daily.dev · Popular · 实时热榜

ChatGPT is experiencing a major worldwide outage starting around 5 AM ET, preventing users from loading chats or interacting with the AI. OpenAI has acknowledged the issue on its status page and is investigating.

Scattering pricing logic across frontend, backend, emails, and background jobs leads to financial discrepancies and debugging nightmares. The backend should be the sole authority for all price calculations, with the frontend acting as a passive display layer. Key practices include defining a single rounding strategy (e.g., Bankers' Rounding), storing amounts in the smallest currency unit (cents), encapsulating currency conversion in one module, and logging all inputs and outputs for traceability

You complained, and you were right. Losing a long streak just because you stepped away on Valentine’s Day or had a life outside the app was frustrating. We got the message. You should never have to choose between a date night and keeping your hard-earned progress. Say hello to streak freezes . Here’s what’s new Stock up in packs: Grab freezes in 3-packs or 5-packs straight from your streak popover. Automatic coverage: Miss a day and a freeze kicks in automatically. No extra steps required. Stash

Launch trailer for Teeto, a cooperative action game where two heroes team up to defeat shadow enemies by absorbing the essence of objects to unlock abilities like vine swinging, boulder smash, and flamethrower.

Merge conflicts are symptoms of process problems, not Git failures. Long-lived branches, large pull requests, and poor team communication are the root causes. Trunk-based development — merging small changes frequently — dramatically reduces conflict size and frequency. The most dangerous conflicts are logical ones Git can't detect, which continuous integration helps surface early. Practical habits like keeping PRs small, using feature flags, and communicating when working in overlapping code are

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.

A hands-on walkthrough of customizing the Cosmic desktop environment on Pop OS, covering applets, tiling, themes, fonts, icon packs, custom cursors built with xcursor-gen, frosted glass windows, and a dynamic wallpaper switcher written in Rust. The author forks a monochrome icon pack to create pixelated variants, designs pixel-art cursors from scratch, and builds a small Rust app called Twilight to swap wallpapers between light and dark mode. The post also touches on gaming on Pop OS with Nvidia

webkul/inertia is a framework-agnostic PHP server-side adapter for Inertia.js, equivalent to inertia-laravel but for plain PHP applications. It implements the full Inertia protocol including standard HTML visits, XHR JSON visits, stale-asset 409 handshakes, and partial reloads with lazy prop resolution. WordPress is supported transparently with automatic detection of WP helpers. Installation is via Composer, and runnable examples are provided for both React 18 and a vanilla JS client demonstrati

Nub is a new all-in-one JavaScript toolkit built on top of Node.js (not replacing it) that consolidates a package manager, TypeScript file runner, script runner, package runner, and Node version manager into a single Rust binary. Created by Colin (the creator of Zod), it claims to be 5x faster than pnpm for installs, 19x faster than npx for package running, and 24x faster than pnpm run for scripts. It supports full TypeScript (including enums, namespaces, decorators), automatic .env loading, JSO

A detailed walkthrough of the system design behind Katabench, a .NET coding platform that safely executes arbitrary C# code submitted by users. The architecture uses three Hetzner VPS servers: one for the .NET API and Postgres, one for isolated code execution sandboxes, and one for deployment and observability. All servers are connected via a self-hosted Headscale/WireGuard VPN with no public exposure beyond ports 80/443. Code submissions flow through NATS JetStream to a worker that spins up thr

Solod 0.3 is a new release of Solod (So), a strict subset of Go that transpiles to plain C with no runtime and manual memory management. This version adds OS thread support via a `conc` package (threads, channels, worker pools), mutex and atomic primitives via `sync` and `sync/atomic`, a streaming token-level JSON encoder/decoder, and several safety features including escape analysis, leak detection via `mem.Tracker`, nil-pointer panics with stack traces, and C sanitizer integration. Tooling imp

