Obsidian 已推出全新的社区站点 community.obsidian.md，这是一个集中式的插件与主题目录兼开发者仪表板。自 2020 年 API 发布以来，社区已开发了超过 4,000 个插件和主题，总下载量突破 1.2 亿次。新平台旨在让扩展的构建、分发、发现与使用对所有人来说更简单、更安全。

社区站点改进了浏览、搜索、筛选和排序功能，覆盖数十个分类，如集成、 Bases 和图表。每个项目都有详细页面，展示截图、信息和安全评分卡。新增标签用于标识付费插件与官方集成，作者还可以在个人资料中添加赞助链接和社交媒体信息。

新的开发者仪表板让作者可以提交、管理并跟踪项目。所有现有的插件、主题及来自 GitHub 的排队提交都已自动迁移。开发者只需登录并连接 GitHub 账户即可认领项目。仪表板简化了提交流程，审核结果通常在几分钟内就能得到。

此次发布引入了对所有社区项目的自动化审查，覆盖每个版本的安全性和代码质量，而不再只检查初次提交。这有效缓解了随着编码代理加速插件创建而产生的积压问题。系统会检查是否遵守开发者政策、代码最佳实践以及已知漏洞；对热门插件、精选项目和社区标记的问题仍会保留人工复审。在上线前几天，使用新系统处理了超过 2,300 个排队提交。

插件安全性通过自动化扫描得到显著提升，扫描项包括代码质量、安全漏洞和恶意软件。每个项目的评分卡都会显示自动化检测的结果。后续计划将加入功能权限披露（显示插件访问的内容）、经过验证的作者标识，以及更清晰的插件功能与作者信息透明度。

使用 Obsidian 的团队将获得更完善的工具，用于管理社区插件并向成员分发私有插件。发布官方插件的团队现在可以在社区目录中申请官方标识。现有的团队部署与安全控制仍然可用。

常见问题部分解答了关于迁移与过渡的疑问。用户可以在新站点上浏览，可能会发现此前需要手动获取的早期访问插件已能通过目录获取。开发者可通过五步流程提交新项目，审核速度更快。当前自动化审查不通过的现有插件暂时仍可继续使用，但最终需要达到新的标准。开发者可以在本地运行自动化审查或在发布更新前预览扫描结果。该平台目前需要 GitHub 与 Obsidian 账户，未来可能会支持更多托管平台。 Obsidian has launched the new Community site at community.obsidian.md, a centralized directory and developer dashboard for plugins and themes. Since the API release in 2020, over 4,000 community plugins and themes have been created, with total downloads surpassing 120 million. The new platform aims to make building, distributing, discovering, and using extensions easier and safer for everyone.

The Community site offers improved browsing, searching, filtering, and sorting capabilities across dozens of categories like Integrations, Bases, and Charts. Each project has a detail page with screenshots, information, and a safety scorecard. New labels identify paid plugins and official integrations, while authors can customize their profiles with sponsorship links and social media.

A new developer dashboard allows authors to submit, manage, and track their projects. All existing plugins, themes, and queued submissions from GitHub have been automatically migrated. Developers can claim their projects by signing in and connecting their GitHub accounts. The dashboard provides a streamlined submission process with results typically available within minutes.

The launch introduces automated reviews for all community projects, scanning every version for security and code quality rather than just initial submissions. This addresses the growing backlog as coding agents accelerate plugin creation. The system checks adherence to developer policies, code best practices, and known vulnerabilities. Manual reviews will continue for popular plugins, featured projects, and community-flagged issues. Over 2,300 queued submissions were processed in the final days before launch using the new system.

Plugin safety receives major enhancements through automated scans checking every version for code quality, security vulnerabilities, and malware. Scorecards display the status of automated checks for each project. Future improvements will include capability disclosures showing what plugins access, verified author labels for trusted developers, and better transparency about plugin functionality and authors.

Teams using Obsidian will gain better tools for managing community plugins and distributing private plugins to team members. Teams publishing official plugins can now apply for the Official badge in the Community directory. Existing team safety controls for deploying plugins remain available.

The FAQ section addresses common questions about the transition. Users can explore the new site and may find previously manual early-access plugins now available through the directory. Developers can submit new projects through a five-step process with faster review times. Existing plugins failing automated review will continue working for now but will eventually need to meet new standards. Tools are available to run automated reviews locally or preview scans before releasing updates. The platform currently requires GitHub and an Obsidian account, with plans to potentially add other hosting platforms in the future.

高级开发人员难以把自己的专业性传达清楚，往往不是因为知识不足，而是因为他们与业务其他部门"说的是两种语言"。问题的核心在于框架不同：市场、销售、产品和高管关注的是如何快速把想法投到市场、尽快验证并减少不确定性；而高级开发关注的是管理复杂性以保证系统稳定。这种根本性的错位导致了冲突。当开发以维护成本或代码复杂性为由回绝新功能时，他们是在表达运营层面的担忧，但并未触及业务对不确定性的根本焦虑。

把业务想象成同时运行两条循环会更容易理解：一条是速度循环——快速探索市场、从反馈中学习、尽快降低不确定性；另一条是稳定循环——确保付费用户能持续获得可靠、可理解且可修复的服务。高级开发主要处在第二条循环，他们害怕复杂性，因为复杂性直接威胁到他们必须维护的稳定性。每增加一个功能或特殊用例，系统就更难理解、排查和维护，风险随之上升，客户服务也可能受到影响。

