Post by @nolan@toot.cafe

RE: https://toot.cafe/@nolan/117519051277607728

Working on a new version of the benchmark that's a bit more heavyweight and SPA-flavored, with a few more frameworks based on feedback. Feeling pretty energized and inspired; it's nice to be doing benchmarks again. I was getting a bit tired of writing opinion pieces. 😅

Nolan Lawson

@nolan@toot.cafe 4 days ago

New post: "The AI cost of JavaScript frameworks" https://nolanlawson.com/2026/10/28/the-ai-cost-of-javascript-frameworks/

Which JavaScript frameworks are actually "good for agents"? Let's benchmark it!

0 Boosts

0 Likes

Nick Chomey

@nickchomey@mastodon.social 6 hr. ago

@nolan it would be great if you added Datastar (eg with hono) to it. It's like htmx+alpine, but smaller, faster, more powerful and flexible, and more spec-compliant.

https://data-star.dev/

Can pass the clanker this doc
https://data-star.dev/docs.md

https://data-star.dev/docs.md

Datastar

The hypermedia framework.

Nolan Lawson

@nolan@toot.cafe 2 hr. ago

@nickchomey Thanks, I was not aware of this one. I've already started working on the v2 of the benchmark; I'll see if I can fit it in. 🙂 Otherwise it's always possible to run the benchmark yourself!

Nolan Lawson

@nolan@toot.cafe 19 hr. ago

Honestly I was starting to wonder what, if anything, I could contribute in this weird new world of AI coding. Writing mopey poetry like "We mourn our craft" is great, but at my core, I really want to do deep, novel work that pushes the boundaries of what's possible with software. Having some knowledge of performance, benchmarks, and web frameworks (and where measurements can go wrong) really helps here. Expertise is not dead! But the goalposts have shifted a bit.

jokeyrhyme

@jokeyrhyme@aus.social 17 hr. ago

@nolan thanks so much! between you and @danluu , I love that technology (and especially AI/LLM) myths/speculation/vibes can be dispelled with actual testing and analysis