raven-uxn now has an x86-64 assembly backend (yay!)
and it's about 2x faster (yay!!)
and the first draft was written by Claude (booo!)
and then I rewrote most of it, which made it even faster (yay!)
and introduced a memory corruption bug (booo?)
which Claude is better at debugging than I am (.......?)
--------
if you too have complicated feelings about our robot buddies, you may enjoy my writeup:
@mjk > I would have made fewer changes to a human PR, because I'm sensitive to completely ripping up someone's work.
I too find that's a huge difference from the way I approach regular PRs. If someone comes with a LLM-assisted PR and there's anything in it that makes my life harder as a reviewer or maintainer, I respond without regard for churn or seemingly gratuitous changes. These things don't have feelings or memory.
@mjk Oh no, you were said to destroy the Siths, not join them! :/
Jokes aside, I’m trying to figure out why I feel so much despair when my programming heroes adopt this tech. I think a difference might be ”the reason why we create projects”, is it to realize an idea or to do the work/process? Maybe. I would rather not have a x86_64 backend in my project than to use this tech. But.. I suspect you would not have paid someone else the same amount you paid big tech now to implement the backend before LLMs were good enough? LLMs makes it feel like you did it yourself to a higher degree than delegation to another human would? To me it feels like, what’s the point in doing a project if you’re not doing it, might as well play guitar or something.
I’ll make sure to read the post nonetheless!
@gustav It's a good question – if someone had come to me and said "I'll implement a rough draft of the x86-64 backend for $35", I'm not sure how I would have reacted.
And there's also the question of whether the vibes be different if I did this experiment with a local model, instead of paying Anthropic?
I care about the craft of programming and quality of the resulting code, which is why I spent days rewriting the initial implementation (git shows 924 insertions(+), 1504 deletions(-) on a 2351-line file), but I wouldn't have gotten started without the AI boost – too much else on my plate, and the barrier to getting started was high...
Matt Keeter
Paul Khuong
gustav