NerdNextDoor 6d ago

EDIT: This might be a better idea. https://mastodon.social/@mrmasterkeyboard/116309403462425727

FUCK IT, let's just fucking fork OpenSolaris's last code. Best way to get away from AI/LLMs.

It's already written for us, it's a good OS and kernel and honestly it's an awesome relic of computing.

cc: @thing

Show threadNerdNextDoor 6d ago

@thing or something like that, forking another OS without any AI stuff is the best option instead of writing our own from scratch. That would take years. Sure, both options are fun but one is less painful from my experience.

Doesn't have to be OpenSolaris but it's one of the best options we have. I don't exactly want to pull an EVi on FreeBSD or any of the others currently.

Show threadhazel 6d ago
@mrmasterkeyboard @thing i saw somewhere someone proposing that we could fork linux instead of a nother kernel. it's probably a good idea and would allow us to cherry pick commits from mainline

if something comes out of this discussion, i probably wont have enough time to notice, consider me in, im already planning to make a distro loosely based on chimera linux
Show threadNerdNextDoor 6d ago

@hazel @thing Yeah but Linux is millions of lines with thousands of contributors and maintainers... it's not viable at all unless we somehow group a bunch of Linux contributors or something. Even then, we don't entirely know if AI code slipped in at some point into Linux. Best bet would be 2024 as that when AI started becoming more coherent in code.

That's why elsewhere in this thread, I proposed making our own kernel and OS with our own software or projects that are against AI code.

Show threadReid 6d ago
@mrmasterkeyboard @hazel @thing starting from scratch is, to say the least, a nightmarishly gargantuan task. Making a UNIX kernel in and of itself isn't *too* outlandish, but supporting hardware with it is. OpenBSD and Haiku have decades of development and strong communities behind them, yet hardware support is still pretty hit or miss for both (for both you have to cherry pick the hardware you intend to use them on, and Haiku doesn't even have 3D acceleration on *any* graphics card)
Show threadNerdNextDoor 6d ago
@reiddragon @thing @hazel well, it's still fun though. not to mention, it's a great way to start unslopping our computers.
Show threadReid 6d ago
@mrmasterkeyboard @thing @hazel you're not really unslopping anything if the OS can't actually be used as a main OS on real hardware
Show threadNerdNextDoor 6d ago
@reiddragon @thing @hazel yes, but still it's a step in the right direction. honestly, I'm just desperate just to try and avoid AI and go back to how things were.
Show threadLixou 🏝️6d ago
@mrmasterkeyboard @reiddragon @thing @hazel
If you're really trying to unslop stuff (which, with your position about not using any of LLVM because of ai contributions, seems pretty clear), I don't think you'd have high hopes of "normal" hardware. You'll have specs and linux kernel impl as reference which could also be slop, even the card firmware maybe. If you want to unslop, you'd need to focus on stuff like open source riscv chips and hardware.
Show threadNerdNextDoor 6d ago
@lixou @reiddragon @thing @hazel I was honestly considering RISC-V.
Show threadthe thing 6d ago
@mrmasterkeyboard @lixou @[email protected] @hazel longterm, we need open source both hardware and software anyways.
Show threadNerdNextDoor 
@thing @lixou @reiddragon @hazel yes, we do, and non-slop ones too, hence my idea.
Mar 29 at 1:38pmWeb
Show threadLixou 🏝️6d ago
@mrmasterkeyboard @thing @reiddragon @hazel like we have nonslop spade as HDL (and many others, I'm just biased hehe), but like.. non slop optimizers and languages which could ideally be formally verified pretty easily.. 🥀