mb21
@mb21.bsky.social
Hacking on https://mastrojs.github.io – a static site generator that runs in your browser, and accompanying web dev intro course.
https://github.com/mb21/
https://github.com/mb21/
Regarding the medium web, can recommend this classic: alistapart.com/article/dao/
A Dao of Web Design – A List Apart
Web designers often bemoan the malleable nature of the web, which seems to defy our efforts at strict control over layout and typography. But maybe the problem is not the web. Maybe the problem is …
alistapart.com
November 13, 2025 at 7:07 AM
Regarding the medium web, can recommend this classic: alistapart.com/article/dao/
Their point is that this approach would have avoided 20 different variants. Sure, if your process is just to move pixel for pixel from figma/photoshop to css, then tailwind will suit you better.
November 13, 2025 at 7:04 AM
Their point is that this approach would have avoided 20 different variants. Sure, if your process is just to move pixel for pixel from figma/photoshop to css, then tailwind will suit you better.
It’s saying you should primarily style html elements. And then you can add some utility classes for exceptions etc.
Not to add one class for each css property-value combination on every html element on your page, like tailwind does.
Not to add one class for each css property-value combination on every html element on your page, like tailwind does.
November 12, 2025 at 8:00 PM
It’s saying you should primarily style html elements. And then you can add some utility classes for exceptions etc.
Not to add one class for each css property-value combination on every html element on your page, like tailwind does.
Not to add one class for each css property-value combination on every html element on your page, like tailwind does.
You’re also using Servo?! 🙌
November 11, 2025 at 5:03 PM
You’re also using Servo?! 🙌
Sounds like something old-school enterprises would think is more secure than the latest ever-green although probably it’s not? Either way, no compatibility data on caniuse makes it somewhat hard to support.
November 1, 2025 at 7:20 AM
Sounds like something old-school enterprises would think is more secure than the latest ever-green although probably it’s not? Either way, no compatibility data on caniuse makes it somewhat hard to support.
Oh, interesting! I have to admit, first time I hear of it 😬Doesn’t it have the same browser engine with the same features as regular Firefox?
I usually look for iPad Safari as the lowest one to support.
I usually look for iPad Safari as the lowest one to support.
November 1, 2025 at 6:50 AM
Oh, interesting! I have to admit, first time I hear of it 😬Doesn’t it have the same browser engine with the same features as regular Firefox?
I usually look for iPad Safari as the lowest one to support.
I usually look for iPad Safari as the lowest one to support.
That and too many dependencies.
October 30, 2025 at 6:54 AM
That and too many dependencies.
And not only build systems! All programming! ✊
October 28, 2025 at 7:14 AM
And not only build systems! All programming! ✊
Maybe it’s the browser window that’s not level?
October 27, 2025 at 7:50 AM
Maybe it’s the browser window that’s not level?
Very far. But only because I haven’t touched the CSS of that static site for basically forever.
October 26, 2025 at 6:39 AM
Very far. But only because I haven’t touched the CSS of that static site for basically forever.
Sass and SCSS were awesome. Thanks for the service!
But nowadays, unless you need to support Safari <17, you get almost the same features with native CSS.
But nowadays, unless you need to support Safari <17, you get almost the same features with native CSS.
October 25, 2025 at 7:12 PM
Sass and SCSS were awesome. Thanks for the service!
But nowadays, unless you need to support Safari <17, you get almost the same features with native CSS.
But nowadays, unless you need to support Safari <17, you get almost the same features with native CSS.
OpenAI’s new Atlas “browser”
October 23, 2025 at 10:09 AM
OpenAI’s new Atlas “browser”
Agree Dan writes well. Although he still didn’t convince me about RSCs.
October 19, 2025 at 12:29 PM
Agree Dan writes well. Although he still didn’t convince me about RSCs.
But I doubt it makes a measurable difference unless you have a huuuuuge page with loooots of elements in the affected layout. In the real world, JS is taxing the CPU _way_ more.
October 19, 2025 at 9:28 AM
But I doubt it makes a measurable difference unless you have a huuuuuge page with loooots of elements in the affected layout. In the real world, JS is taxing the CPU _way_ more.
Don’t have a link, but I remember Jen Simmons saying that with the two passes on the socials somewhen.
October 19, 2025 at 8:55 AM
Don’t have a link, but I remember Jen Simmons saying that with the two passes on the socials somewhen.
If you want to rationalize using flexbox, you can always say that it requires only one pass by the layout engine, whereas grid requires two. Then proceed to load in 1MB of JS…
October 19, 2025 at 8:32 AM
If you want to rationalize using flexbox, you can always say that it requires only one pass by the layout engine, whereas grid requires two. Then proceed to load in 1MB of JS…
The idea with JSR was to put a bit of pressure on them, right? Unfortunately, not sure it’s working 🫠
Although personally, I’ve been quite happy with JSR.
Although personally, I’ve been quite happy with JSR.
October 17, 2025 at 5:25 PM
The idea with JSR was to put a bit of pressure on them, right? Unfortunately, not sure it’s working 🫠
Although personally, I’ve been quite happy with JSR.
Although personally, I’ve been quite happy with JSR.
x86 assembly isn’t structured enough for you? granted, I also favour RISC IAs…
October 15, 2025 at 9:11 PM
x86 assembly isn’t structured enough for you? granted, I also favour RISC IAs…