沟通失败正发生在这两条循环碰撞的地方。业务把新功能视为学习和减不确定性的机会；高级开发把同一请求视为可能破坏稳定性的复杂来源。对那些在时间和不确定性中奔跑的业务方而言，开发对复杂性的担忧常被看作阻挠。开发者在沟通上会失败，往往是因为他们只用内部的复杂性指标来解释问题，而没有承认业务试图减少不确定性的目标。

那么高级开发该如何更有效沟通？作者 Tuhin Nair 建议做一个简单但有力的重新表述：不要一上来就谈复杂性问题，而是把自己的方案用"减少不确定性"的角度来呈现。他提出的关键句是："我们能试试更快的方法吗？"这句话既承认了业务对速度的诉求，"更快"暗示有替代路径，"试试"接受不完美但可能"够用"。开发者可以借此倡导简化、复用现有工具或避免不必要的投入，同时把这些作为更快获得市场反馈的手段来表述。

但 AI 的兴起改变了这一动态。 AI 能让任何人快速生成代码，加速那个以速度和不确定性为主的业务循环，代价却是稳定性。 AI 往往会生成更难理解、调试与维护的代码，而且并不对所产出的系统承担责任。这使得高级开发的角色比以往更重要，但他们的工作也在从单纯的构建者转向以"编辑者"为主——对产出承担最终责任的人。

为此，Nair 提出一个结构性解决方案：把两条业务循环在系统层面解耦。设想有一个"速度"版本的系统，让 AI 代理、初级开发和非技术人员可以在其中快速构建并发布功能以获取市场反馈，而不必为长期稳定性担忧；验证有价值的功能再进入"规模"版本，由高级开发作为编辑进行策划，实现设计良好、稳定且易于理解的最终系统。这样，业务既能保持速度与试错的动力，高级开发也能在负责的系统设计和稳定性上发挥真正价值，通过落地的成果而非空谈复杂性来证明自己的作用。 Senior developers often struggle to communicate their expertise not because they lack knowledge, but because they speak a different language than the rest of the business. The core issue lies in how problems are framed. While marketers, salespeople, product managers, and executives are driven by the need to reduce uncertainty, getting ideas to market quickly to test what works, senior developers are focused on managing complexity to maintain system stability. This creates a fundamental misalignment. When developers push back on feature requests by citing maintenance costs or code complexity, they are expressing their own operational concerns, but failing to address the business core anxiety about uncertainty.

To understand this dynamic, it helps to think of a business as running in two simultaneous loops. The first loop is about speed, exploring the market, learning from feedback, and reducing uncertainty as fast as possible. The second loop is about stability, ensuring that paying customers continue to receive a reliable, understandable, and fixable service. Senior developers live primarily in the second loop, fearing the monster of complexity because it directly threatens the stability they are responsible for. Every new feature or special case added to a system makes it harder to understand, debug, and maintain, which increases risk and threatens the continued service of customers.

The communication failure happens when these two loops collide. The rest of the business sees a new feature request as an opportunity to learn and reduce uncertainty. The senior developer sees the same request as a source of complexity that could destabilize the system. This leads to a disconnect where developer concerns about complexity feel like stonewalling to business stakeholders who are racing against time and uncertainty. Developers fail when they try to explain away a business problem using their own internal metrics, focusing only on complexity management instead of acknowledging the business goal of uncertainty reduction.

So, how can senior developers communicate more effectively? The author, Tuhin Nair, suggests a simple but powerful reframing. Instead of leading with the problems of complexity, senior developers should present their solutions in terms of uncertainty reduction. The proposed magic phrase is, "Can we try something quicker?" This phrase acknowledges the business need for speed, "quicker" implies an alternative path, and "try" accepts imperfection while suggesting a potentially good enough solution. It allows a developer to advocate for simplification, re-using existing tools, or avoiding unnecessary work, all while framing these actions as a faster way to get the market feedback the business craves.

However, the rise of AI has disrupted this entire dynamic. AI accelerates the business loop of speed and uncertainty reduction by allowing anyone to generate code quickly, but it does so at the cost of stability. AI is a destabilizer, creating code that is often harder to understand, debug, and maintain. Crucially, AI takes no responsibility for the systems it helps create. This makes the senior developer role more critical than ever, but their job is shifting from being solely a builder to being primarily an editor, the person who takes on the responsibility for what gets built.

Nair proposes a structural solution to this problem: decoupling the two business loops into separate systems. Imagine a 'Speed' version of a system, where AI agents, junior developers, and non-technical staff can rapidly build and ship features for market feedback without worrying about long-term stability. This feeds into a 'Scale' version of the system, a well-designed, stable, and understandable version curated by senior developers acting as editor. Features get tested in 'Speed' and then, if they prove valuable, are properly integrated into 'Scale'. This way, the business gets its speed and momentum, while senior developers get the space to apply their expertise in responsible system design and stability, truly communicating their value through action rather than just words about complexity.

