Vite just solved its biggest architectural debt. 🔥 Since its inception, Vite ran two separate bundlers — esbuild for dev, Rollup for production. It worked, but it meant two plugin systems, subtle inconsistencies, and the classic "works in dev, breaks in prod" headache. Vite 8 replaces both with Rolldown — a single Rust-based bundler powering everything. The results are staggering: ⚡ Linear: 46s → 6s production builds ⚡ Beehiiv: 64% build time reduction ⚡ Ramp: 57% faster ⚡ Mercedes-Benz.io: up to 38% reduction But speed isn't even the real story. The real win? One bundler = one plugin API, one transformation pipeline, zero dev/prod drift. Most existing plugins work without changes. Vite 8 also marks the arrival of a fully integrated stack: Vite + Rolldown + Oxc. Parsing, bundling, minification — all tightly coordinated. The new plugin-react v6 already dropped Babel entirely thanks to Oxc. Other goodies I'm excited about: → Native tsconfig paths support (goodbye alias plugins) → Built-in devtools → Browser console forwarding to terminal (huge for AI coding agents) → Built-in emitDecoratorMetadata support 🙌 I wrote a deep dive covering the architecture shift, migration path, and what's coming next. Link in the comments 👇 #Vite #Rolldown #JavaScript #TypeScript #WebDevelopment #FrontendEngineering #DeveloperTools #Rust #Performance #OpenSource

Using Dev Containers for AL development? Your debugger might silently fail to open source files when it hits a breakpoint. There's an open issue on the AL GitHub repo affecting developers who use the default /workspaces mount path in Dev Containers. The AL debugger hits the breakpoint, pauses execution, but can't resolve the file path to show you the source code. You're staring at a call stack with no code to read. The problem is a path mismatch. When you create a Dev Container with the default configuration, VS Code mounts your workspace under /workspaces/your-project. But the AL debugger resolves source file paths relative to a different root. The paths don't align, so the debugger knows where it stopped but can't map that location back to an actual file in your editor. This is particularly frustrating because everything else works. IntelliSense runs fine. Publishing works. The app deploys and executes. It's only when you need to step through code that the debugger breaks down silently. The workaround reported by developers is to customize your devcontainer.json to explicitly set the workspace mount path to match what the AL debugger expects. Alternatively, you can configure the launch.json with explicit path mappings using the "breakOnNext" and source path settings. If you're setting up AL Dev Containers, test your debugger before you invest hours in the development workflow. A broken debugger in a containerized setup defeats the purpose of the whole environment. Have you hit path resolution issues with the AL debugger in containers? Repost to help others ♻️ and follow me for more tips! 🚀 #msdyn365bc #businesscentral #al #docker #devcontainers #debugging

Great code doesn’t just work. It communicates. Write it so the next developer can read it without stress. Clear structure. Simple logic. Clean naming. If you could build one thing this year, what would it be? #NGSSolution #CodingTips #CleanCode #SoftwareDevelopment

Most developers use Claude Code at maybe 20% of its potential. After months of real-world use, here are the hacks that actually move the needle: 1. CLAUDE.md Setup > Stop re-explaining your codebase every session. Run /init once. Claude analyzes your project and creates a persistent memory file with your stack, conventions, and rules. What used to take 15 minutes of onboarding? Gone. Forever. 2. Custom Slash Commands > Automate what you repeat. Build commands like /bug-triage, /deploy, or /feature-plan inside .claude/commands/. One command can fetch GitHub issues, analyze root causes, and propose a fix automatically. 3. Plan Before Coding > Always. 2 minutes of planning saves 20 minutes of refactoring. Ask Claude to map dependencies, understand the codebase, and give you a roadmap before writing a single line. 4. Draft PRs via Claude > Low risk, high reward. Let Claude handle PR creation. Review everything before marking it ready. You stay in control while Claude handles the grunt work. 5. MCP Integrations > Drive a Ferrari in all gears. ==> Connect Context7 to pull live docs on the fly ==> Connect Playwright for browser automation ==> Stop copy-pasting documentation and hoping Claude remembers obscure APIs 6. Parallel Agents > Multitask like a team. Run multiple Claude sessions simultaneously. - One fixing bugs - One building features - One writing docs Use Git worktrees to keep environments clean and isolated. 7. Double-Check Prompt > Your secret QA weapon. After any output, say: "Double check every claim and make a table of what you verified." This one prompt alone has saved me hours of debugging. The reality? Claude Code is not just an autocomplete tool. It is an AI dev platform. The developers shipping fastest are not the ones with the best syntax knowledge. They are the ones who have mastered the workflow. You are not a coder anymore. You are a director. Repost if this helped someone on your network level up. What is your favourite Claude Code hack? Drop it below. #ClaudeCode #DeveloperProductivity #AITools #SoftwareDevelopment #Anthropic #CodingTips #AIEngineering

