Joel Drapper 🇬🇧🇺🇦

This looks like it will reduce battery life by about 37%.

Brilliant. I’m going to try a bunch of my libraries with it now Zeitwerk is supported. ❤️

Good idea. I can see why you might want something like this, especially if you can’t get that value at boot time for the constant. But it definitely adds a lot of confusion. Not worth the cost.

I would get this optimisation by assigning an interpolated regex to a constant (without the /o). Then you can re-use the same instance again and again with the constant interpolated content.

That feels like a good mental model for what /o is doing, but with significantly more indirection.

Agreed. I’m working on it. Will have more to share soon.

Because it allows you to name small procedures and compose them together, allowing you to think at a higher level of abstraction.

Do you have 15 methods that take `(name, address, number, state)` or do you make an object?

If you’re going to make an object, you could have the methods on the module pass that object around, or you could just put all the methods on the object itself. That object is the service object.

Okay, but now have that method call 15 private methods that all need to work with the `name`. Sure, you could just put all those methods on a module, but what if you need to pass `name` and `address` and `number` and some state that needs to track from one method to another?

screenshot of the release PR for SvelteKit remote functions

we just released SvelteKit remote functions. feel like 'minor change' undersells it a little tbh

✅ await in components svelte.dev/docs/svelte/...
✅ remote functions svelte.dev/docs/kit/rem...
🔜 async SSR

truly can't wait to ship the rest of this stuff

Incredibly sad news. Piotr was brilliant. 💔

Because React is badly designed in my opinion. I typically use a web component (custom HTML element) to mount small svelte components. But I think Inertia looks great if you can go all in on a Svelte SPA.

I’ve noticed this too but I prefer Bluesky.

I’m assuming it’s a recording studio.

It’s amazing. 🤩

Link?

I think Stimulus may have been somewhat useful before web components and before some of the new HTML/CSS capabilities were available.

jQuery was also useful before querySelector and Flash was useful before <video>.

Its time has come and gone.

7. You don’t need Stimulus for “sprinkles”, since custom HTML elements allow us to easily attach behaviour to elements.
8. You don’t need Stimulus for popovers, dialogs, drop-downs, details-reveals, carousels, toggles, etc. You can do all this with just HTML and CSS now.

1. Stimulus is larger than Svelte.
2. You need to write more code in Stimulus.
3. Stimulus has no templating support.
4. Stimulus has no state management, so you need to handle callbacks by hand.
5. Stimulus cannot coordinate asynchronous state.
6. Stimulus is procedural, Svelte is object-oriented.

Basically as soon as you need state or templating, and probably as soon as you need to make a fetch request.

With popovers, dialogs, details, invokers, checkboxes, labels, and the CSS `:has` selector, there’s a ton of stuff you can do with HTML and CSS only.

Show me a Stimulus controller and I will rewrite it in either vanilla HTML/CSS/JS or Svelte.

The result will be smaller, faster, more maintainable, more flexible and easier to understand. The JS bundle will be smaller too.

I cannot understand why anyone uses Stimulus today. It provides no value.

Introducing the Herb Linter, Formatter, and a Vision for the Future of Rails Views Marco Roth - marcoroth.dev

Love what Marco is doing to push the ecosystem forwards. marcoroth.dev/posts/rails...

The new Phlex beta (2.4.0.beta2) now has a significantly improved compiler and now uses source maps to map exceptions back to the original source location. github.com/yippee-fun/p...

I will always recommend using Mise and OrbStack in the same way I recommend using Vim motions.

**Eventually.**

But we need friendlier tools for beginners and this Mac app looks *amazing* for that. Great work @afomera.dev 👏

Tricky command line interfaces shouldn’t be a blocker.

Something like this

You can already use a very small (maybe 10 lines) web component instead of turbo mount. Then it feels native to HTML.

This is working really well. It’s much nicer than having to keep two separate packages in sync.

Wow! It only took them a decade or so to add three lines of code to actually respect user tab preferences. They had the setting, just ignored it for the last 10 plus years.

You know it’s almost September, right?

The reason for doing this was I have a package that includes Ruby, TypeScript and CSS that I want to install all at once.

I installed a Ruby gem via NPM. How’s your morning going?