Jeff Geerling 一直批评 Bambu Lab 将其 3D 打印机转向"始终在线"的云模式。在 Bambu Lab 把云服务设为默认后，他通过 OPNsense 防火墙切断打印机的互联网连接，停止固件更新，将设备锁定在开发者模式，并把工作流从 Bambu Studio 换到 OrcaSlicer，以此维持对自己 P1S 的控制权。他承认自己希望完全拥有已购硬件的想法并不普遍，但 Bambu Lab 近期的做法促使他更严厉地发声。

导火索是 Bambu Lab 对一个名为 OrcaSlicer-bambulab 的 OrcaSlicer 分支的反应。该分支允许用户在不通过 Bambu 云的情况下使用打印机全部功能。 Bambu Lab 没有置之不理，而是以冒充和安全风险为由威胁开发者采取法律行动，尽管该分支照搬了 Bambu Studio 上游的 AGPL 授权代码。 Geerling 认为这种回应既令人费解又不成比例，尤其是在该分支在收到停止函前使用量极少的情况下。

Bambu Lab 在一篇博客中将该分支描述为严重的安全威胁，声称其会注入伪造的身份元数据，可能对其服务器造成结构性漏洞。 Geerling 指出讽刺之处：Bambu 自己的某个分支在 2022 年曾导致 Bambu 用户的遥测误发到 Prusa 的服务器上，Prusa 并未以法律威胁回应。他认为 Bambu 在滥用从中受益的 AGPL 许可，同时也误解了开源文化和基本安全原则——毕竟公开的 user-agent 字符串并不能作为防御 DDoS 的可靠手段。

Geerling 把 Bambu 的强硬做法与可行的温和处理方式对比，例如仅以商标为由要求开发者在分支名中去掉 "bambulabs" 。然而，Bambu 却公开将潜在的基础设施问题归咎于一名开发者。这与去年他们把社区的反弹称为"令人遗憾的误导信息"的做法一脉相承。更荒谬的是，涉事开发者此前还在 Bambu 的 GitHub 上帮助 Bambu Studio 用户解决 Linux 和 Wayland 问题。

更大的问题在于 Bambu Lab 正在逐步封闭其生态，背离了曾帮助他们建立社区的开源原则。 Geerling 指出，最简单也最有利可图的选择本来是无为而治，但他们选择压制那一小部分高级用户。 Louis Rossmann 已承诺出资 1 万美元帮助该开发者应对 Bambu 的法律威胁，Geerling 也愿意出力，但开发者可能并不想继续成为 Bambu 的靶心。对于重视开源和硬件所有权的用户来说，最实际的结论或许就是避开 Bambu Lab 。 Jeff Geerling has been a vocal critic of Bambu Lab's shift toward an always-connected cloud model for their 3D printers. After Bambu Lab started pushing their cloud solution as the default, he took steps to maintain control over his own P1S printer by blocking its internet access via his OPNsense firewall, stopping firmware updates, locking it into Developer mode, and switching from Bambu Studio to OrcaSlicer. He acknowledges he's unusual in wanting full ownership of hardware he purchased, but Bambu Lab's recent actions have pushed him to speak out more forcefully.

The immediate trigger for this post is Bambu Lab's response to a fork of OrcaSlicer called OrcaSlicer-bambulab, which allowed users to access all printer features without routing prints through Bambu's cloud. Instead of ignoring this small project, Bambu Lab threatened the developer with legal action, accusing the fork of impersonation and posing a security risk, despite the fork using Bambu Studio's upstream AGPL-licensed code verbatim. Geerling finds this response baffling and disproportionate, especially since the fork had minimal uptake before Bambu's cease and desist.

Bambu Lab published a blog post framing the fork as a serious security threat, claiming it injected falsified identity metadata and could cause structural vulnerability to their servers. Geerling points out the irony, noting that Bambu's own fork once caused Bambu users' telemetry to hit Prusa's servers in 2022, and Prusa didn't respond with legal threats. He argues that Bambu is misusing the AGPL license they benefit from and misunderstanding both open source culture and basic security principles, since a public user agent string is hardly a robust defense against DDoS attacks.

Geerling contrasts Bambu's heavy-handed approach with how the situation could have been handled, such as simply asking the developer to remove "bambulabs" from the fork's name for trademark reasons. Instead, they publicly blamed one developer for potential infrastructure problems, echoing a pattern from last year when they dismissed community backlash as "unfortunate misinformation." The developer in question had previously helped Bambu Studio users with Linux and Wayland issues on Bambu's own GitHub, making the legal threat especially absurd.

The broader issue is Bambu Lab's increasing lock-in of their ecosystem, moving away from the open source principles that helped build their community. Geerling notes that it would have been easier and more profitable for Bambu to do nothing, but they chose to suppress a tiny fraction of power users. Louis Rossmann has pledged $10,000 to help the developer fight Bambu's legal threats, and Geerling would contribute too, but the developer may not want to remain in Bambu's crosshairs. The practical takeaway for users who value open source and hardware ownership might be to simply avoid Bambu Lab altogether.

本文记录了在浏览器中构建真实大气散射着色器的一个月历程，灵感来自一张航天飞机 Endeavour 在日落时分拍摄的、地球高层大气色彩斑斓的照片。目标不是简单地绘制蓝色渐变背景，而是把天空模拟成光与空气分子、尘埃及其他颗粒在体积中相互作用的结果，采用像 raymarching 、 Rayleigh 和 Mie 散射以及臭氧吸收等技术。

作者首先用 raymarching 为大气着色器打下基础，采样大气密度并计算光学深度。具体做法包括用 Rayleigh 密度函数模拟随高度稀薄的大气分布，并应用 Beer's Law 计算透射率，即光穿过大气后剩余的强度。随后引入 Rayleigh 相位函数，考虑入射阳光与视线间的夹角，从而呈现出更真实的蓝天效果：较短波长的光散射更强。

为增强真实感，文章加入了 Mie 散射，用以模拟光与较大颗粒（如尘埃和气溶胶）的相互作用，在太阳周围形成朦胧的光晕；还加入了臭氧吸收，它会吸收部分波长的光，使天空颜色在太阳附近和黄昏时向更深的蓝紫色偏移。文中通过具体的密度与相位函数以及标高和散射系数等常数给出了这些效果的实现细节。

