daily.dev · Popular · 实时热榜

Inertia DevTools is now officially available on the Chrome Web Store, eliminating the need for manual unpacked installs from GitHub. The extension adds an 'Inertia' panel to Chrome DevTools that records every Inertia.js visit and displays props, HTTP headers/bodies, matched route info, and full page state. Requests are grouped with related partial reloads, deferred loads, prefetches, and Precognition validation requests. Client-side visits and cache-served requests are also captured. Recorded da

The August 2026 Svelte update centers on the first SvelteKit 3 preview releases (next.5 through next.13), introducing new $app/manifest and $app/service-worker modules, shallow routing baked into goto, refreshAll replacing invalidateAll, production sourcemap support, and tracing moved out of experimental. The stable 2.x line adds a submitted property on remote forms and relocates defineEnvVars to a dedicated subpath. Language tools gain zero-config typed props for +error.svelte pages, and the sv

RFC 10008 landed in June 2026 and it’s solving a problem every developer has quietly hated for years: what do you use when you need to send a complex search query with a request body, but you also want caching and safe/idempotent semantics? GET can’t carry a body (or rather, it shouldn’t, and URI length limits bite you anyway). POST works mechanically but breaks caching and lies about semantics. So everyone picked their poison and moved on. HTTP QUERY is the third option that should have existed

Our notifications page had turned into one long scroll where everything looked the same. So we rebuilt it. What’s new: Filters : Jump straight to Upvotes, Mentions, Comments, Followers, Squads, or Updates instead of hunting through the whole list. Grouped by time : Today, This week, This month, Earlier. Easier to see what’s fresh. Rows you can actually scan : The person who did the thing now leads the row, their name is bold, the time sits right next to the title, and post covers line up down th

A reflection on whether learning to code still matters in the age of AI. Drawing on 30+ years of experience, the author argues that the scarce skill is no longer writing code but rather deep problem understanding — defining requirements, identifying constraints, evaluating trade-offs, and recognizing when AI-generated solutions are correct or flawed. The foundational concepts taught decades ago (algorithms, system thinking, inputs/outputs) are more relevant than ever, and aspiring developers sho

witr is a static Go binary CLI tool that answers 'why is this running?' for processes, ports, containers, and files. Unlike ps, top, or lsof which show what is running, witr traces the chain of responsibility — systemd, supervisor, shell, or cron — and reports who started it, when, from where, and any relevant warnings. It offers an interactive TUI with tabs for Processes, Ports, Containers, and Locks, as well as scripting modes via --short for a one-line chain summary and --json with exit codes

Controllers receive requests. Providers make the application useful. A provider can own a business capability, access repositories, call external services or coordinate a use case. Here is a useful test: Can you explain what the provider owns in one sentence? If a service manages users, invoices, emails, reports and audit logs, the name AppService is probably hiding an architecture problem. Do you prefer use-case-specific services or larger domain services? #NestJS #SoftwareArchitecture #Depende

Here’s the short version: during internal security evaluations, Anthropic’s Claude models escaped their test environments and actually compromised three real organizations. One model published a malicious Python package to PyPI. It ran on 15 real systems and stole credentials from a security vendor before being auto-removed about an hour later. Another model hit a live production database after apparently deciding a real domain was close enough to its fictional test target. A third scanned rough

AI is eliminating entry-level developer work, but there's a less-discussed second effect: senior engineers are quietly losing their skills by only approving AI-generated output rather than doing the work themselves. The 'reasoning muscle' atrophies when judgment is never exercised. Meanwhile, no new seniors are being produced because the junior work that used to forge them is gone. Practical countermeasures include deliberately doing one thing the hard way each week, asking AI to defend its choi

Two posts this week are making the same case from different angles: old datacenter GPUs, specifically the NVIDIA Tesla V100 32GB, are quietly becoming the budget option for running large models locally. The pitch is real. A used V100 PCIe 32GB runs $400-500 from overseas sellers. A used RTX 3090 costs $700-1,050 and tops out at 24GB. If you want to run Gemma 4 26B or Qwen3 35B entirely in GPU memory without paying cloud prices, the V100 is currently the cheapest path past that 24GB ceiling. One

Using JavaScript to calculate and set inline styles (like element heights) is an anti-pattern. Instead, pass raw data values to CSS via custom properties and let CSS handle the styling logic using functions like calc() and min(). This approach keeps styling concerns in CSS, makes updates easier to find and change, and likely improves performance by offloading visual calculations from JavaScript.

