hu3

I understand there's an avalanche of vibe slop PRs.

But to be clear their AI instance is as clear-cut as can be. Their instance IS INDEED to "prevent AI-assisted code to reach their codebase".

> The following is not allowed in postmarketOS:

> Submitting contributions fully or in part created by generative AI tools to postmarketOS.

source: https://docs.postmarketos.org/policies-and-processes/develop...

Probably not and why would it need such constraint?

Not even humans can do that. Documentation needs to at least be reverse-engineered and understood before implementation.

I'm pointing out that their expectation of AI-free OS is pointless.

Because AI-assisted code is most probably already present in devices they use.

And I dare say that even for PostmarktOS:

1) There's no way they can prevent AI-assisted code to reach their codebase.

2) They will most probably change this policy in the future lest other forks/projects outpace them in terms of utility and they get reduced to a carriage in a car world.

Not parent commenter but this is bound to happen.

And I highly doubt iOS and Android are free from LLM assisted code at this point.

It's more like mentioning excavators in posts about the best vehicles to do groceries with.

Yep. It's like a crossfit vegan religion at this point.

You don't even have to ask. They will tell you and usually add nothing to the conversation while doing so.

Quite off-putting.

With the amount of money being thrown in R&D, I don't doubt the actual number is astounding.

Really? That's a lot of presumption and reductionism to LLMs enthusiasts.

People of varied ages, already leverage LLMs on a daily basis. And LLMs will only get better.

Yesterday, Opus did work for me that would have taken me weeks. And the result was verified with a comprehensive suite of unit tests plus smoke tests by myself. The code looks exactly as the rest of the code in the 10y+ old, hand-written, enterprise project, no slop.

And you actually should be afraid of being left behind in dev related fields if you don't use LLMs. In most areas in fact.

Once the market corrects for LLM assisted production, the expectations will raise. So right now there is a small window to leverage LLMs as a time saving advantage before it becomes the norm and everyone is forced to use it because expecttions will reflect that.

I suspect those who praise Gemini use it mostly for JS/CSS/HTML because that's where it shines for me.

For complex code I have been having using Sonnet/Opus as usual with a mix of GPT5.3-Codex.

It's still valid as as example to the language community of how to apply these optimizations.

> Strip markdown decoration: Headers (##), bold (*), italic (*), horizontal rules (---). These are visual aids Claude doesn't need.

In my experience, LLMs do understand header levels in markdown. So removing these is detrimental.

---

> Compress table formatting: Markdown tables with alignment rows and padding are verbose for what they communicate.

This is a good idea imo.

For all I know Linux started a hobby project from one person.

So entertain me if you disagree.

Isn't that how Linux started?

If they do, it could be because safety is a gradient and one variable among many in software development, albeit a very important one when it comes to browsers.

I wonder where did you get the idea that Firefox was all Rust. Made me curious.

I expected a coparison to `bun build` in the transformer TS -> JS part.

But I guess it wouldn't be an apples to apples com parison because Bun can also run typescript directly.

Maybe, but forcing code generation in a certain way could ruin hello worlds and simpler code generation.

Sometimes the user just wants something simple instead of enterprise grade.

Humans are nondeterministic.

So this is a moot point and a futile exercise in arguing semantics.

Not my experience. What doesn't work for you?

I use Opus 4.6 (for complex refactoring), Gemini 3.1 Pro (for html/css/web stuff) and GPT Codex 5.3 (workhorse, replaced Sonnet for me because in Copilot it has larger context) mostly.

For small tools. But also for large projects.

Current projects are:

1) .NET C#, Angular, Oracle database. Around 300k LoC.

2) Full stack TypeScript with Hono on backend, React on frontend glued by trpc, kysely and PostgreSQL. Around 120k LoC.

Works well in both. I'm using plan mode and agent mode.

What helps a ton are e2e playright tests which are executed by the agent after each code change.

My only complain is that it tends to get stutters after many sessions/hours. A restart fixes it.

$39/mo plan.

you can't really compare Karpathy with Einstein.

One of them is barely known outside some bubbles and will be forgotten in history, the other is immortal.

Imagine what Einstein could do with today's computing power.

it's because Apple blocks access to iMessage and other Appe services from non Apple os.

If you, like me, don't care about any of that stuff you can use anything plus use SoTA models through APIs. Even raspberry pi works.

$3k Ryzen ai-max PCs with 128GB of unified ram is said to run this reasonably well. But don't quote me on it.

Not to mention these people basically self declare as non-hireable or mark themselves a target on their back for the next layoff culling.

Terrible for career.

It's like complaining that power tools are bad because they promote industrialised products over hand made detailed work.

By looking at the code?

> And I asked LLM to show me the code for each answer, on all 5 engines.

They same way I would have done without AI, but it sped up finding the relevant parts to a velocity that made it viable in my limited time.

Doing things faster/easier means I now do most of these things whereas I didn't before.

Because I have limited time and energy. Take learning as an example:

I couldn't afford to spend a weekend learning the tradeoffs made by the top 5 WebGL JavaScript game engines AND generate the same demos for all of them to compare DX, and performance on my phone. And as I had more questions about their implementation I would have to scavenge their code again, for each question.

A sample of the questions I had (and as I asked it would suggest new question for things I didn't know I should ask):

- Do they perform sorting or are their drawing immediate? sort on z? z and y? z/y and layers? immediate'ish + layers? Frustum culling supported? What's their implementation for it if any?

- What are their GPU atlas strategies? fixed size? multiple with grouping by drawing frequency to reduce atlas switching? 2048? 4096? How many atlases? Does it build the atlas at boot or does it support progressive atlas sprite loading? How does it deal with fragmentation? What does it use for packing algo? Skyline ir something more advanced? How is their batch splitting behaviour and performance characteristics?

- Does it help with ECS? How is their hierarchical entity DX, if any? Does it math with matrices for transformations or simpler math? Shaders support? Do they use an Uber shader for most things? And what about polygons? Also, how do they help with texture bleeding? What's their camera implementation? Do they support spatial audio?

...and so on. Multiply the number of questions by at least 10.

And I asked LLM to show me the code for each answer, on all 5 engines.

This kind of learning just wasn't feasible for me before with my busy life.

So when I say "easier" it often means "made possible".

Finally let's not forget most of us in HN are incredibly privileged and can afford to learn futile things on the weekend. But for a great part of the less privileged population, having access to easier learning is LIFE CHANGING.

- learning: easier

- searching: better

- photo edit/enhance/filter: easier and acessible

- text summarization: better

- quick scripts/tools: faster

- brainstorming/iterating ideas: faster

- generating list of names: faster

- rephrasing text: better

- researching topics: faster

- stackoverflow: i'm finally free. won't be missed by me

- coding: debatable but for me LLMs made possible projects that weren't before due to scope or lack of expertise

Folks around me use mostly Instagram which ironically is also from Meta.

Zuck is always one step ahead.

it's toxic because of how it sometimes materializes:

- predictable

- incessant

- with belittling/reductionist/arrogant/elitist/combative phrasings

- handwaving rust shortcomings and tradeoffs

I see a lot of game devs in discord experimenting with Electrobun to release desktop games.

I think it's going to eat a piece of the Electron pie for Steam indie games.

Most stay with bun after seeing how fast and seamless it is to run typescript games with instant auto reload:

bun --watch game.ts

Out of their way?

They have been watching talented folks waste their lives away reverse engineering hardware/software that they possess all the schematics of.

If they really wanted to help, all they had to do is send a single e-mail with a zip file.

The distortion field is unbelievable.