HNHacker News
TopNewBestAskShowJobs

typpilol

246 karma · joined July 30, 2025

Network Engineer. I build random stuff. https://github.com/Nick2bad4u

[ my public key: https://keybase.io/nick2bad4u; my proof: https://keybase.io/nick2bad4u/sigs/To1plwGDQHjbNElo1gY76u5Q-Rij4z7hEA4WfAdq03o ]

submissionscomments
typpilol··on Zed is now available on Windows
How can people justify using editors like this when they are missing basic features?

Like what's the benefit?

typpilol··on Pwning the Nix ecosystem
Their leadership and community is also a disaster
typpilol··on Zed is now available on Windows
No lol
typpilol··on New lab-grown human embryo model produces blood cells
We're growing humans ?
typpilol··on Show HN: CSS Extras
Don't forget the people that have to tell us how much they hate JavaScript on every single web dev post too
typpilol··on LLMs are getting better at character-level text manipulation
I asked this in another thread and it would only be better with unlimited compute and memory.

Because without those, then the llm has to encode way more parameters and way smaller context windows.

In a theoretical world, it would be better, but might not be much better.

typpilol··on Palisades Fire suspect's ChatGPT history to be used as evidence
I still haven't once talked to an LLM for personal reasons. It's always been to get information.

Talking to an LLM like a human is like talking to a mirror. You're just shaping their responses based on what you say. Quite sad to see stuff like the "myboyfriendisai" reddit

typpilol··on DDoS Botnet Aisuru Blankets US ISPs in Record DDoS
Yup. Spectrum is Michigan will give you up to 2gbps down but not anything more than 200mbps up
typpilol··on NanoChat – The best ChatGPT that $100 can buy
Why hasn't anyone made a tokenizer that's 1 character per token. Is it because it requires an insane amount of compute?

Or would the loss of efficiency make it dumber then modern tokenizers?

typpilol··on NanoChat – The best ChatGPT that $100 can buy
Won't work at all. Or if it does it'll be so slow since it'll have to go to the disk for every single calculation so it won't ever finish.
typpilol··on MAML – A new configuration language
I can't decide between yaml and json5/json

It seems like we will be forced to use both forever though

typpilol··on Meta Superintelligence Labs' first paper is about RAG
Tell me more
typpilol··on Microsoft Amplifier
You're not wrong.

I'd never encourage anyone to blind commit the messages But if they are correct they seem a lot more useful than 90% of commit messages.

I found the biggest mistakes that I've seen other people do are like - they move a file, and the commit message acts like it's a brand new feature they added because the llm doesn't put it together it's just a moved file

typpilol··on Microsoft Amplifier
I'd rather have in-depth commit messages then three word ones
typpilol··on Fears over AI bubble bursting grow in Silicon Valley
Especially when you see Bitcoin and how their adoption rate is

Crypto was / is a true bubble Lots of paper value but hardly any adoption

typpilol··on They Don't Have the Money: OpenAI Edition
Their moat is 700m active users.

What's facebooks moat? There's tons of social media sites. Facebooks moat is the 3b users.

This comment is so idiotic it's starting to annoy me.

"WHaTs tHe MoAt" for a company with almost 1b active users

typpilol··on Does our “need for speed” make our wi-fi suck?
Really? My spectrum has been super reliable in Michigan. Way better than when I had Comcast here
typpilol··on Does our “need for speed” make our wi-fi suck?
The reality is that most people only have a single cord coming into the house

So they would have to do quite a bit of work to run cable. Also people living in apartments that cant just start drilling through walls.

I'd say most ppl use wifi because they have too, not pure convenience

typpilol··on Embracing the parallel coding agent lifestyle
Co-pilot has a new execute prompt tool in preview that lets it spin out its own requests to LLMs
typpilol··on Embracing the parallel coding agent lifestyle
Typically people use sub agents from what I've seen to work in different parts of large code bases at a time.

If your hitting merge conflicts that bad all the time you should probably just have a single agent doing the work. Especially if they're intertwined rightly

typpilol··on Embracing the parallel coding agent lifestyle
The best way to get decent core I've found is test suites and a ton of linting rules.
typpilol··on Reasoning LLMs are wandering solution explorers
Makes a lot more sense when you replace reasoning with generating more context increase the chance of a correct answer
typpilol··on LLMs are mortally terrified of exceptions
Weird thing I've noticed.

Some LLMs can output nerd font glyphs and others can't.

If I recall grok code fast can but codex and sonnet can't

typpilol··on Designing a Low Latency 10G Ethernet Core (2023)
True. If you put in the whole system maybe.
typpilol··on Kurt Got Got
Edge has some basic typo squat protection
typpilol··on Two things LLM coding agents are still bad at
Ask a model to show you the seahorse emojii and you'll get a storm of "but wait!"
typpilol··on Tonight's restaurant dinner fell off the Sysco truck
I'm genuinely confused at the parent comment, and I don't understand how they came away from the article with an anti union sentiment lol
typpilol··on The least amount of CSS for a decent looking site (2023)
Runtime polymorphism is great but I wouldn't consider that to be tree shaking.
typpilol··on Gemini 2.5 Computer Use model
I've used it a few times with vscode.

Although I find the electron mcp is better for what I'm doing at the moment.

typpilol··on Designing a Low Latency 10G Ethernet Core (2023)
I feel like it's also got to be super stressful

Imagine getting a call you system is down and is costing them millions per hour

← PreviousPage 4 of 13Next →