My biggest frustration with app dev currently, by far the worst possible scenario is happening

_brian on X: "You can just build cross platform apps in C https://t.co/FdV2u4bj41" / X You can just build cross platform apps in C https://t.co/FdV2u4bj41

Yeah on twitter someone posted a neat example and a shout out to the repo
x.com/_brianpmaher...

People in terms how they interact with each other maps pretty well to how they interact with chat bots, so it's very easy to fall into this trap, part of it is how these bots add flair to their messages to make them seem more human..

With remote functions it's going to be hard to justify not using it.

To me, I see it from the joystick pov, when I move the joystick up it's y+, right x+
And z is elevation, I noticed it's mostly a taste thing, but most important is to be consistent with it

docs.swmansion.com/TypeGPU/fund...
Praying for the day this becomes the norm to writing shaders instead of the mess we have to go through

And this is where my pain point with the meta conversation comes in, I can't tell what is a real argument and what is just hype stuff, I don't mind a serious conversation about the issues we might face with llms, but I can't tell if these conversations are just hype/attention grabs or genuine.

And when look at llms as a store of knowledge rather than the dictator of information, it actually becomes much easier to understand, right now if you Google frontend learning course most likely you'll see react courses more so than anything, in that way it's no different to llms defaulting to react

With these people ofc llms would be ground breaking, but it'll help them leap to what? And for how long can they rely on it before going back to basics, and if so can llm help them in that? It's at best training wheels and a modern search engine, using it as the end all be all is silly and dangerous

Will it change something? It'll have an impact similar to react on the lower end, currently at my company we're looking for interns, and every person we interview doesn't understand the underlying ideas, they understand react, concepts like if statements and for loops don't register to them

Every time a big thing happen and people view it as like "the thing that makes most sense, therefor why evolve?" but in reality every 10 15 years a big jump happens and we can describe what we want in a better approach, llms regurgitating react code is not "better description".

and already alot of what llms suggest is bad for these evolutions, the fact is that llms will freeze knowledge at the lower end of the skill distribution, but largely it won't change the trajectory, it is the same way jquery was seen as the end of web dev, react would be the end of it all etc etc

My coworker argued this with me but honestly the merit of this argument is kinda weak, react it self is going through two evolutions with rsc and the "compiler".

Checkout type gpu, they're building probably the thinnest and type safe webgpu abstraction layer I've seen docs.swmansion.com/TypeGPU/
There is hardly any documentations but the samples can get you somewhere.

This is generally my view on it all outside of the conspiracy drama, which obviously were huge back then.

A third and final problem is that while rsc on its own is nice, it did not fit typical software departments work environment, where there is a static frontend and a separate (usually even different lang) backend, and two devs working on a slice, this setup made it hard to accept rsc as a sale pitch.

A secondary problem was the general vibe of "clique", it felt like alot of what I'm seeing around rsc is stuff that is being discussed offline, and I unfortunately don't have the access to these conversations so I didn't understand the reasoning behind them, it is cool tech though.

Personally my main problem was not knowing which is specifically a react behaviour and which is nextjs behaviour, it's cool that vercel crafted such a situation, but man it became extremely hard to know who caused what problem, the early months of rsc were brutally unstable..

An introduction to Shader Art Coding YouTube video by kishimisu

Its fine in some cases, alot of people with bad accent rely on it with content creation like here
youtu.be/f4s1h2YETNY?...
The creator is French and his accent is very heavy, so he uses ai to make it easier to understand, I feel like alot of people are reflexively anti ai which is very short sighted

Amazing work,

Side note but I wish the desktop story in react native was better, the prospects of using skia in desktop environment is so nice for small game dev

Love sokol so much, it's such a well thought out api

Woah.

I played with zero and got it working on expo, and it's great. My only issue is the postgres requirement. Otherwise, this might be the most straightforward way to achieve a great local first development environment.

Yeah, at least here locally jobs are everywhere, but the way companies use ai to expedite the hiring process ironically made it a lot worse.

Thanks for the reminder.

`Error.isError` to stage 3, per 2024.12.02 TC39 · tc39/proposals@6db6ffb

`Error.isError` was just approved for stage 3 in today's @tc39.bsky.social!

- github.com/tc39/proposa...
- github.com/tc39/proposa...
- npmjs.com/error.iserror

Agreed, but it is sad that services like archive are paying the fee for webscrapers that are doing it for free..

Thanks Tom!

It's crazy, whole year felt like a week