接着讨论光照部分，解释如何通过考虑从太阳到每个采样点路径上的光衰减来正确渲染日出和日落。实现方法是在主 raymarching 循环内嵌套一个"光线步进"循环，用于在光源方向上采样透射率，从而让着色器能准确表现一天之中天空颜色从正午明亮的蓝色到日落时温暖的橙红色的变化。

文章还把平面天空着色器改造为适用于三维行星周围大气的后期效果：通过屏幕空间 UV 和场景深度缓冲重建世界空间坐标，使 raymarching 能考虑场景几何，从而产生随距离变得更朦胧的大气雾效。作者还详述了为处理行星尺度启用对数深度缓冲的方法，并用光线与球体相交测试来确定大气壳层的起点和终点。

一个附加章节探讨了日食处理，介绍了如何实现太阳可见度函数，根据 moon 等天体的视半径和位置判定它们对太阳的遮挡程度。另一个附加示例展示了通过调整大气常数来模拟不同行星环境，例如 Mars 上尘埃弥漫的橙色大气及其独特的蓝色日落。最后，文章还介绍了一种受 Sebastian Hillaire 工作启发的更高性能的基于查找表（LUT）的方法：将昂贵的散射计算预先烘焙到透射率、天空视角和空气透视等 LUT 纹理中，然后在最终通道合成，从而相较于完全 raymarching 的实现获得显著的性能提升。 This article documents a month-long journey into building realistic atmospheric scattering shaders in the browser, inspired by a photo of the space shuttle Endeavour against the colorful layers of Earth's upper atmosphere at sunset. The goal was to move beyond simple blue gradient backgrounds and instead simulate the sky as the result of light interacting with air molecules, dust, and other particles in a volumetric way, using techniques like raymarching, Rayleigh and Mie scattering, and ozone absorption.

The author begins by laying the foundation for an atmosphere shader using raymarching to sample atmospheric density and calculate optical depth. This involves using the Rayleigh density function to model how air thins with altitude and applying Beer's Law to compute transmittance, which represents how much light survives traveling through the atmosphere. The Rayleigh phase function is then introduced to account for the angle between incoming sunlight and the view ray, resulting in a realistic blue sky where shorter wavelengths scatter more strongly.

To enhance realism, the article covers the addition of Mie scattering, which models light interaction with larger particles like dust and aerosols, creating a hazy glow around the sun. Ozone absorption is also incorporated, which removes certain wavelengths of light and shifts the sky's color toward deeper blues and purples, especially near the sun and during twilight. The implementation details for these effects are provided through specific density and phase functions, along with constants for scale heights and scattering coefficients.

The discussion then shifts to lighting, explaining how to properly render sunsets and sunrises by accounting for light attenuation along the path from the sun to each sample point. This is achieved by introducing a nested "lightmarch" loop within the main raymarching loop to sample transmittance in the direction of the light source. This allows the shader to accurately depict the changing colors of the sky throughout the day, from the bright blue of midday to the warm oranges and reds of sunset.

The article progresses to transforming the flat sky shader into a post-processing effect suitable for rendering atmospheres around 3D planets. This involves reconstructing world-space coordinates from screen-space UVs and the scene's depth buffer, allowing the raymarching to account for scene geometry and create atmospheric fog that makes distant objects appear hazier. The author also details enabling logarithmic depth buffers to handle the large scales involved in planetary rendering and using ray-sphere intersection tests to define the start and end points of the atmosphere shell.

A bonus section explores handling eclipses by implementing a sun visibility function that determines how much a celestial body like the moon obstructs the sun based on their apparent sizes and positions. Another bonus demonstrates how tweaking atmospheric constants can simulate different planetary environments, such as the dusty, orangy atmosphere of Mars with its distinctive blue sunsets. The article concludes with an exploration of a more performant LUT-based approach inspired by Sebastian Hillaire's work, which precomputes expensive scattering calculations into textures like transmittance, sky-view, and aerial perspective LUTs, then composes them in a final pass for a significant performance boost over the fully raymarched version.

欧盟正加大对主要社交媒体平台的监管力度，重点打击官员所称在 TikTok 和 Instagram 上的"成瘾性设计"。欧盟委员会主席 Ursula von der Leyen 在丹麦峰会上宣布，欧盟预计将在今年晚些时候推出针对这些平台的正式法规。此次整治聚焦无限滚动、自动播放和推送通知等功能，这些功能被认为对儿童和青少年尤其有害。 von der Leyen 还批评 Meta 未能执行其 13 岁最低年龄要求，指出未成年人可以轻易绕过现有的审核机制。

除界面设计外，欧盟还在调查通过算法将儿童引入"兔子洞"、使其接触有害内容的平台，例如宣扬饮食失调或自残的视频。为配合这些工作，委员会开发了一款自有的年龄验证应用，声称符合全球最高的隐私标准。成员国将很快能够将该技术整合进其数字钱包。 von der Leyen 强调，"年龄验证技术已经可用"，平台不再有借口。一项正式法律提案最早可能在夏季提出，具体取决于欧盟儿童安全在线专家小组的调查结果。

此轮监管也发生在欧盟与美国科技巨头关系紧张的背景下。过去两年，欧盟已对 Apple 、 Meta 和 Google 等公司处以超过 70 亿美元的反垄断和竞争罚款。这些处罚引发了美方批评，Donald Trump 签署备忘录，表示将考虑采取关税等应对措施，称欧盟对美企不公。欧盟还对其他平台展开调查，包括 Elon Musk 的 X，因其传播由其 Grok 聊天机器人生成的露骨非自愿内容。

