Hyeseong Kim

Spaces like @e18e.dev, where maintainers can gather, share their journeys, and collaborate, are more important than ever. Hats off to everyone who keeps making the e18e community such a special place in the ecosystem. And the same to all of you who are out there building bridges between oss projects

A new text encodings library, iconv-nano, was posted in the @e18e.dev discord. A day later, the author (Jeremy) is collaborating with @hyeseong.kim on making it ~80% smaller! 🤯 Always impressed by how e18e has managed to join up people with knowledge in such niche subjects 😄

we implemented batched reads in the v3 full-text engine, leading to a ~9x performance improvement. postings are stored as columns, and we were reading them a row at a time sometimes optimization is not about being clever, but simply avoiding mistakes tpuf.link/v3/2026-10-02

ECMAScript excitement 😉 Congrats to @peet.bsky.social from TechAtBloomberg advancing JSON.parse Options to Stage 2.7 at @TC39 today 🎉 There are two independent opt-ins. 🔶 deep freezing the entire object 🔶 non-array objects get a null prototype Names may change ⚠️

Cloudflare Containers now start 6x faster, let your agent choose each sandbox's image and instance type at runtime, and support filesystem snapshots in public beta, all controlled from a Durable Object. https://cfl.re/4z9xz6Q #BirthdayWeek

we're changing tpuf's storage architecture so we can keep pushing the search frontier the new engine (v3) now passes CI, but we have to make it fast before we ship it to prod we're grinding query plans in public until v3 wins on everything follow along: turbopuffer.com/v3

ECMAScript excitement 😉 Congrats to @ashley-c.bsky.social from TechAtBloomberg on advancing the Composites proposal to Stage 2 at @tc39.es today 🎉 Composites are objects that are compared by value instead of identity. They are an evolution of Record & Tuple proposal.

Preact 11 is out 🎉 We’ve been hard at work at making our diffing use modern features like moveBefore , leveraging ESM for tree-shaking Preact Compat and ensuring you have a great experience with resumed hydration.

Several nice additions here. util.debounce(fn, ...) util.throttle(fn, ...) SlidingWindowHistogram More FFI, net, and vfs goodness nodejs.org/en/blog/rele...

A new documentary about @vscode.dev is out! 🎬🎞️🍿 Check out The Story of VS Code, the story of the editor that brought the TypeScript experience to everyone! www.youtube.com/watch?v=kHL3...

Ecosystem performance meets Astro. @florian-lefebvre.dev is here to explain how the @e18e.dev project is making your favorite framework smaller and faster ⚡️

Now I’m actually celebrating! @nodejs.org just merged github.com/nodejs/node/... Which means it will now be possible to bundle whole applications into ZIP archives and run them straight from the bundle.

Node.js 26.9.0 is out, it enables by default node:ffi and adds support for Web Worker API (behind a flag). Full changelog and download links at nodejs.org/en/blog/rele...

I've ported my unicode-segmenter JS library to Rust. It performs text segmentation 4–5x faster than existing libraries for the grapheme-specific path. Check it out: github.com/cometkim/fas...

I thought I’d have some breathing room right after leaving my job, but my home setting is taking a huge amount of time. I'm not sure if I'll be able to finish everything in this month.

React 19.3 is now available! This release makes View Transitions and Fragment Refs stable, and adds browser(), Trusted Types support, and Context in Server Components. react.dev/blog/2026/09...

DwarfStar running DeepSeek v4.1 Flash on a 128GB M5 Max at steady 15 t/s. I didn't expect with SSD streaming it could be so fast. Recent SSD streaming changes to retain the right experts surely helped, but also maybe DS4.1 uses the same experts more. Will push online when ready QA > ASAP.

ulottie compiler is now SSR-ready, which means the compiler can extract the initial frame as a static markup. (even it includes dynamic bodymovin expressions) Your Lottie animation can be in the initial document, and the client side script only hydrates.

Who said that interface files are no longer needed? Writing interfaces yourself and delegating implementation is a great method to collaborate with your agent.