What changed CodePen 2.0 is a full rebuild, five years in the making. The old setup had three separate editors — Classic Pen, Projects, and a handful of other tools — each with their own constraints. Version 2.0 collapses all of that into one file-based environment. The biggest structural change is a real file system. The old three-file limit is gone. You can now build multi-page projects, organize assets however you want, and use ES module imports across files. It works the way a local dev envi

Anthropic has released Claude Opus 5, a new model positioned between Opus 4.8 and the top-tier Fable 5, at the same price as its predecessor ($5/M input, $25/M output tokens). It sets new state-of-the-art results on Frontier-Bench and GDPval-AA coding/knowledge work benchmarks, triples the next-best model on ARC-AGI 3, and outperforms all competitors on OSWorld 2.0 computer use at roughly one-third the cost of Fable 5. Key improvements include stronger agentic behavior (self-verification, iterat

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

The Node.js project has launched a beta preview of its redesigned API documentation at beta.docs.nodejs.org. Built with a new standalone tool called doc-kit, the redesign introduces built-in search with keyboard shortcuts, a unified design system shared with the nodejs.org website, a persistent sidebar, per-page table of contents, mobile-friendly layout, and llms.txt support for AI tools. Content is still generated from the same Markdown source files, and all existing features like ESM/CJS switc

Several significant web platform developments are converging: Chrome is experimenting with out-of-order HTML streaming via declarative partial updates using processing instruction markers and the template element, enabling servers to deliver page sections as their data becomes available without JavaScript workarounds. Chrome is also experimenting with rendering real DOM elements inside canvas, preserving browser interactions like text selection and accessibility. ECMAScript 2026 has been officia

Arcovia is a CLI tool that analyzes React, Next.js, and Vite codebases and produces a deterministic architecture score using static analysis — no AI, no network connection required. The author describes the motivation (replacing gut-feel architectural reviews with a reproducible number), the multi-layered scoring model (weighted category deductions, diminishing penalties for repeated findings, capped critical-risk adjustments), and the engineering challenges involved: making the dependency graph

CodePen 2.0 has officially launched, combining the Classic Pen and Project editors into a single, extensible editor. Key new features include one-click deployment to a *.codepen.app subdomain with custom domain support, a built-in file system, version history, real-time collaborative editing, and a Blocks system for mixing technologies via the CodePen Compiler. The editor now supports Vite-powered Vue, Tailwind, Lightning CSS, MJML, and Nunjucks. The Classic Editor remains accessible but is no l

A senior developer's philosophy on writing software for long-term maintainability. Key principles include prioritizing readability over cleverness, treating naming as documentation, writing comments that explain 'why' not 'what', choosing boring but appropriate architecture, respecting the cost of dependencies, investing in documentation and tests, and designing APIs that feel natural. The core mindset shift: write code for your future self who has forgotten all context, not for your present sel

A personal account of finally committing to Vim after years of hesitation. The author argues that Vim's modal editing, home-row navigation, and composable command language reduce physical strain and mental friction, ultimately making editing feel like an extension of thought. Key benefits highlighted include ergonomic comfort (no reaching for mouse or arrow keys) and the expressive command language (e.g., `ciw`, `dt)`) that lets experienced users edit at the speed of thought. The author encourag

Here is information on what will be added new to JavaScript. The version is still being developed, but we already know that a Temporal object will be added, which allows you to work conveniently with the date. https://tc39.es/ecma262

A personal story from a recent university graduate in Australia navigating the harsh job market, low starting salaries, and the country's housing crisis. After landing a first full-time job with disappointing pay, the author took on two jobs seven days a week to afford housing and savings goals, only to eventually resign from the second job after their team was offshored. The piece reflects on the emotional toll of overworking, the tension between financial ambition and personal wellbeing, and t

A developer built an x86 CPU emulator entirely in CSS, using new CSS features like conditional logic and custom functions to track processor memory and registers. It compiles small C programs into real 8086 machine code and executes them inside a stylesheet, currently working in Chromium. JavaScript is not required — CSS animations alone keep the processor running.