监管审查的加强还受到美国司法进展的推动。今年 3 月，Meta 和 YouTube 在一宗备受关注的法院裁决中败诉，法院认定无限滚动和自动播放等设计功能助长了青少年的成瘾并损害心理健康。欧盟委员会最近也认定 Meta 违反了 Digital Services Act，未能阻止 13 岁以下用户使用其平台。这些判定为全球保护儿童在线安全的努力增添了动力。去年 12 月，Australia 成为首个对 16 岁以下实施社交媒体禁令的国家；包括 Spain 、 France 和 the U.K. 在内的多个欧洲国家也在提出类似立法，限制儿童访问社交媒体平台。 The European Union is intensifying its regulatory efforts against major social media platforms, specifically targeting what officials call "addictive design" features on TikTok and Instagram. EU Commission President Ursula von der Leyen announced these plans at a summit in Denmark, stating the bloc expects to introduce formal regulation for platforms later this year. The crackdown focuses on features like endless scrolling, autoplay, and push notifications that are seen as particularly harmful to children and teenagers. Von der Leyen also criticized Meta for failing to enforce its own minimum age requirement of 13, noting that minors can easily bypass existing checks.

Beyond design features, the EU is investigating platforms that allow children to access harmful content through algorithmic "rabbit holes," including videos promoting eating disorders or self-harm. To support these efforts, the Commission has developed its own age verification app, which it claims meets the highest privacy standards globally. Member states will soon be able to integrate this technology into their digital wallets, with von der Leyen emphasizing that "the technology for age-verification is available" and platforms have no more excuses. A formal legal proposal could come as early as summer, pending findings from the EU's Special Panel of experts on Child Safety Online.

This regulatory push comes amid broader tensions between the EU and U.S. tech giants. Over the past two years, the bloc has imposed over $7 billion in fines on companies like Apple, Meta, and Google for antitrust and competition violations. These penalties have drawn criticism from U.S. officials, with President Donald Trump signing a memorandum to consider tariffs in response to what he views as unfair targeting of American companies. The EU has also launched investigations into other platforms, including Elon Musk's X, for spreading explicit non-consensual content generated by its Grok chatbot.

The increased scrutiny follows significant legal developments in the United States, where Meta and YouTube lost a high-profile court ruling in March. The court found that design features like infinite scrolling and autoplay contributed to addiction and mental health harms in teenagers. More recently, the EU Commission determined that Meta breached the Digital Services Act by failing to keep under-13s off its platforms. These findings are adding momentum to global efforts to protect children online, with Australia becoming the first country to enforce a social media ban for under-16s in December. Several European nations, including Spain, France, and the U.K., are now proposing similar legislation to restrict children's access to social media platforms.

作者以自己从生物信息学实验室转向担任 IntelliJ Rust 项目负责人的经历为例，认为学习软件设计更适合通过实战而不是依靠正规教育。大学课程能打好基础，但真正的理解来自于在真实项目中动手、解决问题。因为软件工程本质上是讲逻辑的，好奇的人可以通过试验和阅读逐步掌握。

一个重要的洞见是 Conway's Law：软件架构会映射出构建它的组织的社会结构。作者认为，科学软件与工业软件的差别更多源自激励机制而非技术水平。例如，学术界受到发表期限等压力，会影响科学软件的开发方式。他建议两条策略：一是在可能的情况下去影响激励机制（如 TIGER_STYLE 这样的项目所示）；二是接受现有约束，在其范围内高效地工作。

以 rust-analyzer 为例，作者指出它既"深"（具有编译器级的复杂性）又"广"（包含大量 IDE 功能），因此需要针对不同类型的贡献者采取不同策略。为了吸引专注的核心开发者，他把优先级放在干净、快速的构建系统和核心架构的高质量代码上；而对于添加孤立功能的偶发贡献者，他则降低了准入门槛，允许质量要求较低的代码，只要这些改动的失败可以被限制住、对用户不可见。这种务实做法在质量与社区增长之间取得了平衡。

他也提醒，迎合既有激励机制存在风险，因为未来的需求难以预测。 rust-analyzer 最初只是一个用于验证 rustc 架构的实验性原型，后来竟发展成了完整的编译器；同样，uutils 最初是 Rust 开发者的练习项目，后来成为 Ubuntu 的 coreutils 实现。这些例子说明项目常会因环境变化而超出最初的设想。

在具体学习资源方面，作者推荐了若干有影响力的作品。 Gary Bernhardt 的 "Boundaries" 演讲提供了实用的设计建议，并引发了他对软件架构的更深入思考。他自己关于测试的随笔反映了在意识到许多传统测试观念无效之后总结出的宝贵经验。 Pieter Hintjens 关于 ZeroMQ 的著作让他接触到以 Conway's Law 为出发点的思维和"乐观合并"等概念，这些都影响了 rust-analyzer 的开发模式。另有值得关注的还有 Jamii 的反思性编码文章和 Ted Kaminski 的博客，后者把软件开发作为一个连贯的理论体系来探讨。

尽管像 Software Engineering at Google 和 Ousterhout 的 A Philosophy of Software Design 这样的书很有价值，作者认为它们仍不及亲身实践带来的改变深刻。他的结论是：在软件设计上达到精通，主要靠做、观察和不断迭代，而不是依赖某一本权威书籍。整个过程需要接受不确定性，适应社会性动态，并通过现实世界的挑战不断打磨自己的理解。 The author, drawing from personal experience transitioning from a bioinformatics lab to leading the IntelliJ Rust project, argues that software design is best learned through practice rather than formal education. He notes that while university courses provide foundational knowledge, real understanding comes from hands-on experience and problem-solving in actual projects. This learning process is accessible because software engineering is fundamentally logical and can be grasped by curious individuals through experimentation and reading.

