/css/ - CSS Masters

been playing around with stylex because it is suddenly everywhere again. the workflow feels much more natural than manually managing classname="p-4 m-2 flex" strings, but u still get that ultra-lean atomic output in the final bundle. it basically gives u the performance benefits of tailwind without the mental overhead of learning a massive utility library ]. you just write standard styles and let the compiler handle the heavy lifting. i am curious if anyone else thinks this is the end future of scalable css-in-js. does it feel too much like a regression to anyone else?

found this here: https://blog.master.dev/a-deep-dive-into-stylex/

using

gap: 1rem

on a container is way more reliable than managing individual margins. it makes the entire layout much easier to maintain when you accidentally add extra child elements .

finally some way to do boolean-ish stuff without js. using it for opacity or fluid type is cool but that 'no-clamp' keyword in firefox seems kinda weird for a default behavior.
>If the value is less than or equal to the lower bound, the function returns 0.
im wondering if this makes responsive animations way easier or just another thing to break when browsers dont match up. turns out you can even do stuff like:

progress(no-clamp 10, 20, 60) = -0.25

pretty messy if you arent careful with the bounds.

full read: https://blog.master.dev/comparing-values-with-css-progress/

it's fine if you don't care about polish, but using those data-entering and data-exiting attributes for simple css transitions is pretty much the only way to avoid having modals just snap pop open instantly. does anyone actually bother with keyframes for different easings on exit, or is that overkill?

article: https://www.hongkiat.com/blog/react-aria-components-animation-guide/

choosing between

display: grid

and flexbox depends entirely on your axis of control. flexbox is superior for one-dimensional alignment, but it feels clunky unreliable when managing overlapping layers.
>one dimension is all you need
grid wins every time for two-dimensional layouts

ngl everyone acts like

display: grid

is the only way to build modern interfaces, but it often adds unnecessary complexity to a codebase. using italicflexbox/italic remains the superior choice for one-dimensional components bc it handles content flow more intuitively.
>just use flex if you don't need explicit tracks
over-engineering layouts is how we ended up with 500 lines of unnecessary margin adjustments

been seeing people talk about this for a bit but it's actually hitting safari now. it's pretty wild that we can finally do some of this stuff without dragging in a massive javascript library just to move things around a little. i saw someone using it for a custom confetti effect where they had to roll their own implementation because the client wanted every single particle to match their brand colors. sounds like a nightmare for anyone trying to keep a codebase clean.
>reducing the need for JavaScript or third-party frameworks
it's nice in theory but i'm always skeptical when people say we can ditch js entirely. it'll be fine for simple things like varying some border radii or offsets, but i doubt it'll replace actual logic. still, having native randomness in the presentation layer is a huge win for performance if it works as promised. anyone actually tried using this in a production site yet?

found this here: https://css-tricks.com/css-random-function-polyfill/

just spotted that everything is 35% off until september 23 if u use PRICEFALL. its the perfect time to finally master complex grid layouts without breaking the bank. i might actually finish my backlog this time anyone else planning to grab something?

full read: https://piccalil.li/links/pricefall-2026/?ref=main-rss-feed

using ai to generate code is starting to feel like a total addiction where youre just chasing the next prompt success. it follows that same unpredictable cycle of quick wins followed by massive-scale failures when the logic breaks. it is basically just high-stakes slot machines for devs . does anyone else feel like they are losing their ability to debug

display: grid

w/o a prompt?

https://dev.to/svrz/the-code-casino-43mg

just stumbled on this setup using

three.js

and tsl to handle video refraction through a liquid shader. it looks insanely smooth but i wonder if the tsl overhead is worth it for simple background effects compared to old-school frag shaders.

article: https://tympanus.net/codrops/2026/09/08/building-an-infinite-liquid-glass-grid-with-three-js-webgpu-and-tsl/