A developer shares six months of experience building Tabularis, a cross-platform database client, using Tauri instead of Electron or native frameworks. The core argument: a database client is a systems program, so having a real Rust backend (SQLx, Tokio, keyring, SSH tunnels) matters more than bundle size. The post honestly catalogs the real costs of Tauri's shared-webview model — targeting three different webview implementations (WKWebView, WebKitGTK, WebView2), Linux packaging nightmares (AppI

What is Nub? Nub is a new JavaScript toolkit that sits on top of Node.js rather than replacing it. In a single native binary, it bundles a package manager, script runner, Node version manager, and TypeScript runtime. The pitch is straightforward: get meaningfully faster tooling without abandoning Node.js compatibility or migrating to a different runtime like Deno or Bun. The benchmark claims The Nub team published benchmarks showing a warm reinstall of 1,100+ packages completing in 346ms. For co

OOMWOO is a fully open source robotic vacuum cleaner that runs entirely without cloud services. Built around a Raspberry Pi running ROS 2, it uses an inexpensive 2D lidar for room mapping and integrates with Home Assistant. Most structural parts require a 3D printer, but the software is largely complete and the project is actively developed on GitHub with many areas open for contribution. The project is positioned as a community-driven alternative to proprietary vacuums like Roomba amid growing

A Staff Software Engineer at Aiven argues that AI is creating a structural imbalance in software development: code generation has accelerated dramatically while code review has not scaled to match. AI-generated PRs are often verbose and harder to review, yet management KPIs still reward writing code over reviewing it. The problem extends to testing, where AI can produce high coverage numbers by gaming tests rather than verifying real behavior, leading to gradual system erosion. The deeper concer

A few months ago, I wrote about why a frontend developer working mainly with Vue and TypeScript decided to explore Rust. It started with curiosity. I kept seeing Rust behind tools I already used or admired: ripgrep, Zed, Biome, Alacritty, Tauri, Ratatui, and many others. Tauri became the bridge for me. I could continue building interfaces with familiar web technologies while using Rust for the system-level parts of the application. From there, I started experimenting with: 🦀 CLI tools 🖥️ Termina

A Korean games culture site called 2P_Game_Arcade is hosting a game development contest where all entries must fit within 1.44MB — the capacity of a single HD floppy disk. The entire fileset, including game engine, libraries, and assets, must stay within this limit. Browser-based or streaming games are ineligible. Technology choices are otherwise unrestricted. The prize pool is ₩1.14 million (roughly $750), split among the top three entries. The contest closes September 4, 2026. The competition

A developer built a complete LLM pipeline from scratch in pure Rust — no Python, no PyTorch — called aarambh-ai. The 14-crate workspace covers pretraining, SFT fine-tuning, LoRA/QLoRA, GRPO reinforcement learning, INT4/GGUF quantization, Flash Attention v2 CUDA kernels, a thinking engine, a safety layer, and an online self-learning loop. Four model scales (25M to 1.3B parameters) share identical code. Key technical details include: AdamW with β₂=0.95 (not 0.999), Cholesky decomposition for stabl

A developer shares their hybrid folder structure approach for organizing Vue 3 projects, combining responsibility-first organization at the root level with feature-first grouping for feature-specific code, plus consistent file naming conventions. The post explains the reasoning behind this architecture with production-inspired examples and discusses trade-offs compared to purely responsibility-first or feature-first approaches.

The post reflects on whether learning to code is still worthwhile in an era of AI coding assistants and widespread tech layoffs. It argues that many layoffs are rooted in pandemic-era overhiring rather than AI alone, and warns against 'vibe coding' — using AI-generated code without understanding it — because real systems break in ways that require deep knowledge to debug. It also raises concern about companies cutting junior hiring, which could create a future experience gap. The conclusion is t