A key insight emphasized is Conway's Law, which states that software architecture mirrors the social structure of the organization building it. The author suggests that differences between scientific and industrial software stem more from incentive structures than technical knowledge. For example, academic pressures like publishing deadlines shape how scientific software is developed. He advises two approaches: influencing incentive structures when possible, as seen in projects like TIGER_STYLE, or adapting to existing constraints by accepting limitations and working effectively within them.

Using rust-analyzer as a case study, the author explains how its dual nature, being both deep (compiler-level complexity) and broad (many IDE features), required different strategies for different contributor types. To attract dedicated core developers, he prioritized a clean, fast build system and high-quality code in the central architecture. For casual contributors adding isolated features, he lowered the barrier to entry by allowing less rigorous code, as long as failures were contained and invisible to users. This pragmatic approach balanced quality with community growth.

He cautions that adapting to incentive structures carries risks, since future needs are unpredictable. Rust-analyzer began as an experimental prototype to inform rustc's architecture but evolved into a full-fledged compiler itself. Similarly, the uutils project started as a learning exercise for Rust developers but became Ubuntu's coreutils implementation. These examples illustrate how projects can outgrow their original intent due to shifting circumstances.

For concrete learning resources, the author recommends several influential works. Gary Bernhardt's "Boundaries" talk offers practical design advice and sparked his deeper inquiry into software architecture. His own essay on testing reflects hard-won insights after realizing much conventional testing wisdom is unhelpful. Pieter Hintjens' writings on ZeroMQ introduced him to Conway's Law thinking and concepts like optimistic merging, which influenced rust-analyzer's development model. Other notable mentions include Jamii's reflective essay on coding and Ted Kaminski's blog, which approaches software development as a coherent theoretical framework.

While books like Software Engineering at Google and Ousterhout's A Philosophy of Software Design are valuable, the author finds them less transformative than experiential learning. He concludes that mastery in software design comes primarily from doing, observing, and iterating, rather than from any single authoritative source. The journey involves embracing uncertainty, adapting to social dynamics, and continuously refining one's understanding through real-world challenges.

本合集为复古技术媒体的完整档案，收录了 1983 年至 2007 年间的操作系统截图与桌面环境。图库涵盖多种历史计算平台，既有早期图形界面（如 Visi On 1.0 、 SunTools 、 GEM Desktop 、 HP-UX），也有更先进的系统（如 NeXTstep 、 OS/2 Presentation Manager 、 Windows 3.0，以及来自 SGI 、 DEC 、 Sun 的各类 Unix 工作站）。每个条目均详尽记录硬件、显示分辨率、色深和具体软件版本等技术细节，构成了近三十年图形用户界面演变的视觉档案。

这些截图突显了计算史上的若干重要里程碑——从原始的窗口系统向复杂桌面环境的转变。值得注意的示例包括 Acorn Archimedes 上的 Arthur OS 、创新但短暂的 BeOS on the BeBox 、 Apple 曾雄心勃勃但最终放弃的 Copland 项目，以及从 Rhapsody 到 Mac OS X Server 与 Public Beta 的演进。合集还收录了许多小众与专业系统，如用于 Lisp 编程的 TI microExplorer 、 Symbolics Genera，以及各类基于 X11 的桌面（OpenWindows 、 Motif 、 CDE），展示了该时期用户界面设计的多样性。

档案的另一独特之处在于策展人个人技术历程的记录，收录在"策展人主要桌面环境的演变"一节。该时间线追溯了他们自 1996 年的 OS/2 Warp Connect，经 Windows NT 、 IRIX 、 NEXTSTEP 到 Mac OS X 的使用历程，为更广泛的技术变迁提供了个人视角。该合集既可作为技术参考，也是一段对个人计算革命中快速创新与实验的怀旧回顾，保留了那些曾经开创且如今逐渐淡出的系统的外观与使用感受。 This collection showcases a comprehensive archive of retrotechnology media, specifically focusing on operating system screenshots and desktop environments spanning from 1983 to 2007. The gallery features screen captures from a diverse array of historical computing platforms, including early graphical interfaces like Visi On 1.0, SunTools, GEM Desktop, and HP-UX, as well as more advanced systems such as NeXTstep, OS/2 Presentation Manager, Windows 3.0, and various Unix-based workstations from SGI, DEC, and Sun. Each entry is meticulously documented with technical details about the hardware, display resolution, color depth, and specific software versions, offering a detailed visual record of the evolution of graphical user interfaces over nearly three decades.

The screenshots highlight significant milestones in computing history, such as the transition from rudimentary windowing systems to sophisticated desktop environments. Notable examples include the Arthur OS on Acorn Archimedes, the innovative but short-lived BeOS on the BeBox, Apple's ambitious but ultimately abandoned Copland project, and the progression from Rhapsody to Mac OS X Server and the Public Beta. The collection also captures niche and specialized systems, such as the TI microExplorer for Lisp programming, Symbolics Genera, and various X11-based desktops like OpenWindows, Motif, and CDE, illustrating the diversity of approaches to user interface design during this era.

