daily.dev · Popular · 实时热榜

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

Technical folder structures (components, hooks, services, utils) make React repos easy to browse but hard to change, because a single feature's logic is scattered across unrelated directories. The key insight is organizing code around ownership of change rather than file type. This means colocating a feature's components, state, validation, API calls, and business rules under one feature directory. State placement should reflect lifecycle and source of truth, not convenience. Business rules belo

A provocative argument that developers — even those writing critical code — are not generating enough AI-assisted code. The core thesis is that AI-generated code doesn't have to go into production to be valuable: it can be used as throwaway verification tooling, custom debuggers, test harnesses, and exploratory scripts. The author distinguishes between 'important code that ships' and 'slop code that verifies,' arguing that cheap code generation should dramatically expand the latter category. Eng

Reddit has required login to access old.reddit.com, citing security concerns. The author argues this is misleading PR framing — the real reason is to make scraping harder by forcing users through a heavier JavaScript-dependent frontend. Old Reddit serves plain HTML efficiently, while New Reddit loads 5x more data and tracks user behavior via heartbeat events. The author contends that calling plain HTML 'unsafe' is disingenuous, and that Reddit's actual goal is protecting its LLM-training data fr

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

Laravel 13.21 ships several developer-facing additions: a #[RouteKey] PHP attribute for declaratively setting the route model binding column on Eloquent models (replacing getRouteKeyName() overrides), a base64 validation rule that enforces strict RFC 4648 encoding, a #[RequestAttribute] contextual attribute for injecting request bag values directly into controller parameters, and expanded Image component output formats (PNG, GIF, AVIF, BMP) alongside a fix for the AVIF MIME mapping. The release

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 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 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

Substack shipped an AI detection tool this week, and the most interesting thing about it isn’t the tech. It’s the framing. The platform partnered with Pangram to let readers scan any post, note, reply, or comment (100+ words) for an estimate of how much was written by a human versus AI. The scan is private, visible only to whoever requests it. Writers can also run it on their own drafts before publishing, and optionally add an “AI Author’s note” disclosing their process. No bans, no penalties. J

The HTTP QUERY method, now standardized as RFC 10008, solves a long-standing problem with fetching filtered data over HTTP. GET requests struggle with complex query parameters due to URL length limits, and sending a body with GET is poorly supported. Using POST for reads is a common workaround but carries wrong semantics, is not idempotent, and is not cached by default. QUERY acts as 'GET with a body': it is safe, idempotent, and cacheable (with cache keys derived from both URI and body), and ac

Redential is a developer credential tool that scans your local git history to generate a verifiable profile of what you actually built, without exposing your code. A CLI tool runs locally, extracts patterns and skills from your repos, and creates a shareable public profile. The key differentiator is a live defense mechanism where developers answer real-time questions about their own work history, making the credential harder to fake than a CV or cloned GitHub profile. The CLI is open source and

A small fix for safely restoring rate-limit data after Redis outages evolved into conditional updates, a cleaner API, and ultimately Trypema v2. https://davidoyinbo.com/engineering-logs

A comedic sketch show episode from daily.dev featuring short skits about AI agents, tech culture, privacy, and developer humor. Scenes include AI impersonation security failures, DuckDuckGo blocking YouTube ads, self-hosted deployments replacing Microsoft, South Korea offering free AI to citizens, IBM multi-agent systems, and a social engineering gag where an AI agent impersonates a colleague to approve a production deployment.

Seven months in, System76’s COSMIC desktop is doing something rare: actually shipping. Version 1.3 just landed with 186 contributors, 16,000+ merged pull requests, and a feature list that’s starting to make other Linux desktops look slow. The headline grab is dual-monitor workspace handling. On KDE, switching workspaces moves both monitors together. On COSMIC, each monitor gets its own independent workspace. It sounds like a small thing until you’ve lived without it. One user who switched from K

Laravel Time Machine is a new open-source performance profiler for Laravel applications that records every stage of the HTTP request lifecycle — from bootstrap through termination — with millisecond-level timings. It provides a Gantt-style timeline dashboard at /time-machine, captures SQL queries with bindings and execution times, flags slow requests (>500ms) and slow queries (>50ms), and stores profiles as flat JSON files without requiring database migrations. Custom instrumentation is availabl