Z AI's GLM-5.3: Built to Code. Ready for Cyber Defense. (open-weight soon) - Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model - A major leap in cybersecurity, setting a new standard among open models z.ai/blog/glm-5.3

I'm p confident a lot of people will resonate with this brettcodes.com/im-done-usin... We need more perspectives from people with lots of experience as developers!

Today we’re launching Bluesky Protocol Services, a new brand and website for the infrastructure Bluesky operates. And Jetstream v2, the best way for developers to use atproto at scale, now with network replay. atproto.com/blog/introdu...

Previously, when I thought I needed a simple Kotlin library for this, I was overwhelmed by the amount of configuration and had to give up on publishing, but the world has become so much more convenient now. github.com/cometkim/kourl

Umm, it seems Claude doesn't really like the AOT approach. It keeps resisting it, calling it an overcomplication, and arguing that runtime dispatch is a simpler solution. Of course, that is true... but it wasn't what I wanted.

Changesets v3 is here! 🦋 After months of chipping away at issues alongside @bluwy.me, @haglund.dev, and many other contributors, we’re excited to finally share it. changesets.dev/blog/announc...

Astro 7.2 is out now! Experimental incremental static builds are here: only re-render the pages that actually changed and watch your build times drop. Plus opt-out session support, background astro preview, and relative logger entrypoints. astro.build/blog/astro-7...

had another good chat about @e18e.dev and what the community has been up to. check it out! 🎉 big thanks to @josh.bressers.name for having me along and for the good conversation 🙏

I found a real bug in Kubernetes through webernetes, my port of Kubernetes to TypeScript. github.com/kubernetes/k... I found it by using webernetes to visualise how liveness probes work and noticed they didn’t work as I expected. Thought it was a webernetes bug but it was a Kubernetes bug!

An enormous amount of the world’s CI and automation runs headless Chrome, all of it on x86—until last week. Chrome now ships official arm64 Linux builds. AWS says Graviton uses up to 60% less energy for the same performance. Chrome for Testing coming soon 🥳 calibreapp.com/blog/chrome-...

Cloudflare OS is an open-source platform that lets everyone in your company build apps, automate work, and safely access internal systems, shaped around what your organization knows and how it operates. https://cfl.re/4fGBl0w

And we are live now with @honk.foo youtube.com/live/L8Edgas...

Don't be a meat proxy gruhn.me/blog/2026-08...

gRPC support is now available in Cloudflare Workers. You can now route, transform, and serve gRPC services natively at the edge without maintaining dedicated proxy infrastructure or translation layers. Read the launch post: https://cfl.re/4fMsitT

github.com/cometkim/rky... allows Rust's rkyv from/to JS applications. You can (zero-copy) deserialize wire messages sent from JS in Rust, or vice versa! Would be useful for JS-Rust interop projects, e.g., a Rust CLI with Cloudflare Workers, a web app with a Rust backend, etc.

github.com/cometkim/uni... unicode-segmenter `countGrapheme` speedup by avoiding unnecessary allocation and iterator overhead. If you are using `Intl.Segmenter` simply to count numbers, you could consider this. It is almost 15-30x faster (varying depending on the input corpus).

After much effort, it seems I have approached the lower threshold of the library's bundle size. When I make the code smaller, its compressed size actually becomes larger.

📣 The moment is here. 📣 TypeScript 7 is officially released! 7️⃣ devblogs.microsoft.com/typescript/a...

OMG, @jddalton.bsky.social has a wild proposal for Rspack's native binding: - size: 38.4 MiB -> 13.8 MiB (-64%) - first load: 853ms -> 615ms (-28%)

Cloudflare Research is building a global consensus service called Meerkat that uses a new consensus algorithm called QuePaxa. We plan to use Meerkat to build a strongly consistent, fault-tolerant key-value store, and other applications. https://cfl.re/3SKpK7L

unicode-segmenter/grapheme is now only 2.5 KiB mingzipped. Faster than ever. github.com/cometkim/uni... Last dance with Claude Fable 5

uhh… TIL: <marquee direction="down" behavior="alternate"> <marquee behavior="alternate">This text will bounce</marquee> </marquee> developer.mozilla.org/en-US/docs/W...