A unique aspect of the archive is the curator's personal journey through these technologies, documented in a section titled "Evolution of the curator's primary desktop environments." This timeline traces their own computing experience from OS/2 Warp Connect in 1996, through Windows NT, IRIX, and NEXTSTEP, to Mac OS X, providing a personal narrative that contextualizes the broader technological evolution. The collection serves as both a technical reference and a nostalgic look back at the rapid innovation and experimentation that characterized the personal computing revolution, preserving the look and feel of systems that were often groundbreaking in their time but have since faded into obscurity.

AWS 上的 Claude 平台现已全面可用，AWS 客户现在可以通过原生的 AWS 身份验证、计费和承诺抵消，访问完整的 Claude API 功能集。对于已经在 AWS 上构建的团队而言，这意味着可沿用现有的 IAM 凭证与权限来使用 Claude，且计费可合并到同一张 AWS 发票中。

平台支持的高级功能包括 Claude Managed Agents（用于大规模部署 agent）、用于提升 agent 智能的 advisor 策略，以及代码执行、网络搜索、 Skills 、 Files API 、 Prompt Caching 、批量处理（Batch Processing）和 MCP connector 等实用工具。 Anthropic 的开发环境 Claude Console 也一并提供，包含 prompt improver 、 generator 和评估工具。目前可用的模型有 Claude Opus 4.7 、 Sonnet 4.6 和 Haiku 4.5，新模型在发布后会立即上架。

该产品面向想要完整 Claude 平台体验的 AWS 用户。服务由 Anthropic 运营，数据处理位于 AWS 边界之外，因此在功能上与原生 Claude API 完全一致，且可在第一时间访问所有新能力和 beta 功能。这使得它对追求最新模型特性和简化云运营的工程团队非常有吸引力，例如使用 Claude Code 的团队。 ReliaQuest 、 OpenRouter 和 Emergent 等早期采用者表示，这一方案让他们更快将先进 AI 集成到网络安全、工程和平台工作流中，同时保持现有云治理和工具链不变。

公告还明确区分了 Claude Platform on AWS 与 Amazon Bedrock 上的 Claude：两者虽都运行在 AWS 上，但 Bedrock 将 AWS 作为数据处理方并在其基础设施内处理数据，这对有严格数据驻留要求的客户至关重要。相比之下，AWS 上的 Claude 平台在 AWS 边界之外运行，更适合将功能广度和迭代速度放在首位而非将所有处理留在 AWS 内部的客户。企业应根据合规需求做出选择：需要数据完全留在 AWS 内部的可选 Bedrock，追求完整功能与更快更新的可选 AWS 上的 Claude 平台。

从落地角度看，AWS 客户可以沿用熟悉的上手流程：身份验证通过 IAM，审计通过 CloudTrail，计费纳入 AWS 承诺，从而无需对接额外供应商或管理新账户。该平台支持全球与美国的推理地域，并在大多数 AWS 商业区可用，便于安全、财务与工程团队在统一的云合同和治理框架下，直接访问 Anthropic 的最新模型与工具。

部分功能仍属 beta，例如 Managed Agents 、 advisor 策略、 Files API 和 Skills，但已作为平台功能提供。当前在 Bedrock 并拥有 private offers 的客户需与 Anthropic 或 AWS 客户经理沟通，妥善转移折扣，因为折扣无法追溯。总体而言，AWS 上的 Claude 平台被定位为 Anthropic 的一项首创产品，使 AWS 客户能在现有云环境中更便捷地构建更强大的 agent 与应用，减少运营负担折衷，并即时使用不断演进的 Claude API 。 The Claude Platform on AWS is now generally available, giving AWS customers a new way to access the full Claude API feature set with native AWS authentication, billing, and commitment retirement. This launch means that teams already building on AWS can use their existing IAM credentials and permissions to work with Claude, with billing consolidated onto a single AWS invoice. Customers can access advanced tools like Claude Managed Agents for deploying agents at scale, the advisor strategy for boosting agent intelligence, and practical utilities such as code execution, web search, skills, files API, prompt caching, batch processing, and the MCP connector. The Claude Console, Anthropic's development environment, is included and provides a prompt improver, generator, and evaluation tools. Currently, Claude Opus 4.7, Sonnet 4.6, and Haiku 4.5 are available, with new models shipping as they launch.

This offering is designed specifically for AWS users who want the complete Claude Platform experience. Anthropic operates the service and data is processed outside the AWS boundary, which means full feature parity with the native Claude API and day-one access to all new capabilities and betas. That makes it a strong fit for engineering teams, like those using Claude Code, who want cutting-edge model features and a streamlined cloud operating model. Several early adopters, including ReliaQuest, OpenRouter, and Emergent, highlight that this setup let them integrate advanced AI into cybersecurity, engineering, and platform workflows faster, while keeping existing cloud governance and tooling intact.

The announcement clarifies that this is different from Claude on Amazon Bedrock. While both run on AWS, Bedrock keeps AWS as the data processor within its infrastructure, which is important for customers with strict data residency requirements. The Claude Platform on AWS runs outside the AWS boundary, so it appeals to those who prioritize feature breadth and velocity over keeping all processing inside AWS's own perimeter. Companies are encouraged to choose based on their compliance needs: fully in-AWS data handling with Bedrock versus full feature access and control with the Claude Platform on AWS.

From a practical standpoint, AWS customers get a familiar onboarding path. Authentication happens via IAM, auditing through CloudTrail, and billing through AWS commitments, so teams don't have to integrate a separate provider or manage new accounts. The platform supports global and U.S. inference geographies, with availability across most AWS commercial regions. For organizations, this means security, finance, and engineering teams can align on a single cloud contract and governance model while still getting direct access to Anthropic's latest models and tools.