Vibe coding tools like Cursor and Claude Code have automated the typing-heavy tasks that once defined junior developer work, but they haven't made the role obsolete. The junior role has shifted from writing first drafts to critically reviewing AI-generated code, spotting security flaws, missing permission checks, and architectural mismatches. The key failure mode for juniors is accepting generated diffs without understanding them. Seniors must adapt mentorship to focus on architecture and prompt

A podcast discussion arguing that vibe coding — using AI tools to rapidly generate software without deep technical knowledge — is fundamentally misleading as a path to production-ready software. While AI coding assistants are genuinely useful for prototyping and making experienced developers more productive, they don't enable non-developers to build scalable, maintainable systems. The conversation highlights how architectural thinking is critical for long-term software health, and that startups

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

It’s no secret that AI has flooded our timelines everywhere, and here at daily dev too. Heck, we’re even working to throttle it down a little algorithmically. I wanted to dig into our data and see if it’s actually what developers read. If so much content is being published about it, then people are probably reading it, right? I’m guessing you thought the same as me. Yesterday I ran a poll here , and 55% out of 720 people, voted AI as the most read topic on daily dev. So I rolled up my sleeves, s

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

Meta has open-sourced Asterisk, an internal React component library built on top of StyleX that has been refined across 13,000 internal applications over eight years. It ships with over 150 accessible components and a powerful theming system that generates design tokens from a few high-level inputs like a brand color, typography scale, border radius, and motion speed. Unlike shadcn/ui, Asterisk keeps components in a central package so upstream fixes are automatic, while still allowing deep custo

A researcher named Lauren Leek investigated how Google Maps ranks restaurants in London, revealing that its algorithm — based on relevance, proximity, and 'prominence' — creates a vicious cycle where new or lesser-known restaurants struggle to gain visibility without reviews, while chains and high-traffic venues dominate. Leek also uncovered undisclosed paid placements in Maps results. Using a machine learning model, she built a public dashboard to surface hidden gem restaurants that punch above

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.

Microsoft PowerToys' Command Palette recently added a stable dock feature that serves as a customizable alternative to the Windows 11 Taskbar. The dock lets users pin commands, extensions, apps, system tools, and even resource monitors (CPU/GPU stats) to three configurable zones. It supports multi-monitor setups, flexible alignment, color and theme customization, and label toggling. Key limitations include no system tray replication, no thumbnail previews, and the native Taskbar cannot be fully

For a long time, I used AI like Google. I’d ask one question, get one answer, and move on. Then I changed one thing. I started treating AI like a teammate instead of a tool. Now my workflow looks like this: ✅ Brainstorm architecture ✅ Review code before PRs ✅ Debug tricky errors ✅ Generate test cases ✅ Improve documentation ✅ Challenge my design decisions The biggest productivity boost didn’t come from better prompts. It came from having better conversations. AI isn’t replacing developers. It’s

A personal account of how picking up self-hosting as a hobby led to cancelling multiple paid subscriptions. By learning Docker and experimenting with self-hosted apps like BentoPDF (replacing Adobe Acrobat) and Karakeep (replacing Pocket), the author discovered that many recurring payments were funding convenience rather than irreplaceable functionality. The key takeaway: self-hosting doesn't mean rejecting all subscriptions, but it forces each one to justify its cost.

A Laravel developer explains why they built an open-source application-level quota manager package after repeatedly implementing ad-hoc usage-limiting logic across projects. Unlike Laravel's built-in RateLimiter (which throttles requests), this package tracks business-level usage quotas such as monthly PDF exports, AI credits, or API budgets. Key design decisions include a fluent builder API, calendar-aware periods without manual date math, Eloquent model integration for automatic key resolution

A video transcript arguing that AI-assisted 'vibe coding' is fundamentally unsuitable for infrastructure and reliability engineering. The core argument is that UI features have tight feedback loops (caught in dev or tests), but infrastructure failures manifest as multi-hour production outages with no clear root cause. Recent real-world outages (Grok uploading git repos, Railway losing Google Cloud API access, COD outage without postmortem) are cited as evidence that infrastructure complexity can

He posted on the kernel mailing list this week that Linux is not an anti-AI project. Anyone who objects can fork the kernel or walk away. “Linux is not one of those anti-AI projects, and if somebody has issues with that they can do the open-source thing and fork it. Or just walk away.” He called AI “a tool, just like other tools we use. And it’s clearly a useful one.” Anybody who doubts that, he said, “clearly hasn’t actually used it.” Here is the twist. Just months ago Torvalds called 90% of AI

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