You were told not to swipe. But you still did. Now you know what Code and Core really does. 👀 #CodeAndCore #DigitalProducts #WebDevelopment #TechPartner

A good developer writes code that works. A valuable developer writes code that saves business time and cost. What I prioritize when building products: • Clear system design before coding • Reusable and testable components • Fast load times and efficient queries • Easy future scaling Technology should support business growth, not slow it down. #SoftwareEngineering #TechForBusiness #MERNStack #OpenToOpportunities

Why tsc --declaration and Module Federation fall short for type sharing — and what I built instead If you've worked with microfrontends, you've probably hit this wall: ❌ tsc --declaration generates .d.ts for everything — all transitive dependencies, internal types, private members. No tree shaking. No way to share only your public API contract. ❌ Module Federation's type plugins (dts-plugin, dts-loader) are tied to webpack. If you use Vite, Rspack, or a custom setup — you're on your own. I got tired of this and built ts-remote — a framework-agnostic toolchain: 📦 Builder — generates a consolidated .d.ts with real tree shaking: → Only explicitly exported types are included → Private/protected class members are stripped automatically → Internal dependencies are inlined, not leaked → Multiple entry points → single .d.ts file → 3-5x faster than tsc 🔗 Fetcher / CLI — downloads and caches remote types by URL 🔌 TS Plugin — add to tsconfig.json and your IDE resolves remote types automatically What makes it different: ✅ Tree shaking for .d.ts — only your public API, nothing extra ✅ Framework-agnostic — works with any bundler and deployment strategy ✅ Full IDE support — autocomplete, go-to-definition, type errors out of the box ✅ Zero dependencies — only peer dep on TypeScript 🔗 GitHub: https://lnkd.in/gPAr7Ka2 📦 npm: https://lnkd.in/ghArxWEU Contributions and feedback are very welcome! ⭐ #typescript #microfrontends #frontend #opensource #dx #webdevelopment #modulefederation

“Can you just add one small change?” — User *…while sitting comfortably* 😌 “Sure…” — Developer *…carrying the entire system on their back* 🪨😅 This perfectly sums up the reality of software development. What looks like a “small change” from the outside often means: 🔧 Code refactoring 🔗 Dependency checks 🧪 Testing & edge cases 🚀 Deployment & monitoring Behind every simple feature is a developer solving unseen complexity. Next time you work with a developer, remember: 👉 Simple for you ≠ Simple in code Respect the process. Respect the builder. 💻 #DeveloperLife #SoftwareDevelopment #CodingReality #TechLife #ProgrammerHumor #BuildInPublic

As engineering teams push for faster delivery and better developer experience, tooling can quietly become either a bottleneck or a multiplier. At nCircle Tech, leading teams has taught me one thing : velocity is not just speed - it’s clarity, confidence, and the right tools. Vite has already been a game-changer with its lightning fast dev server and instant HMR. But Vite+ pushes that foundation further by focusing on developer comfort and sets new performance benchmarks⚡. 🔗 please do new feature check here - https://lnkd.in/df3aWK7k Why It Matters for Engineering Leaders ? Vite+ unifies your entire web development workflow into a single, powerful command-line interface. 1) ⏩ dev & build - Blazingly fast builds with hot module replacement. 2) ✅ check - Format, lint, and type-check in one pass. 3) 🧪 Test - Feature rich test runner. 4) 🏃♂️➡️ Run - Execute built-in commands and package.json scripts with automated caching and dependency-aware execution. 5) 📦 Pack - Library packaging with best practices baked in. Great teams don’t just ship fast — they ship with clarity and tools like Vite+ help create that clarity by letting developers focus on solving real problems, not wiring configs. If you’re building modern frontends and want to accelerate both productivity and confidence across your team, Vite+ deserves a serious look 👀. #vite #react #frontend #javascript #typescript #webdevelopment #leadership #engineeringmanagement

🚨 Use #ClaudeCode and go FAST! ⚡ Prompt "Read the README and generate all the tests" ⚡ Just saying YES to everything! ⚡ If it doesn't work, prompt "try again" ⚡ Still not working? "think hard and try again" 👍 Repeat until it pass all the tests! Many developers are doing this right now 🫠 Even if the code pass all your tests (also generated, not crafted), that doesn't mean it works. Passing N tests just means that it won't fail in those N cases, but there is always more!