Some features are still in beta, such as Managed Agents, the advisor strategy, Files API, and Skills, but they are included now as part of the platform. Customers already on Bedrock with private offers should coordinate with their Anthropic or AWS account executives to transfer discounts correctly, since discounts cannot be applied retroactively. Overall, the Claude Platform on AWS is positioned as a first-of-its-kind offering from Anthropic, letting AWS customers build more capable agents and applications inside their existing cloud environment, with fewer operational overhead trade-offs and immediate access to the evolving Claude API surface.

"They Live Adblocker" 是基于流行扩展 uBlock Origin Lite (uBOL) 的一个富有创意的分支，它从 John Carpenter 1988 年的邪典电影 They Live 中汲取了独特元素。与单纯隐藏那些绕过网络层屏蔽的广告不同，这个扩展会把这些广告替换为白色方块，方块上随机显示电影中的标语，如 "OBEY" 、 "CONSUME" 、 "SLEEP" 、 "SUBMIT" 和 "NO INDEPENDENT THOUGHT" 。这个想法源自开发者在 2015 年的一篇博客，最终被付诸实现。

项目改变了 uBO Lite 处理"元素隐藏"过滤的方式。标准插件通过注入 CSS 、使用 display: none 来隐藏广告元素；而该分支会拦截这些注入点，改用 CSS 的 ::after 叠加层绘制白色遮罩。每个被屏蔽的广告会通过 data-ubol-they-live 属性获得一条随机标语。扩展还用 MutationObserver 来处理页面初始加载后动态插入的广告。

安装流程与 Chromium 内核浏览器（如 Chrome 、 Brave 、 Edge）上的开发者模式方法相同：从 GitHub 下载最新发布包、解压并在 chrome://extensions 以"未打包扩展"方式加载。不过，开发者提示，要真正看到那些带有 OBEY 等字样的方块，需将过滤模式从默认的 "Basic" 切换为 "Optimal" 或 "Complete" 。因为 Basic 模式只在网络层阻止请求，不会在 DOM 中创建被拦截的元素；而只有通过元素隐藏规则筛出的广告才会被替换成带标语的遮罩。

从源码构建要求 Node 版本 22，步骤包括递归克隆仓库并运行 uBlock Origin 的标准构建脚本。修改后的文件放在 davmlaw/uBlock 的 they-live 分支的一个独立子模块中，使元素替换逻辑与上游代码分离。关键改动包括新增的 they-live.js（包含标语列表和 CSS 生成器）以及对若干处理过滤注入点的脚本文件的修改。

项目有几条重要说明：首先，它明确标注为个人兴趣的 fork，与官方 uBlock Origin 无关。将原本隐藏的元素重新显示有时可能破坏页面布局——因为站点的 CSS 可能假定广告位会收缩。用户自定义的元素隐藏过滤仍照常生效，不会应用 They Live 效果；另外，通过网络层屏蔽掉的广告不会有标语叠加，因为根本不存在可供修改的 DOM 元素。

尽管改动相对简单，该扩展在 GitHub 上已获 103 星，说明不少用户喜欢在浏览时看到这种带有颠覆意味的小彩蛋，当广告侥幸绕过网络过滤时还能看到一条电影式的评论。 The "They Live Adblocker" is a creative fork of the popular uBlock Origin Lite (uBOL) browser extension with a unique twist from John Carpenter's 1988 cult film They Live. Instead of simply hiding advertisements that slip through network-level blocking, this extension replaces them with white tiles featuring iconic slogans from the movie, including "OBEY," "CONSUME," "SLEEP," "SUBMIT," and "NO INDEPENDENT THOUGHT." The concept originated from a 2015 blog post by the developer, who finally decided to bring the idea to life.

The project modifies how uBO Lite handles cosmetic filtering. While the standard extension injects CSS to hide ad elements with `display: none` rules, this fork intercepts those injection sites and instead applies a white-box mask using a CSS `::after` overlay. Each blocked advertisement receives a random phrase from the movie's slogan list through a `data-ubol-they-live` attribute. The extension also employs a MutationObserver to handle dynamically loaded ads that appear after the initial page load.

Installation follows the standard developer mode process for Chromium-based browsers like Chrome, Brave, and Edge. Users download the latest release package from GitHub, extract it, and load it as an unpacked extension through `chrome://extensions`. However, the developer notes that to actually see the OBEY tiles, users need to change their filtering mode from the default "Basic" to either "Optimal" or "Complete." This is because basic mode only blocks ads at the network level, preventing DOM elements from being created altogether. The cosmetic replacement only works on ads that uBO Lite filters through cosmetic rules.

Building from source requires Node version 22 and involves cloning the recursive repository and running the standard uBlock Origin build script. The modified files reside in a separate submodule at `davmlaw/uBlock` on the `they-live` branch, keeping the cosmetic replacement logic separate from the upstream codebase. Key modifications include a new `they-live.js` file containing the phrase list and CSS generator, plus changes to several scripting files that handle the actual filter injection points.

The project comes with several important caveats. First and foremost, it's explicitly marked as a personal hobby fork and not affiliated with the official uBlock Origin project. Making previously visible elements can occasionally break page layouts where site CSS assumed ad slots would collapse. User-defined custom cosmetic filters still function normally without the They Live treatment, and network-level blocks don't get the slogan overlay since there's no DOM element to modify.

Despite being a relatively simple modification, the extension has garnered 103 stars on GitHub, suggesting that plenty of internet users appreciate having a bit of subversive commentary injected into their browsing experience every time an advertisement manages to slip past the network-level defenses.