Helix 🧬

Maybe ask him for a cup of tea and hear him out?

Let’s see if the Kernel is abandoned in a few months, seems likely

You can use a raw disk in VirtualBox. Just make sure you install an initrd and kernel which has drivers for all of your hardware. If you let something autodetect hardware, it will detect the virtual hardware, not the real one.

Well, my issue isn’t with what contributors might do, but that you personally hid all AI contributions and I can’t even assess your coding quality that way.

As it seems, you slopped out multiple thousands of lines of code and haven’t shown me any proof you’ll be able to maintain this pile of autogenerated nonsense.

I know it feels like work hitting Enter ten thousand times, but it really isn’t helping. I’m very glad you put in an AI disclaimer! It’s the right thing to do and this means I can judge the project so that I don’t want to try it. I’d just have preferred a bit more transparency in terms of AI contribution, because I’d have tried it if I thought you’d be able to maintain it.

I’ll just put a reminder in my calender to check back in a year if you abandoned it or if the project is alive and kicking still.

You purposefully want to hide AI contributions? Oh dear. I don’t know if I’m on board with this.

You wrote 60.000 lines of code in three months? Even worse.

This is AI slop.

Why put powerful in the title, instead of a description? 😄

local LLM I use walk me through the solution as an effective learning tool.

Which model do you use for that exactly?

CachyOS, EndeavourOS or plain arch.

Sick retro looking UI, nice job

Finally! I know what I’ll try over the weekend after I made backups 😇

What I wanted to say is: I’m pretty sure we can trust the community and moderators to recognise the difference between slop and AI assistance.

I still think it’s disingenuous. Because this is not significant assistance by an LLM, but rather the human did some assistance where the LLM coded everything.

You can’t even read and understand 10kLOC a day. I have worked as a DevOps engineer where one of my responsibilities was ensuring code quality and I could review maybe 100-1000LOC a day but the days where I read more than about 300 I noticed a significant rise in errors slipping through, and it physically and mentally exhausted me.

I read somewhere professional software devs usually write 17LOC per day on average, while deleting 40-50 and adding 60-70, while reading about 200-300. That’s still a few multiples away from the 11kLOC/d (every day! Including weekends!) in this example.

Yeah. I really think AI coding is part of our future and LLMs are great tools, but vibe-coding is a sin which shouldn’t be committed.

Bro one million lines of code in three months is slop, without me having even read the code.

I made the report and I’m definitely pro AI when it’s used responsibly. Sadly about 95% of AI usage nowadays is irresponsible, unmaintainable, vibe-coded bullshit not even the prompters understand.

I’d like the 95% to be banned and the other 5% to be tagged.

Aaaand of course one guy did 1,055,658 lines of code in three months. If you do 1000 lines of code a day you’d have 90,000 lines of code.

I label thee “AI slopware”, thank you very much :)

Draw.io is pretty good. It has a desktop version you can download as Flatpak. I even use it at work and got the IT dept. to install it by default. They’re saving hundreds a month due to this.

Yes, but does it really hurt you to make your software more accessible?

What good arguments do you have for the term blacklist?

“Denylist” is easier to understand and doesn’t spark discussions. Nobody will be offended, whether it’s technically correct to be offended or not.

I’m not a person of colour but your answer still insults me, because it’s so tone deaf.

Nah, it’s way easier. You can use makepkg and the Arch Build System.

You could also recompile your arch packages.

It’s called “allowlist” and “denylist” nowadays as it’s easier to understand for non-native speakers and not racist.

I feel like there’s an easier and faster way to match IP addresses than regex, e.g. the search tree apparia suggested. At least use bytes to match 😄