了其在生产环境中的应用价值。

By Parth Jain, Rakesh Sukumar, Yingwu Zhao, Renzo Sanchez & Nathan FisherHow we built a living map of our distributed infrastructure to help engineers understand dependencies, troubleshoot faster,...

On sensemaking, the stories organizations tell themselves, and why crisis is sometimes the only thing that reveals the truth. There is a room in California that, for years, everyone called the...

Anthropic's Claude Code CLI had its full TypeScript source exposed after a source map file was accidentally included in version 2.1.88 of its npm package. The 512,000-line codebase was archived to...

Go 语言的 sync/v2 提案通过引入泛型，旨在解决 sync 包的类型安全和性能问题，改进 API，提升并发编程体验。尽管引发讨论，支持者认为性能提升和开发体验改善是必要的。

PGConf.EU schedule can feel like a parallel query gone wild, so many great talks but not enough CPU. I built this guide to help my fellow database nerds skip the overwhelm and enjoy the best...

字节与MAP团队提出的FR3E算法旨在解决大语言模型在强化学习中的探索不足问题。该算法通过“先返回、再探索”的框架，动态平衡利用与探索，显著提升模型的推理能力和多样性。实验结果表明，FR3E在多个数学基准上表现优异。

New AI model integrates petabytes of Earth observation data to generate a unified data representation that revolutionizes global mapping and monitoring

New AI model integrates petabytes of Earth observation data to generate a unified data representation that revolutionizes global mapping and monitoring