rich harris's Avatar

rich harris

@rich-harris.dev

my life is derp and i do derp shit

20,579
Followers
700
Following
1,635
Posts
23.04.2023
Joined
Posts Following

Latest posts by rich harris @rich-harris.dev

Preview
Community Session: Svelte on Vercel Learn more about Svelte on Vercel.

@svelte.dev devs / Svelte-curious folks - join us for a live session with the team!

I'll be chatting with @rich-harris.dev, @ell.iott.dev and @dummdidumm.bsky.social + @eveporcello.com to share more on @vercel.com Academy.

Thursday 12th March, 10AM PST

community.vercel.com/live/34169-c...

11.03.2026 14:38 πŸ‘ 8 πŸ” 4 πŸ’¬ 1 πŸ“Œ 3

ok tell me more about these 'calculations'. because i think you are missing the point. we are talking about _language models_ with context window constraints β€” a preference for concise, human-understandable abstractions is basically written into their DNA

11.03.2026 14:53 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

nope bsky.app/profile/rich...

11.03.2026 12:06 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

not merely a good framework but an excellent framework. just not the _best_ framework 😎

10.03.2026 22:08 πŸ‘ 4 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

hiring managers making decisions based on some half-remembered linkedin hustleslop probably aren't the most sophisticated individuals. but it's their funeral, commitment to legacy technology in an era of free rewrites is a competitive disadvantage

10.03.2026 20:22 πŸ‘ 17 πŸ” 0 πŸ’¬ 0 πŸ“Œ 1

clown event still possible

10.03.2026 17:29 πŸ‘ 69 πŸ” 3 πŸ’¬ 2 πŸ“Œ 0
Preview
Best practices β€’ Svelte Docs Best practices β€’ Svelte documentation

you'd have to ask them, but it sounds like it overlaps a lot with the `svelte-core-bestpractices` skill we just released, which pulls from svelte.dev/docs/svelte/...

(we may add more docs-powered best practices skills in future, e.g. for SvelteKit)

10.03.2026 17:16 πŸ‘ 24 πŸ” 2 πŸ’¬ 0 πŸ“Œ 1
Preview
How we Rewrote 130K Lines from React to Svelte in Two Weeks We ported Strawberry's entire frontend from React to Svelte in two weeks using coding agents. It made the browser 2x faster.

myth: LLM-assisted coding means everyone is forced to use The Popular Framework forever

reality: agents will happily migrate your entire app to The Good Framework

inertia is a powerful drug but we are entering a world without moats, plan accordingly

strawberrybrowser.com/blog/react-t...

10.03.2026 16:27 πŸ‘ 205 πŸ” 24 πŸ’¬ 13 πŸ“Œ 7

πŸŽ‚

08.03.2026 10:45 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

if you noticed your svelte build/lint/whatever times just got faster, this is why. such a concrete demonstration of the value of open source

06.03.2026 19:20 πŸ‘ 130 πŸ” 8 πŸ’¬ 1 πŸ“Œ 0
Preview
[focusgroup] Multi-directional focus groups Β· Issue #1390 Β· openui/open-ui Apologies if this has already been addressed somewhere; couldn't find an open issue or anything in the demos. The current behaviour of focusgroup is (IIUC) for the down/right arrow keys to focus th...

ask and ye shall receive! github.com/openui/open-...

05.03.2026 22:21 πŸ‘ 6 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

1) what

05.03.2026 17:46 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Svelte goes beyond the DOM
Svelte goes beyond the DOM Svelte is well loved as a web framework, and now the team is working to take it multi-platform. Svelte’s custom renderers API will enable native apps, terminals, and more to be build with Svelte.…

This evening, @paolo.ricciuti.me will join @codetv.dev's The Web Dev Podcast to talk about our work on the Svelte Custom Renderers Initiative! Check it out if you’re interested in rendering beyond the DOM with Svelte πŸ™Œ

Tune in at 18:30 CET πŸ‘‰Β www.youtube.com/watch?v=nPac...

#Svelte #SvelteKit

05.03.2026 09:17 πŸ‘ 20 πŸ” 5 πŸ’¬ 1 πŸ“Œ 0

yep. i see a lot of 'people i used to respect have been brainwashed by the AI cult' lamentations and it's frustrating to see the exact reverse phenomenon β€” extremely smart people whose critical faculties have apparently been neutered by seething hatred of anything AI-adjacent

05.03.2026 14:00 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

(unless it's not a new abstraction layer but something categorically different, which is an interesting topic but something I would have to be convinced of, I think)

03.03.2026 23:35 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Yeah that's fair. Totally possible that the job is much more about writing specifications in future (which, ugh, booooring). I'm just disputing the idea that this new abstraction layer will bypass the existing layers. That's not how abstraction layers work 😁

03.03.2026 23:34 πŸ‘ 4 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

certainly possible. rule no. 1 is that nobody knows anything. i just don't see any reason to actively prefer the hypothesis that agents will choose goop over the 'curated opinions' that they've already demonstrated competence with

03.03.2026 22:42 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

i would argue you're conflating 'more complex' and 'faster'. regardless: given limited context windows, and the availability of documentation and examples, agents operating within structures will always outperform free-range ones, even if the free-range ones could outperform humans

03.03.2026 21:39 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

LLMs are excellent pattern-matchers. Phrased differently: they excel when presented with patterns

03.03.2026 21:19 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

If the code is unmaintainable by humans but maintainable by agents, then there's an implicit premise: that agents aren't just faster/less fussy/less lazy than humans, they're superintelligent. To me it veers into magical thinking

03.03.2026 21:18 πŸ‘ 4 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

meta-frameworks provide two things:

1) missing pieces in the underlying technology/platform
2) structure/opinion

both are lumped together as 'DX' but they're almost opposite. 1) will indeed inevitably flatten. for you, 2) is a distraction, but for me that's where the real excitement begins

03.03.2026 20:29 πŸ‘ 13 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

au contraire β€” dropping next is _so_ contentious that vinext had to be created. that's because the structure/conventions/opinions (the bit that we/agents actually code against) are where the value lies. i don't think that's just 'stopgap' (though flattening is undeniably occurring)

03.03.2026 18:48 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

maybe i'm biased but i think the precise opposite is true. what agents crave is structure. left to their own devices they will happily generate unmaintainable goop. libraries? pfft they don't bother with libraries half the time, they just write their own

03.03.2026 17:10 πŸ‘ 39 πŸ” 0 πŸ’¬ 5 πŸ“Œ 2
02.03.2026 15:00 πŸ‘ 118 πŸ” 9 πŸ’¬ 0 πŸ“Œ 0

cc @paolo.ricciuti.me @jyc.dev β€” generally I've found them pretty good at figuring out non-interactive mode, but clearly it's not bullet-proof

01.03.2026 17:47 πŸ‘ 1 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0
kylo ren yelling MORE

kylo ren yelling MORE

26.02.2026 21:50 πŸ‘ 11 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

guessed why this had happened, checked, was not disappointed

26.02.2026 12:15 πŸ‘ 17 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

i would say that this genre of issue has now jumped the shark, but presumably that's also forbidden

25.02.2026 18:17 πŸ‘ 29 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0

Why do I have to pretend that I'm going to print something in order to save it as a PDF. Why do I have to engage in a little ruse.

23.02.2026 21:43 πŸ‘ 19302 πŸ” 2927 πŸ’¬ 345 πŸ“Œ 1
Web Platform Design Principles

we try! we often reference the W3C's Priority of Constituencies www.w3.org/TR/design-pr...

24.02.2026 01:20 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0