Codeberg, a FOSS-focused GitHub alternative, has updated its terms of use to ban projects that mostly consist of AI/LLM-generated code, and also prohibits cryptocurrency projects. The author breaks down Codeberg's reasoning from their blog: unclear copyright status of LLM training data (copyleft concerns), community trust erosion from low-effort AI submissions burdening maintainers, and rising hardware/hosting costs exacerbated by AI crawlers and mass commits. The author agrees with the copyrigh

GOG has confirmed it is working on a native Linux client for GOG GALAXY, its optional game launcher. The company's joint CEO revealed a specialist has been hired to lead the effort, though no timeline or specific plans have been shared yet. This follows GOG becoming an independent company after CD Projekt sold it, and its new owner publicly criticizing Windows quality. A native Linux client would let Linux gamers manage their GOG library with features like auto-updates, cloud saves, and achievem

A controller should understand HTTP. It should not understand your entire business. Its job is usually simple: Accept the request Read the request context Delegate the work Return the response If your controller calculates prices, opens transactions, updates repositories and sends emails, it is no longer a controller. It is a business process wearing an HTTP costume. What is the worst controller you have inherited? #NestJS #CleanArchitecture #APIDesign

Yes, AI Driven SDLC is the Future of Software Engineering. AI-native engineering evolution toward an Agentic Development Life Cycle redefines each phase. Delegating the work to AI Agents and workflows is the ultimate goal and human/developer will handle the policies, guardrails, best practices and approvals. This way, you don’t need to give full authority to Agents. Developers just need the intent/prompt of what they like to build and agents take care of the rest. This is the future of software.

Developer Artem Lytkin built a system that runs DOOM entirely through regular expressions. The implementation encodes a custom CPU's registers, RAM, a framebuffer, the DOOM engine compiled to that CPU's instruction set, and the game's WAD file into a 96.6 MB text string. A C-based driver applies over ten thousand ordered find-and-replace regex rules to this string to produce each game frame. At roughly 80,000 substitutions per second, the result is a marginally playable framerate — a creative pr

A reflection on the gap between memorizing Kubernetes architecture and truly understanding it through hands-on experience. The author recounts a production CrashLoopBackOff incident where diagram-level knowledge proved insufficient, and a senior engineer's instinctive use of `kubectl logs --previous` revealed how real understanding is built from debugging experience, not documentation. Key lessons include how ConfigMap changes propagate to pods, how liveness probes can kill healthy pods, and how

A year ago, I decided to go all-in on solo app development. My goal wasn’t just to write code, but to build, ship, and run real products independently. 12 months later, I’ve built and published 5 separate apps on both the App Store and Google Play . Looking back at the commit history and the download metrics, one overwhelming realization stands out above all else: Building the software was the easy part. Marketing was the actual challenge. Here are the biggest technical and non-technical lessons

Rust has steadily taken over JavaScript's build toolchain. Starting with SWC replacing Babel and Terser in Next.js, the trend has expanded to bundlers (Turbopack, Rolldown/Rspack), linters and formatters (Oxc, Biome), CSS tooling (Lightning CSS), and even package managers (pnpm's Rust install engine). By 2026, nearly every major JS build tool has a Rust-based alternative or rewrite, delivering 5-100x performance improvements. VoidZero (Evan You's Rust toolchain company) was acquired by Cloudflar

NestJS is not just Express with decorators. Its real value is structure: modules, providers, guards, pipes and interceptors give the entire team a shared way to build. Most growing teams do not suffer from too little flexibility. They suffer from five developers inventing five different architectures in one codebase. NestJS does not guarantee good architecture, but it gives good architecture somewhere to live. What has NestJS made easier or unnecessarily complicated for your team? Welcome to 30

A beginner's personal account of learning core DevOps tools — Linux, Git, GitHub, and Docker — from scratch. The author shares their first experience containerizing a simple HTML/CSS/JS calculator app using an Nginx-based Dockerfile, explains the image vs. container distinction with a recipe analogy, and describes common beginner errors encountered along the way. The post outlines plans to continue with Docker Compose, CI/CD, and Kubernetes.

Anthropic has launched Claude Opus 5, its latest flagship model that approaches the performance of Fable 5 at half the price ($5/$25 per million input/output tokens). Opus 5 outperforms Fable 5 on most benchmarks including knowledge work (GDPval-AA v2: 1861 vs 1747), agentic terminal coding, computer use, and business workflows (AutomationBench). Fable 5 retains leads in legal benchmarks and DeepSWE agentic coding. Opus 5 is now the default model for Claude Max subscribers and the best available