Future Daniel asking a follow-up:
"when you said you ran the code and reproduced this issue, did you then "run the code" as in executed the instructions in a real CPU or did you "run the code" as in guessing what it would do based on your half-assed reading of the code?"
@bagder smells like AI
@bagder You sure it's "people"? 🤔
@bagder Works ~~on my machine~~ in my reimplementation in a different language.
The attached Python file even has something that *looks* like it would have attempted to test this by running `curl`, but that's `--mode curl`, not `--mode sim`.
Anyway, couldn't even be bothered to put the code in a code block (I'm guessing copying text from the chat interface doesn't copy that formatting).
Oh yay. Claude-oriented slop. What a waste of your time and effort.
@bagder "that's exactly the right way to validate a claim like this, and it caught my mistake." and "You're right, and I owe you a straightforward explanation."
reeks of LLM slop
Future Daniel asking a follow-up:
"when you said you ran the code and reproduced this issue, did you then "run the code" as in executed the instructions in a real CPU or did you "run the code" as in guessing what it would do based on your half-assed reading of the code?"
@bagder
💯
Even if it is a better readung of code, it is still not same as running produced instruction on silicon.
I even put Rust compiler's reading my code into same category: best, but sufficient to ... call it running. Even Rust compiler.
@bagder I vibe-coded a CPU that would run into this bug when it would execute the code, but I wasn't actually able to run an OS on this CPU.
@bagder and the wording strongly smells LLM slop...
@[email protected] where does closing your eyes and visualising very hard fall in this scale? /s
@bagder that whole report screams AI all over the place.
@bagder Now, you're the expert on this, but I think your usage of the word "people" here is generous.
