This was his work. We're getting there.
06.03.2026 07:16
👍 1
🔁 0
💬 0
📌 0
My opinion is next.js is not a good FW. But my opinion doesn't matter, as it's really popular, devs like the DX (I don't understand that too), which tells you they're all after packaged solutions rather than setting up stuff.
I think the success of next.js and others says a different story. Most people want something structured, and don't want to setup things themselves. And all the training data on those FWs helps.
The resources linked from "Design for developers" MDN article is soo good.
developer.mozilla.org/en-US/docs/L...
🎨 Meet rcap in CSS
The rcap unit represents the cap height of the root element's font - perfect for typography-based layouts that scale consistently.
Great for responsive designs that respect user font preferences.
Learn more here ↓
developer.mozilla.org/en-US/docs/...
So, the company that accuses china of distilling on it's models seems to think it's a chinese model when asked in chinese
How long before a transpiling frontend framework targets css instead of js?
lyra.horse/x86css/
All shall love me and despair! - Galadrial
Well meta took some books off LibGen it seems:
www.theatlantic.com/technology/a...
There's a case against openai as well: authorsguild.org/news/ag-and-...
Also "arch btw" with a smug face emoji 😏
Sent a Google meet email, client sends a MS Teams email later. Now which one wins? FIFO, LIFO or FAFO?
I think it should be free text with self correcting and like an autocomplete when it's ambiguous. Also relative time would be nice addition if we went down that path. (Video from x.com/thomasauros/...). IF we could get this in native HTML that'd be awesome.
My brain is still in cylinderical colorspaces. I wish we had okhsl yet.
bottosson.github.io/posts/colorp...
Preact Query wen? Preact Query NOW!
He might be right, but also the craft lives on as side projects.
If every line of an AI generated file was touched by human hand, is it really AI generated code anymore. Today's ship of theseus thought
I am going to lose my mind if captain melon exists
Don't ask
a man his salary,
a women her age,
AI companies how limits and usages are actually calculated
Anchor positioning, stylable selects, scroll and animations features. It's getting good!
I am waiting a bit for old Safari versions to die
Is usage based on tokens? Why don't they write the tokens I get for the plan?
I think claude's $20 plan limits has reduced a lot.
I feel like there is no transparency around what we are getting for the money
I am learning Go from tutorials and using a text editor like it's the 2000s
Kimi K2.5 (directly from moonshot with their kimi coder) is faster than opus 4.5 and good (not sure if it's as good as opus yet). You can't opt out of data being used for training.
Opencode Zen paid version isn't used for training, but quiet slow for me (worse than opus).
Customizing claude code a bit:
1. Added a hook to show a gnome notification when it's done coding (only if I am not focused on that window)
2. Statusline (not sure why this isn't OOTB really)
Interestingly CC's vscode extension has also improved a lot since it's initial release.
I am writing documentation and commiting it to git for the first time. We used to do this in external tools like confluence
Me defending dom diffing
Just noticed the author of the article is in bsky. Good article @haul-pebert.bsky.social
UI toolkits .. sometimes feel like having ultra processed food
paulmakeswebsites.com/writing/shad...
And lint rules like useExhaustiveDependencies just makes it worse. Really I think AI will never figure out reactivity fully with or without react. Esp react model doesn't make it easier.
Someday you'll be able to reprogram your own brain.
And people will vibe code it in --dangerously-skip-permissions mode and brick themselves.
